O. Ozan Koyluoglu

Journal Papers

• A. Aghazadeh, H. Nisonoff, O. Ocal, D. H. Brookes, Y. Huang, O. O. Koyluoglu, J. Listgarten, and K. Ramchandran, "Epistatic Net allows the sparse spectral regularization of deep neural networks for inferring fitness functions," Nature communications, vol. 12, no.5225, Sep. 2021.
  [Nature Communications] [pdf preprint] [abstract] [BibTeX]
  Despite recent advances in high-throughput combinatorial mutagenesis assays, the number of labeled sequences available to predict molecular functions has remained small for the vastness of the sequence space combined with the ruggedness of many fitness functions. Expressive models in machine learning (ML), such as deep neural networks (DNNs), can model the nonlinearities in rugged fitness functions, which manifest as high-order epistatic interactions among the mutational sites. However, in the absence of an inductive bias, DNNs overfit to the small number of labeled sequences available for training. Herein, we exploit the recent biological evidence that epistatic interactions in many fitness functions are sparse; this knowledge can be used as an effective inductive bias to regularize DNNs. We have developed a method for sparse epistatic regularization of DNNs, called the epistatic net (EN), which constrains the number of non-zero coefficients in the spectral representation of DNNs. For larger sequences, where finding the spectral transform becomes computationally intractable, we have developed a scalable extension of EN, which subsamples the combinatorial sequence space uniformly inducing a sparse-graph-code structure, and regularizes DNNs using the resulting novel greedy optimization method. Results on several biological landscapes, from bacterial to protein fitness functions, showed that EN consistently improves the prediction accuracy of DNNs and enables them to outperform baseline supervised models in ML which assume other forms of inductive biases. EN estimates all the higher-order epistatic interactions of DNNs trained on massive combinatorial sequence spaces - a computational problem that takes years to solve without leveraging the epistatic sparsity structure in the fitness functions.
  @ARTICLE{Aghazadeh2021Epistatic,
  author = {Aghazadeh, Amirali and Nisonoff, Hunter and Ocal, Orhan and Brookes, David H. and Huang, Yijie and Koyluoglu, O. Ozan and Listgarten, Jennifer and Ramchandran, Kannan},
  title = {Epistatic Net allows the sparse spectral regularization of deep neural networks for inferring fitness functions},
  year = {2021},
  volume={12},
  number={5225},
  doi = {10.1038/s41467-021-25371-3},
  journal = {Nature Communications}
  }
• S. Kadhe, A. Heidarzadeh, A. Sprintson, O. O. Koyluoglu, "Single-server private information retrieval schemes are equivalent to locally recoverable coding schemes," IEEE Journal on Selected Areas in Information Theory, vol.2, no.1, pp.391-402, Mar. 2021.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  The Private Information Retrieval (PIR) problem has recently attracted a significant interest in the information-theory community. In this problem, a client wants to download one or more messages belonging to a database while protecting the identity of the downloaded message(s). In this article, we focus on the scenarios in which (i) the entire database is stored on a single server and (ii) the client has prior side information, namely a subset of messages unknown to the server. Such prior side information is necessary to enable efficient private information retrieval in the single server scenario. In the last decade, there has also been a significant interest in Locally Recoverable Codes (LRCs), a class of storage codes in which each symbol can be recovered from a limited number of other symbols. More recently, there is an interest in cooperative locally recoverable codes, i.e., codes in which multiple symbols can be recovered from a small set of other code symbols. The central problem in this context is given a set of code parameters to design an LRC scheme that includes a locally recoverable code along with encoding, decoding, and repair functions. The paper establishes an equivalence between the single-server PIR schemes and LRC schemes. In particular, we present explicit algorithms that transform a given PIR scheme into an LRC scheme and vice versa. We show that (i) PIR schemes for retrieving a single message are equivalent to classical LRC schemes; and (ii) PIR schemes for retrieving multiple messages are equivalent to cooperative LRC schemes. These equivalence results allow us to recover upper bounds on the download rate for PIR schemes, and to obtain a novel rate upper bound on cooperative LRC schemes. Our results cover schemes based on both linear and non-linear codes.
  @ARTICLE{9335966,
  author={S. {Kadhe} and A. {Heidarzadeh} and A. {Sprintson} and O. O. {Koyluoglu}},
  journal={IEEE Journal on Selected Areas in Information Theory},
  title={Single-Server Private Information Retrieval Schemes are Equivalent to Locally Recoverable Coding Schemes},
  year={2021},
  volume={2},
  number={1},
  pages={391-402},
  doi={10.1109/JSAIT.2021.3053579}
  }
• D. Schwartz and O. O. Koyluoglu, "On the organization of grid and place cells: Neural denoising via subspace learning," Neural Computation, vol.31, no.8, pp.1519-1550, Aug. 2019.
  [Neural Computation] [pdf preprint] [abstract] [BibTeX]
  Place cells in the hippocampus (HC) are active when an animal visits a certain location (referred to as a place field) within an environment. Grid cells in the medial entorhinal cortex (MEC) respond at multiple locations, with firing fields that form a periodic and hexagonal tiling of the environment. The joint activity of grid and place cell populations, as a function of location, forms a neural code for space. In this article, we develop an understanding of the relationships between coding theoretically relevant properties of the combined activity of these populations and how these properties limit the robustness of this representation to noise-induced interference. These relationships are revisited by measuring the performances of biologically realizable algorithms implemented by networks of place and grid cell populations, as well as constraint neurons, which perform denoising operations. Contributions of this work include the investigation of coding theoretic limitations of the mammalian neural code for location and how communication between grid and place cell networks may improve the accuracy of each population's representation. Simulations demonstrate that denoising mechanisms analyzed here can significantly improve the fidelity of this neural representation of space. Furthermore, patterns observed in connectivity of each population of simulated cells predict that anti-Hebbian learning drives decreases in inter-HC-MEC connectivity along the dorsoventral axis.
  @ARTICLE{Schwartz:organization19,
  author={D. {Schwartz} and O. O. {Koyluoglu}},
  journal={Neural Computation},
  title={On the organization of grid and place cells: Neural denoising via subspace learning},
  year={2019},
  month={aug},
  volume={31},
  no={8},
  pages={1519-1550},
  doi={10.1162/neco_a_01208}
  }
• Y. Chen, O. O. Koyluoglu, and A. J. Han Vinck, "Secrecy over multiple-access channel: A game of competition and cooperation," Journal of Universal Computer Science, vol.25, no.8, Aug. 2019.
  [JUCS] [pdf preprint] [abstract] [BibTeX]
  Communication networks have had a transformative impact on our society as they have revolutionized almost all aspects of human interaction. The explosive growth of data traffic has led to an increased demand on improving the reliability, efficiency and security aspects of the systems. In this paper, we focus on the multiple access channel, a communication model where several transmitters communicate to a common receiver (e.g., a cellular telephone network) in the presence of an external eavesdropper. The goal is to explore the competitive yet cooperative relationship between the transmitters in order to obtain an efficient communication under a certain reliability and security guarantee. Moreover, we take a special look into the inner and outer bounds on the secrecy capacity regions over the 2-transmitter DM-MAC with a degraded eavesdropper (assuming that both transmitters are cooperative). We notice that the inner and outer bounds differentiate themselves in the permissible sets of input distributions and thus not tight in general. This leaves the problem of secrecy capacity regions still open.
  @ARTICLE{Chen:Secrecy19,
  author={Y. {Chen} and O. O. {Koyluoglu} and A. J. H. {Vinck}},
  journal={Journal of Universal Computer Science},
  title={Secrecy over Multiple-Access Channel: A Game of Competition and Cooperation},
  year={2019},
  month={aug},
  volume={25},
  number={8} }
• B. Akgun, M. Krunz, and O. O. Koyluoglu, "Vulnerabilities of massive MIMO systems to pilot contamination attacks," IEEE Transactions on Information Forensics and Security, vol.14, no.5, pp.1251-1263, May 2019.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  We consider a single-cell massive multiple-input multiple-output (MIMO) system in which a base station (BS) with a large number of antennas transmits simultaneously to several single-antenna users. The BS acquires the channel state information (CSI) for various receivers using uplink pilot transmissions. We demonstrate the vulnerability of the CSI estimation process to pilot-contamination (PC) attacks. In our attack model, the attacker aims at minimizing the sum rate of downlink transmissions by contaminating the uplink pilots. We first study these attacks for two downlink power allocation strategies under the assumption that the attacker knows the locations of the BS and its users. Later on, we relax this assumption and consider the case when such knowledge is probabilistic. The formulated problems are solved using stochastic optimization, Lagrangian minimization, and game-theoretic methods. A closed-form solution for a special case of the problem is obtained. Furthermore, we analyze the achievable individual secrecy rates under PC attacks and provide an upper bound on these rates. We also study this scenario without a priori knowledge of user locations at the attacker by introducing chance constraints. Our results indicate that such attacks can degrade the throughput of a massive MIMO system by more than 50%.
  @ARTICLE{{8501977,
  author={B. {Akgun} and M. {Krunz} and O. O. {Koyluoglu}},
  journal={IEEE Transactions on Information Forensics and Security},
  title={Vulnerabilities of Massive MIMO Systems to Pilot Contamination Attacks},
  year={2019},
  month={may},
  volume={14},
  number={5},
  pages={1251-1263},
  doi={10.1109/TIFS.2018.2876750}
  }
• G. Calis, S. Shivaramaiah, O. O. Koyluoglu, L. Lazos, "Repair strategies for mobile storage systems," IEEE Transactions on Cloud Computing, May 2019.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  We study the data reliability problem for devices forming a dynamic distributed storage system. Such systems are commonplace in traditional cloud storage applications where storage node failures and updates are frequent. We consider the application of regenerating codes for file maintenance. Such codes require lower bandwidth to regenerate lost data fragments compared to file replication or reconstruction. We investigate threshold-based repair strategies where data repair is initiated after a threshold number of data fragments have been lost. We show that at a low departure-to-repair rate regime, in which repairs are initiated after several nodes have left the system outperforms if repairs are initiated after a single node departure. This optimality is reversed when the node turnover is high. We further compare distributed and centralized repair strategies and derive the optimal repair threshold for minimizing the average repair cost per unit of time. In addition, we examine cooperative repair strategies and show performance improvements. We investigate several models for the time needed for node repair including a simple fixed time model and a more realistic model that takes into account the number of repaired nodes. Finally, an extended model where additional failures are allowed during the repair process is investigated.
  @ARTICLE{Calis:Repair19,
  author={G. {Calis} and S. {Shivaramaiah} and O. O. {Koyluoglu} and L. {Lazos}},
  journal={IEEE Transactions on Cloud Computing},
  title={Repair strategies for mobile storage systems},
  year={2019},
  month={may},
  doi={10.1109/TCC.2019.2914436}
  }
• Z. Liang, D. Schwartz, G. Ditzler, and O. O. Koyluoglu, "The impact of encoding-decoding schemes and weight normalization in spiking neural networks," Neural Networks, vol.108, pp.365-378, Dec. 2018.
  [Neural Networks] [pdf preprint] [abstract] [BibTeX]
  Spike-timing Dependent Plasticity (STDP) is a learning mechanism that can capture causal relationships between events. STDP is considered a foundational element of memory and learning in biological neural networks. Previous research efforts endeavored to understand the functionality of STDP's learning window in spiking neural networks (SNNs). In this study, we investigate the interaction among different encoding/decoding schemes, STDP learning windows and normalization rules for the SNN classifier, trained and tested on MNIST, NIST and ETH80-Contour datasets. The results show that when no normalization rules are applied, classical STDP typically achieves the best performance. Additionally, first-spike decoding classifiers require much less decoding time than a spike count decoding classifier. Thirdly, when no normalization rule is applied, the classifier accuracy decreases as the encoding duration increases from 10ms to 34ms using count decoding scheme. Finally, normalization of output weights is shown to improve the performance of a first-spike decoding classifier, which reveals the importance of weight normalization to SNN.
  @ARTICLE{LIANG2018365,
  author={Z. {Liang} and D. {Schwartz} and G. {Ditzler} and O. O. {Koyluoglu}},
  journal={Neural Networks},
  title={The impact of encoding-decoding schemes and weight normalization in spiking neural networks},
  year={2018},
  month={dec},
  volume={108},
  pages={365-378},
  doi={10.1016/j.neunet.2018.08.024}
  }
• A. S. Rawat, O. O. Koyluoglu, and S. Vishwanath, "Centralized repair of multiple node failures with applications to communication efficient secret sharing," IEEE Transactions on Information Theory, vol.64, no.12, pp.7529-7550, Dec. 2018.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  This paper considers a distributed storage system, where multiple storage nodes can be reconstructed simultaneously at a centralized location. This centralized multi-node repair (CMR) model is a generalization of regenerating codes that allow for bandwidth efficient repair of a single-failed node. This paper focuses on the tradeoff between the amount of data stored and repair bandwidth in the CMR model. In particular, repair bandwidth bounds are derived for the minimum storage multi-node repair (MSMR) and the minimum bandwidth multi-node repair (MBMR) operating points. The tightness of these bounds is analyzed via code constructions. The MSMR point is characterized by codes achieving this point under functional repair for the general set of CMR parameters, as well as with codes enabling exact repair for certain CMR parameters. The MBMR point, on the other hand, is characterized with exact repair codes for all CMR parameters for systems that satisfy a certain entropy accumulation property. Finally, the model proposed here is utilized for the secret sharing problem, where the codes for the multi-node repair problem are used to construct communication efficient secret sharing schemes with the property of bandwidth efficient share repair.
  @ARTICLE{8469091,
  author={A. S. {Rawat} and O. O. {Koyluoglu} and S. {Vishwanath}},
  journal={IEEE Transactions on Information Theory},
  title={Centralized Repair of Multiple Node Failures With Applications to Communication Efficient Secret Sharing},
  year={2018},
  volume={64},
  number={12},
  pages={7529-7550},
  doi={10.1109/TIT.2018.2871451}
  }
• Y. Chen, O. O. Koyluoglu, and A. J. Han Vinck, "Collective secrecy over the K-transmitter multiple access channel," IEEE Transactions on Information Forensics and Security, vol.13, no.9, Sep. 2018.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  This paper studies the problem of secure communication over a K-transmitter multiple access channel (MAC) in the presence of an external eavesdropper, subject to a collective secrecy constraint (i.e., information leakage rate to an eavesdropper on a collection of messages that are from a pre-specified subset of the K transmitters, say subset S in {1, 2, ..., K}, is made vanishing). Since secrecy is of concern only to transmitters {i | i in S} but not to transmitters {i | i not in S}, different transmission strategies could be employed at transmitters {i | i not in S}. Consider the following two scenarios: 1) transmitters {i | i not in S} use deterministic encoders (which are conventionally used for MAC without secrecy), competing for the channel resource (i.e., being competitive) and 2) transmitters {i | i not in S} use stochastic encoders, helping to hide other transmitters' messages from the eavesdropper (i.e., being cooperative). As a result, we establish the respective S-collective secrecy achievable rate regions and demonstrate the advantage of being cooperative theoretically and numerically. To this end, in addition to the standard techniques, our results build upon two techniques. The first is a generalization of Chia-El Gamal's lemma on entropy bound for a set of codewords given partial information. The second is to utilize a compact representation of a list of sets that, together with submodular properties of mutual information functions involved, leads to an efficient Fourier-Motzkin elimination. These two approaches allow us to derive achievable regions in this work, and could also be of independent interest in other context.
  @ARTICLE{8320828,
  author={Y. {Chen} and O. O. {Koyluoglu} and A. J. H. {Vinck}},
  journal={IEEE Transactions on Information Forensics and Security},
  title={Collective Secrecy Over the K-Transmitter Multiple Access Channel},
  year={2018},
  month={sep},
  volume={13},
  number={9},
  pages={2279-2293},
  doi={10.1109/TIFS.2018.2818067} }
• G. Calis and O. O. Koyluoglu, "Architecture-aware coding for distributed storage: Repairable block failure resilient codes," Advances in Mathematics of Communications, vol.12, no.3, Aug. 2018.
  [AIMS] [pdf preprint] [abstract] [BibTeX]
  In large scale distributed storage systems (DSS) deployed in cloud computing, correlated failures resulting in simultaneous failure (or, unavailability) of blocks of nodes are common. In such scenarios, the stored data or a content of a failed node can only be reconstructed from the available live nodes belonging to the available blocks. To analyze the resilience of the system against such block failures, this work introduces the framework of Block Failure Resilient (BFR) codes, wherein the data (e.g., a file in DSS) can be decoded by reading out from a same number of codeword symbols (nodes) from a subset of available blocks of the underlying codeword. Further, repairable BFR codes are introduced, wherein any codeword symbol in a failed block can be repaired by contacting a subset of remaining blocks in the system. File size bounds for repairable BFR codes are derived, and the trade-off between per node storage and repair bandwidth is analyzed, and the corresponding minimum storage regenerating (BFR-MSR) and minimum bandwidth regenerating (BFR-MBR) points are derived. Explicit codes achieving the two operating points for a special case of parameters are constructed, wherein the underlying regenerating codewords are distributed to BFR codeword symbols according to combinatorial designs. Finally, BFR locally repairable codes (BFR-LRC) are introduced, an upper bound on the resilience is derived and optimal code construction are provided by a concatenation of Gabidulin and MDS codes. Repair efficiency of BFR-LRC is further studied via the use of BFR-MSR/MBR codes as local codes. Code constructions achieving optimal resilience for BFR-MSR/MBR-LRCs are provided for certain parameter regimes. Overall, this work introduces the framework of block failures along with optimal code constructions, and the study of architecture-aware coding for distributed storage systems.
  @ARTICLE{Calis:Architecture-aware16,
  author = {Calis, G. and Koyluoglu, O. O.},
  title = {Architecture-aware coding for distributed storage: Repairable block failure resilient codes},
  journal = {Advances in Mathematics of Communications},
  year = {2018},
  month = {aug}
  volume = {12},
  number = {3},
  pages = {465-503},
  doi = {10.3934/amc.2018028}
  }
• O. O. Koyluoglu, Y. Pertzov, S. Manohar, M. Husain, and I. R. Fiete, "Fundamental bound on the persistence and capacity of short-term memory stored as graded persistent activity," eLife, Sep. 2017.
  [eLife] [pdf preprint] [abstract] [BibTeX]
  It is widely believed that persistent neural activity underlies short-term memory. Yet, as we show, the degradation of information stored directly in such networks behaves differently from human short-term memory performance. We build a more general framework where memory is viewed as a problem of passing information through noisy channels whose degradation characteristics resemble those of persistent activity networks. If the brain first encoded the information appropriately before passing the information into such networks, the information can be stored substantially more faithfully. Within this framework, we derive a fundamental lower-bound on recall precision, which declines with storage duration and number of stored items. We show that human performance, though inconsistent with models involving direct (uncoded) storage in persistent activity networks, can be well-fit by the theoretical bound. This finding is consistent with the view that if the brain stores information in patterns of persistent activity, it might use codes that minimize the effects of noise, motivating the search for such codes in the brain.
  @ARTICLE{Koyluoglu:Fundamental16,
  author = {Koyluoglu, O. O. and Pertzov, Y. and Manohar, S. and Husain, M. and Fiete, I. R.},
  title = {Fundamental bound on the persistence and capacity of short-term memory stored as graded persistent activity},
  journal = {eLife},
  year = 2017,
  month = sep
  }
• Y. Chen, O. O. Koyluoglu, and A. Sezgin, "Individual secrecy for the broadcast channel," IEEE Transactions on Information Theory, vol.63, no.9, pp.5981-5999, Sep. 2017.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  This paper studies the problem of secure communications over broadcast channels under the individual secrecy constraints. That is, the transmitter wants to send two independent messages to two legitimate receivers in the presence of an eavesdropper, while keeping the eavesdropper ignorant of each message (i.e., the information leakage rate from each message to the eavesdropper is made vanishing). Building upon Carleial-Hellman's secrecy coding, Wyner's secrecy coding, and the framework of Marton's coding together with techniques, such as rate splitting and indirect decoding, an achievable individual secrecy rate region is established with the characterization of capacity regions for some special cases. In particular, the individual secrecy capacity region for the linear deterministic model is fully characterized, and for the Gaussian model, a constant gap (i.e., 0.5 bits within the individual secrecy capacity region) result is obtained. To illustrate the impact of different secrecy constraints on the corresponding capacity regions, comparisons are made with those satisfying joint secrecy and without secrecy constraints. Overall, when compared with the joint secrecy constraint, the results allow for trading off secrecy level and throughput in the system.
  @ARTICLE{7902110,
  author={Y. {Chen} and O. O. {Koyluoglu} and A. {Sezgin}},
  journal={IEEE Transactions on Information Theory},
  title={Individual Secrecy for the Broadcast Channel},
  year={2017},
  volume={63},
  number={9},
  pages={5981-5999},
  doi={10.1109/TIT.2017.2694552}
  }
• Y. Chen, O. O. Koyluoglu, and A. Sezgin, "Individual secrecy for broadcast channels with receiver side information," IEEE Transactions on Information Theory, vol.63, no.7, Jul. 2017.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  This paper studies the problem of secure communication over the broadcast channel with receiver-side information under the lens of individual secrecy constraints, that is, the transmitter wants to send two independent messages to two receivers, which have, respectively, the desired message of the other receiver as side information, while keeping the eavesdropper ignorant of each message (i.e., the information leakage rate from each message to the eavesdropper is made vanishing). Building upon one-time pad, secrecy coding, and broadcasting schemes, achievable rate regions are investigated, and the capacity region for special cases of either a weak or strong eavesdropper (compared to both legitimate receivers) is characterized. Interestingly, the capacity region for the former corresponds to a line and the latter corresponds to a rectangle with missing corners; a phenomenon occurring due to the coupling between user's rates. Moreover, the individual secrecy capacity region is also fully characterized for the case where the eavesdropper's channel is deterministic. In addition to discrete memoryless setup, Gaussian scenarios are studied. For the Gaussian model, in addition to the strong and weak eavesdropper cases, the capacity region is characterized for the low and high SNR regimes when the eavesdropper's channel is stronger than one receiver but weaker than the other. Remarkably, positive secure transmission rates are always guaranteed under the individual secrecy constraint, unlike the case of the joint secrecy constraint (i.e., the information leakage rate from both messages to the eavesdropper is made vanishing). Thus, this notion of secrecy serves as an appropriate candidate for trading off secrecy level and transmission rate, making secrecy more affordable but still acceptable to the end user.
  @ARTICLE{7914630,
  author={Y. {Chen} and O. O. {Koyluoglu} and A. {Sezgin}},
  journal={IEEE Transactions on Information Theory},
  title={Individual Secrecy for Broadcast Channels With Receiver Side
  Information},
  year={2017},
  volume={63},
  number={7},
  pages={4687-4708},
  doi={10.1109/TIT.2017.2699201}
  }
• G. Calis and O. O. Koyluoglu, "A general construction for PMDS codes," IEEE Communications Letters, vol.21, no.3, pp.452-455, Mar. 2017.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  Partial MDS [(PMDS) also known as maximally recoverable] codes allow for local erasure recovery by utilizing row-wise parities and additional erasure correction through global parities. Recent works on PMDS codes focus on special case parameter settings, and a general construction for PMDS codes is stated as an open problem. This letter provides an explicit construction for PMDS codes for all parameters utilizing concatenation of Gabidulin and MDS codes, a technique originally proposed by Rawat et al. for constructing optimal locally repairable codes. This approach allows for PMDS constructions for any parameters albeit with large field sizes. To lower the field size, a relaxation on the rate requirement is considered, and PMDS codes based on combinatorial designs are constructed.
  @ARTICLE{7740918,
  author={G. {Calis} and O. O. {Koyluoglu}},
  journal={IEEE Communications Letters},
  title={A General Construction for PMDS Codes},
  year={2017},
  volume={21},
  number={3},
  pages={452-455},
  doi={10.1109/LCOMM.2016.2627569}
  }
• B. Akgun, O. O. Koyluoglu, and M. Krunz, "Exploiting full-duplex receivers for achieving secret communications in multiuser MISO networks," IEEE Transactions on Communications, vol.65, no.2, pp.956-968, Feb. 2017.
  [arXiv] [pdf preprint] [abstract] [BibTeX]
  We consider a broadcast channel in which a multi-antenna transmitter (Alice) sends K confidential information signals to K legitimate users (Bobs) in the presence of L eavesdroppers (Eves). Alice uses multiple-input multiple-output (MIMO) precoding to generate the information signals along with her own (Tx-based) friendly jamming (FJ). Interference at each Bob is removed by MIMO zero-forcing. This, however, leaves a “vulnerability region” around each Bob, which can be exploited by a nearby Eve. We address this problem by augmenting Tx-based FJ (TxFJ) with Rx-based FJ (RxFJ), generated by each Bob. Specifically, each Bob uses self-interference suppression to transmit a friendly jamming signal, while simultaneously receiving an information signal over the same channel. We minimize the powers allocated to the information, TxFJ, and RxFJ signals under given guarantees on the individual secrecy rate for each Bob. The problem is solved for the cases when the eavesdropper's channel state information is known/unknown. Simulations show the effectiveness of the proposed solution. Furthermore, we discuss how to schedule transmissions when the rate requirements need to be satisfied on average rather than instantaneously. Under special cases, a scheduling algorithm that serves only the strongest receivers is shown to outperform the one that schedules all receivers.
  @ARTICLE{7792199,
  author={B. {Akgun} and O. O. {Koyluoglu} and M. {Krunz}},
  journal={IEEE Transactions on Communications},
  title={Exploiting Full-Duplex Receivers for Achieving Secret Communications in Multiuser MISO Networks},
  year={2017},
  volume={65},
  number={2},
  pages={956-968},
  doi={10.1109/TCOMM.2016.2641949}
  }
• O. O. Koyluoglu, R. Soundararajan, and S. Vishwanath, "State amplification subject to masking constraints," IEEE Transactions on Information Theory, vol.62, no.11, pp.6233-6250, Nov. 2016.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  This paper considers a state dependent broadcast channel with one transmitter, Alice, and two receivers, Bob and Eve. The problem is to effectively convey ("amplify") the channel state sequence to Bob while "masking" it from Eve. The extent to which the state sequence cannot be masked from Eve is referred to as leakage. This can be viewed as a secrecy problem, where we desire that the channel state itself be minimally leaked to Eve while being communicated to Bob. This paper is aimed at characterizing the tradeoff region between amplification and leakage rates for such a system. An achievable coding scheme is presented, wherein the transmitter transmits a partial state information over the channel to facilitate the amplification process. For the case when Bob observes a stronger signal than Eve, the achievable coding scheme is enhanced with secure refinement. Outer bounds on the tradeoff region are also derived, and used in characterizing some special case results. In particular, the optimal amplification-leakage rate difference, called as differential amplification capacity, is characterized for the reversely degraded discrete memoryless channel, the degraded binary, and the degraded Gaussian channels. In addition, for the degraded Gaussian model, the extremal corner points of the tradeoff region are characterized, and the gap between the outer bound and achievable rate-regions is shown to be less than half a bit for a wide set of channel parameters.
  @ARTICLE{7558108,
  author={O. O. {Koyluoglu} and R. {Soundararajan} and S. {Vishwanath}},
  journal={IEEE Transactions on Information Theory},
  title={State Amplification Subject to Masking Constraints},
  year={2016},
  volume={62},
  number={11},
  pages={6233-6250},
  doi={10.1109/TIT.2016.2605120}
  }
• H. Si, O. O. Koyluoglu, and S. Vishwanath, "Hierarchical polar coding for achieving secrecy over state-dependent wiretap channels without any instantaneous CSI," IEEE Transactions on Communications, vol.64, no.9, pp.3609-3623, Sep. 2016.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  This paper presents a polar coding scheme to achieve secrecy in block fading binary symmetric wiretap channels without the knowledge of instantaneous channel state information (CSI) at the transmitter. For this model, a coding scheme that hierarchically utilizes polar codes is presented. In particular, on polarization of different binary symmetric channels over different fading blocks, each channel use is modeled as an appropriate binary erasure channel over fading blocks. Polar codes are constructed for both coding over channel uses for each fading block and coding over fading blocks for certain channel uses. In order to guarantee security, random bits are introduced at appropriate places to exhaust the observations of the eavesdropper. It is shown that this coding scheme, without instantaneous CSI at the transmitter, is secrecy capacity achieving for the simultaneous fading scenario. For the independent fading case, the capacity is achieved when the fading realizations for the eavesdropper channel is always degraded with respect to the receiver. For the remaining cases, the gap between lower and upper bounds is analyzed. Remarkably, for the scenarios where the secrecy capacity is achieved, the results imply that instantaneous CSI does not increase the secrecy capacity.
  @ARTICLE{Si:Hierarchical16,
  author={Si, H. and Koyluoglu, O. O. and Vishwanath, S.},
  journal={IEEE Transactions on Communications},
  title={Hierarchical polar coding for achieving secrecy over state-dependent wiretap channels without any instantaneous CSI},
  volume={64},
  no={9},
  pages={3609-3623},
  month={Sep.},
  year={2016}
  }
• Y. Yoo, O. O. Koyluoglu, S. Vishwanath, and I. R. Fiete, "Multi-periodic neural coding for adaptive information transfer," Theoretical Computer Science, vol.633, pp.37-53, Jun. 2016.
  [TCS] [pdf preprint] [abstract] [BibTeX]
  Information processing in the presence of noise has been a key challenge in multiple disciplines including computer science, communications, and neuroscience. Among such noise-reduction mechanisms, the shift-map code represents an analog variable by its residues with respect to distinct moduli (that are chosen as geometric scalings of an integer). Motivated by the multi-periodic neural code in the entorhinal cortex, i.e., the coding mechanism of grid cells, this work extends the shift-map code by generalizing the choices of moduli. In particular, it is shown that using similarly sized moduli (for instance, evenly and closely spaced integers, which tend to have large co-prime factors) results in a code whose codewords are separated in an interleaving way such that when the decoder has side information regarding the source, then error control is significantly improved (compared to the original shift map code). This novel structure allows the system to dynamically adapt to the side information at the decoder, even if the encoder is not privy to the side information. A geometrical interpretation of the proposed coding scheme and a method to find such codes are detailed. As an extension, it is shown that this novel code also adapts to scenarios when only a fraction of codeword symbols is available at the decoder.
  @ARTICLE{Yoo:Multi-periodic16,
  author={Yoo, Y. and Koyluoglu, O. O. and Vishwanath, S. and Fiete, I. R.},
  journal={Theoretical Computer Science},
  title={Multi-periodic neural coding for adaptive information transfer},
  volume={633},
  pages={37-53},
  month={Jun.},
  year={2016}
  }
• O. O. Koyluoglu, A. S. Rawat, and S. Vishwanath, "Secure cooperative regenerating codes for distributed storage systems," IEEE Transactions on Information Theory, vol.60, no.9, pp.5228-5244, Sep. 2014.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  Regenerating codes enable trading off repair bandwidth for storage in distributed storage systems (DSS). Due to their distributed nature, these systems are intrinsically susceptible to attacks, and they may also be subject to multiple simultaneous node failures. Cooperative regenerating codes allow bandwidth efficient repair of multiple simultaneous node failures. This paper analyzes storage systems that employ cooperative regenerating codes that are robust to (passive) eavesdroppers. The analysis is divided into two parts, studying both minimum bandwidth and minimum storage cooperative regenerating scenarios. First, the secrecy capacity for minimum bandwidth cooperative regenerating codes is characterized. Second, for minimum storage cooperative regenerating codes, a secure file size upper bound and achievability results are provided. These results establish the secrecy capacity for the minimum storage scenario for certain special cases. In all scenarios, the achievability results correspond to exact repair, and secure file size upper bounds are obtained using min-cut analyses over a suitable secrecy graph representation of DSS. The main achievability argument is based on an appropriate precoding of the data to eliminate the information leakage to the eavesdropper.
  @ARTICLE{6807720,
  author={Koyluoglu, O. O. and Rawat, A. S. and Vishwanath, S.},
  journal={IEEE Transactions on Information Theory},
  title={Secure cooperative regenerating codes for distributed storage systems},
  volume={60},
  number={9},
  pages={5228-5244},
  month={Sep.},
  year={2014}
  }
• H. Si, O. O. Koyluoglu, and S. Vishwanath, "Polar coding for fading channels: Binary and exponential channel cases," IEEE Transactions on Communications, vol.62, no.8, pp.2638-2650, Aug. 2014.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  This work presents a polar coding scheme for fading channels, focusing primarily on fading binary symmetric and additive exponential noise channels. For fading binary symmetric channels, a hierarchical coding scheme is presented, utilizing polar coding both over channel uses and over fading blocks. The receiver uses its channel state information (CSI) to distinguish states, thus constructing an overlay erasure channel over the underlying fading channels. By using this scheme, the capacity of a fading binary symmetric channel is achieved without CSI at the transmitter. Noting that a fading AWGN channel with BPSK modulation and demodulation corresponds to a fading binary symmetric channel, this result covers a fairly large set of practically relevant channel settings. For fading additive exponential noise channels, expansion coding is used in conjunction to polar codes. Expansion coding transforms the continuous-valued channel to multiple (independent) discrete-valued ones. For each level after expansion, the approach described previously for fading binary symmetric channels is used. Both theoretical analysis and numerical results are presented, showing that the proposed coding scheme approaches the capacity in the high SNR regime. Overall, utilizing polar codes in this (hierarchical) fashion enables coding without CSI at the transmitter, while approaching the capacity with low complexity.
  @ARTICLE{6871313,
  author={Si, H. and Koyluoglu, O. O. and Vishwanath, S.},
  journal={IEEE Transactions on Communications},
  title={Polar coding for fading channels: Binary and exponential channel cases},
  volume={62},
  number={8},
  pages={2638-2650},
  month={Aug.},
  year={2014}
  }
• A. S. Rawat, O. O. Koyluoglu, N. Silberstein, and S. Vishwanath, "Optimal locally repairable and secure codes for distributed storage systems," IEEE Transactions on Information Theory, vol.60, no.1, pp.212-236, Jan. 2014.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  This paper aims to go beyond resilience into the study of security and local-repairability for distributed storage systems (DSSs). Security and local-repairability are both important as features of an efficient storage system, and this paper aims to understand the trade-offs between resilience, security, and local-repairability in these systems. In particular, this paper first investigates security in the presence of colluding eavesdroppers, where eavesdroppers are assumed to work together in decoding the stored information. Second, this paper focuses on coding schemes that enable optimal local repairs. It further brings these two concepts together to develop locally repairable coding schemes for DSS that are secure against eavesdroppers. The main results of this paper include: 1) an improved bound on the secrecy capacity for minimum storage regenerating codes; 2) secure coding schemes that achieve the bound for some special cases; 3) a new bound on minimum distance for locally repairable codes; 4) code construction for locally repairable codes that attain the minimum distance bound; and 5) repair-bandwidth-efficient locally repairable codes with and without security constraints.
  @ARTICLE{6655894,
  author={Rawat, A. S. and Koyluoglu, O. O. and Silberstein, N. and Vishwanath, S.},
  journal={IEEE Transactions on Information Theory},
  title={Optimal locally repairable and secure codes for distributed storage systems},
  volume={60},
  number={1},
  pages={212-236},
  month={Jan.},
  year={2014}
  }
• A. El Gamal, O. O. Koyluoglu, M. Youssef, and H. El Gamal, "Achievable secrecy rate regions for the two-way wiretap channel," IEEE Transactions on Information Theory, vol.59, no.12, pp.8099-8114, Dec. 2013.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  The two-way wiretap channel is considered in this paper. Two legitimate users, Alice and Bob, wish to exchange messages securely in the presence of a passive eavesdropper Eve. In the full-duplex scenario, where each node can transmit and receive simultaneously, new achievable secrecy rate regions are obtained based on the idea of allowing the two users to jointly optimize their channel prefixing distributions and binning codebooks in addition to key sharing. The new regions are shown to be strictly larger than the known ones for a wide class of discrete memoryless and Gaussian channels. In the half-duplex case, where a user can only transmit or receive on any given degree of freedom, the idea of randomized scheduling is introduced and shown to offer a significant gain in terms of the achievable secrecy sum-rate. A practical setup is further developed based on a near field wireless communication scenario, and it is shown that one can exploit the two-way nature of the communication, via appropriately randomizing the transmit power levels and transmission schedule, to introduce significant ambiguity at a noiseless Eve.
  @ARTICLE{6600802,
  author={{El Gamal}, A. and Koyluoglu, O. O. and Youssef, M. and {El Gamal}, H.},
  journal={IEEE Transactions on Information Theory},
  title={Achievable secrecy rate regions for the two-way wiretap channel},
  volume={59},
  number={12},
  pages={8099-8114},
  month={Dec.},
  year={2013}
  }
• K. Khalil, O. O. Koyluoglu, H. El Gamal, and M. Youssef, "Opportunistic secrecy with a strict delay constraint," IEEE Transactions on Communications, vol.61, no.11, pp.4700-4709, Nov. 2013.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  We investigate the delay limited secrecy capacity of the flat fading channel under two different assumptions on the available transmitter channel state information (CSI). The first scenario assumes perfect prior knowledge of both the main and eavesdropper channel gains. Here, upper and lower bounds on the delay limited secrecy capacity are derived, and shown to be tight in the high signal-to-noise ratio (SNR) regime. In the second scenario, only the main channel CSI is assumed to be available at the transmitter where, remarkably, we establish the achievability of a non-zero delay-limited secure rate, for a wide class of channel distributions, with a high probability. In the two cases, our achievability arguments are based on a novel two-stage key-sharing approach that overcomes the secrecy outage phenomenon observed in earlier works.
  @ARTICLE{6630484,
  author={Khalil, K. and Koyluoglu, O. O. and {El Gamal}, H. and Youssef, M.},
  journal={IEEE Transactions on Communications},
  title={Opportunistic secrecy with a strict delay constraint},
  volume={61},
  number={11},
  pages={4700-4709},
  month={Nov.},
  year={2013}
  }
• O. O. Koyluoglu and H. El Gamal, "Polar coding for secure transmission and key agreement," IEEE Transactions on Information Forensics and Security, vol.7, no.5, pp.1472-1483, Oct. 2012.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  Achieving information theoretic security with practical coding complexity is of definite interest. This work first focuses on the key agreement problem. For this problem, a new cross-layer secure coding protocol over block fading channels is proposed. The proposed scheme requires only the statistical knowledge about the eavesdropper channel state information (CSI), and, utilizing a privacy amplification technique, reduces the problem of key agreement to a provably secure coding problem per block. Focusing on this secure coding problem, it is shown that polar codes, introduced by Arikan, achieve nonzero perfect secrecy rates for the binary-input degraded wiretap channel while enjoying a remarkably low encoding-decoding complexity. We further show that, in the special case of symmetric main and eavesdropper channels, this coding technique achieves the secrecy capacity. This approach is also extended to the multiple-access channel with a degraded eavesdropper where a nontrivial achievable secrecy region is established. This polar coding method is then utilized in the proposed key agreement protocol, where the secure coding per block is used to create an advantage for the legitimate nodes over the eavesdropper, which is then turned into a private key via the privacy amplification module.
  @ARTICLE{6236066,
  author={Koyluoglu, O. O. and {El Gamal}, H.},
  journal={IEEE Transactions on Information Forensics and Security},
  title={Polar coding for secure transmission and key agreement},
  volume={7},
  number={5},
  pages={1472-1483},
  month={Oct.},
  year={2012}
  }
• O. O. Koyluoglu, C. E. Koksal, and H. El Gamal, "On secrecy capacity scaling in wireless networks," IEEE Transactions on Information Theory, vol.58, no.5, pp.3000-3015, May 2012.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  This work studies the achievable secure rate per source-destination pair in wireless networks. First, a path loss model is considered, where the legitimate and eavesdropper nodes are assumed to be placed according to Poisson point processes with intensities $\lambda$ and $\lambda_e$, respectively. It is shown that, as long as $\lambda_e/\lambda=o((\log n)^{-2})$, almost all of the nodes achieve a perfectly secure rate of $\Omega(\frac{1}{\sqrt{n}})$ for the extended and dense network models. Therefore, under these assumptions, securing the network does not entail a loss in the per-node throughput. The achievability argument is based on a novel multi-hop forwarding scheme where randomization is added in every hop to ensure maximal ambiguity at the eavesdropper(s). Secondly, an ergodic fading model with $n$ source-destination pairs and $n_e$ eavesdroppers is considered. Employing the ergodic interference alignment scheme with an appropriate secrecy pre-coding, each user is shown to achieve a constant positive secret rate for sufficiently large $n$. Remarkably, the scheme does not require eavesdropper CSI (only the statistical knowledge is assumed) and the secure throughput per node increases as we add more legitimate users to the network in this setting. Finally, the effect of eavesdropper collusion on the performance of the proposed schemes is characterized.
  @ARTICLE{6142080,
  author={Koyluoglu, O. O. and Koksal, C. E. and {El Gamal}, H.},
  journal={IEEE Transactions on Information Theory},
  title={On secrecy capacity scaling in wireless networks},
  volume={58},
  number={5},
  pages={3000-3015},
  month={May},
  year={2012}
  }
• O. O. Koyluoglu and H. El Gamal, "Cooperative encoding for secrecy in interference channels," IEEE Transactions on Information Theory, vol.57, no.9, pp.5682-5694, Sep. 2011.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  This paper investigates the fundamental performance limits of the two-user interference channel in the presence of an external eavesdropper. In this setting, we construct an inner bound, to the secrecy capacity region, based on the idea of cooperative encoding in which the two users cooperatively design their randomized codebooks and jointly optimize their channel prefixing distributions. Our achievability scheme also utilizes message-splitting in order to allow for partial decoding of the interference at the nonintended receiver. Outer bounds are then derived and used to establish the optimality of the proposed scheme in certain cases. In the Gaussian case, the previously proposed cooperative jamming and noise-forwarding techniques are shown to be special cases of our proposed approach. Overall, our results provide structural insights on how the interference can be exploited to increase the secrecy capacity of wireless networks.
  @ARTICLE{6006610,
  author={Koyluoglu, O. O. and {El Gamal}, H.},
  title={Cooperative encoding for secrecy in interference channels},
  journal={IEEE Transactions on Information Theory},
  volume={57},
  number={9},
  pages={5682-5694},
  month={Sep.},
  year={2011}
  }
• O. O. Koyluoglu, H. El Gamal, L. Lai, and H. V. Poor, "Interference alignment for secrecy," IEEE Transactions on Information Theory, vol.57, no.6, pp.3323-3332, Jun. 2011.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  This paper studies the frequency/time selective K-user Gaussian interference channel with secrecy constraints. Two distinct models, namely the interference channel with confidential messages and the interference channel with an external eavesdropper, are analyzed. The key difference between the two models is the lack of channel state information (CSI) of the external eavesdropper. Using interference alignment along with secrecy precoding, it is shown that each user can achieve non-zero secure degrees of freedom (DoF) for both cases. More precisely, the proposed coding scheme achieves [(K-2)/(2K-2)] secure DoF with probability one per user in the confidential messages model. For the external eavesdropper scenario, on the other hand, it is shown that each user can achieve [(K-2)/(2K)] secure DoF in the ergodic setting. Remarkably, these results establish the positive impact of interference on the secrecy capacity region of wireless networks.
  @ARTICLE{5773067,
  author={Koyluoglu, O. O. and {El Gamal}, H. and Lifeng Lai and Poor, H. V.},
  journal={IEEE Transactions on Information Theory},
  title={Interference alignment for secrecy},
  volume={57},
  number={6},
  pages={3323-3332},
  month={Jun.},
  year={2011}
  }
• O. O. Koyluoglu and H. El Gamal, "On power control and frequency reuse in the two user cognitive channel," IEEE Transactions on Wireless Communications, vol.8, no.7, pp.3546-3553, Jul. 2009.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  This paper considers the generalized cognitive radio channel where the secondary user is allowed to reuse the frequency during both the idle and active periods of the primary user, as long as the primary rate remains the same. In this setting, the optimal power allocation policy with single-input single-output (SISO) primary and secondary channels is explored. Interestingly, the offered gain resulting from the frequency reuse during the active periods of the spectrum is shown to disappear in both the low and high signal-to-noise ratio (SNR) regimes. We then argue that this drawback in the high SNR region can be avoided by equipping both the primary and secondary transmitters with multiple antennas. Finally, the scenario consisting of SISO primary and multi-input multi-output (MIMO) secondary channels is investigated. Here, a simple zero-forcing approach is shown to significantly outperform the celebrated decoding- forwarding-dirty paper coding strategy (especially in the high SNR regime).
  @ARTICLE{5165318,
  author={Koyluoglu, O. O. and {El Gamal}, H.},
  journal={IEEE Transactions on Wireless Communications},
  title={On power control and frequency reuse in the two user cognitive channel},
  volume={8},
  number={7},
  pages={3546-3553},
  month={Jul.},
  year={2009}
  }

Conference Papers and Abstracts

• A. Aghazadeh, V. Gupta, A. DeWeese, O. O. Koyluoglu, and K. Ramchandran, "BEAR: Sketching BFGS algorithm for ultra-high dimensional feature selection in sublinear memory," in Proceedings of the 2nd Mathematical and Scientific Machine Learning Conference, PMLR 145:75-92, Aug. 2022.
  [PMLR] [pdf preprint] [abstract] [BibTeX]
  We consider feature selection for applications in machine learning where the dimensionality of the data is so large that it exceeds the working memory of the (local) computing machine. Unfortunately, current large-scale sketching algorithms show poor memory-accuracy trade-off in selecting features in high dimensions due to the irreversible collision and accumulation of the stochastic gradient noise in the sketched domain. Here, we develop a second-order feature selection algorithm, called BEAR, which avoids the extra collisions by efficiently storing the second-order stochastic gradients of the celebrated Broyden-Fletcher-Goldfarb-Shannon (BFGS) algorithm in Count Sketch, using a memory cost that grows sublinearly with the size of the feature vector. BEAR reveals an unexplored advantage of second-order optimization for memory-constrained high-dimensional gradient sketching. Our extensive experiments on several real-world data sets from genomics to language processing demonstrate that BEAR requires up to three orders of magnitude less memory space to achieve the same classification accuracy compared to the first-order sketching algorithms with a comparable run time. Our theoretical analysis further proves the global convergence of BEAR with O(1/t) rate in t iterations of the sketched algorithm.
  @INPROCEEDINGS{aghazadeh2022bear,
  title={BEAR: Sketching BFGS Algorithm for Ultra-High Dimensional Feature Selection in Sublinear Memory},
  author={Amirali {Aghazadeh} and Vipul {Gupta} and Alex {DeWeese} and O. O. {Koyluoglu} and Kannan {Ramchandran}},
  booktitle = {Proceedings of the 2nd Mathematical and Scientific Machine Learning Conference},
  series = {Proceedings of Machine Learning Research},
  volume = {145},
  pages = {75--92},
  year = 2022,
  month = aug
  }
• S. Kadhe, N. Rajaraman, O. O. Koyluoglu, and K. Ramchandran, "FastSecAgg: Scalable secure aggregation for privacy-preserving federated learning," in Proc. 2020 CCS Workshop on Privacy-Preserving Machine Learning in Practice, Orlando, FL, Nov. 2020. (Appeared in ICML 2020 as well.)
  [CCS] [pdf preprint] [abstract] [BibTeX]
  Recent attacks on federated learning demonstrate that keeping the training data on clients' devices does not provide sufficient privacy, as the model parameters shared by clients can leak information about their training data. A 'secure aggregation' protocol enables the server to aggregate clients' models in a privacy-preserving manner. However, existing secure aggregation protocols incur high computation/communication costs, especially when the number of model parameters is larger than the number of clients participating in an iteration -- a typical scenario in federated learning. In this paper, we propose a secure aggregation protocol, FastSecAgg, that is efficient in terms of computation and communication, and robust to client dropouts. The main building block of FastSecAgg is a novel multi-secret sharing scheme, FastShare, based on the Fast Fourier Transform (FFT), which may be of independent interest. FastShare is information-theoretically secure, and achieves a trade-off between the number of secrets, privacy threshold, and dropout tolerance. Riding on the capabilities of FastShare, we prove that FastSecAgg is (i) secure against the server colluding with 'any' subset of some constant fraction (e.g. ~10%) of the clients in the honest-but-curious setting; and (ii) tolerates dropouts of a 'random' subset of some constant fraction (e.g. ~10%) of the clients. FastSecAgg achieves significantly smaller computation cost than existing schemes while achieving the same (orderwise) communication cost. In addition, it guarantees security against adaptive adversaries, which can perform client corruptions dynamically during the execution of the protocol.
  @INPROCEEDINGS{Kadhe:ICML2020,
  author={S. {Kadhe} and N. {Rajaraman} and O. O. {Koyluoglu} and K. {Ramchandran}},
  booktitle={2020 CCS Workshop on Privacy-Preserving Machine Learning in Practice},
  title={FastSecAgg: Scalable Secure Aggregation for Privacy-Preserving Federated Learning},
  year={2020},
  month={Nov.},
  address={Orlando, FL}
  }
• S. Kadhe, N. Rajaraman, O. O. Koyluoglu, and K. Ramchandran, "FastSecAgg: Scalable secure aggregation for privacy-preserving federated learning," in Proc. 2020 ICML International Workshop on Federated Learning for User Privacy and Data Confidentiality, Vienna, Austria, Jul. 2020.
  [ICML] [pdf preprint] [abstract] [BibTeX]
  Recent attacks on federated learning demonstrate that keeping the training data on clients' devices does not provide sufficient privacy, as the model parameters shared by clients can leak information about their training data. A 'secure aggregation' protocol enables the server to aggregate clients' models in a privacy-preserving manner. However, existing secure aggregation protocols incur high computation/communication costs, especially when the number of model parameters is larger than the number of clients participating in an iteration -- a typical scenario in federated learning. In this paper, we propose a secure aggregation protocol, FastSecAgg, that is efficient in terms of computation and communication, and robust to client dropouts. The main building block of FastSecAgg is a novel multi-secret sharing scheme, FastShare, based on the Fast Fourier Transform (FFT), which may be of independent interest. FastShare is information-theoretically secure, and achieves a trade-off between the number of secrets, privacy threshold, and dropout tolerance. Riding on the capabilities of FastShare, we prove that FastSecAgg is (i) secure against the server colluding with 'any' subset of some constant fraction (e.g. ~10%) of the clients in the honest-but-curious setting; and (ii) tolerates dropouts of a 'random' subset of some constant fraction (e.g. ~10%) of the clients. FastSecAgg achieves significantly smaller computation cost than existing schemes while achieving the same (orderwise) communication cost. In addition, it guarantees security against adaptive adversaries, which can perform client corruptions dynamically during the execution of the protocol.
  @INPROCEEDINGS{Kadhe:ICML2020,
  author={S. {Kadhe} and N. {Rajaraman} and O. O. {Koyluoglu} and K. {Ramchandran}},
  booktitle={2020 ICML International Workshop on Federated Learning for User Privacy and Data Confidentiality},
  title={FastSecAgg: Scalable Secure Aggregation for Privacy-Preserving Federated Learning},
  year={2020},
  month={Jul.},
  address={Vienna, Austria}
  }
• S. Kadhe, O. O. Koyluoglu, and K. Ramchandran, "Communication-efficient gradient coding for straggler mitigation in distributed learning," in Proc. 2020 IEEE International Symposium on Information Theory (ISIT 2020), Los Angeles, CA, Jun. 2020.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  Distributed implementations of gradient-based methods, wherein a server distributes gradient computations across worker machines, need to overcome two limitations: delays caused by slow running machines called stragglers, and communication overheads. Recently, Ye and Abbe [ICML 2018] proposed a coding-theoretic paradigm to characterize a fundamental trade-off between computation load per worker, communication overhead per worker, and straggler tolerance. However, their proposed coding schemes suffer from heavy decoding complexity and poor numerical stability. In this paper, we develop a communication-efficient gradient coding framework to overcome these drawbacks. Our proposed framework enables using any linear code to design the encoding and decoding functions. When a particular code is used in this framework, its block-length determines the computation load, dimension determines the communication overhead, and minimum distance determines the straggler tolerance. The flexibility of choosing a code allows us to gracefully trade-off the straggler threshold and communication overhead for smaller decoding complexity and higher numerical stability. Further, we show that using a maximum distance separable (MDS) code generated by a random Gaussian matrix in our framework yields a gradient code that is optimal with respect to the trade-off and, in addition, satisfies stronger guarantees on numerical stability as compared to the previously proposed schemes. Finally, we evaluate our proposed framework on Amazon EC2 and demonstrate that it reduces the average iteration time by 16% as compared to prior gradient coding schemes.
  @INPROCEEDINGS{9174120,
  author={S. {Kadhe} and O. O. {Koyluoglu} and K. {Ramchandran}},
  booktitle={2020 IEEE International Symposium on Information Theory (ISIT 2020)},
  title={Communication-Efficient Gradient Coding for Straggler Mitigation in Distributed Learning},
  year={2020},
  month={Jun.},
  address={Los Angeles, CA},
  doi={10.1109/ISIT44484.2020.9174120}
  }
• S. Kadhe, A. Heidarzadeh, A. Sprintson, and O. O. Koyluoglu, "On an equivalence between single-server PIR with side information and locally recoverable codes," in Proc. 2019 IEEE Information Theory Workshop (ITW 2019), Visby, Sweden, Aug. 2019. (Appeared in ITA 2019 as well.)
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  Private Information Retrieval (PIR) problem has recently attracted a significant interest in the information-theory community. In this problem, a user wants to privately download one or more messages belonging to a database with copies stored on a single or multiple remote servers. In the single server scenario, the user must have prior side information, i.e., a subset of messages unknown to the server, to be able to privately retrieve the required messages in an efficient way.In the last decade, there has also been a significant interest in Locally Recoverable Codes (LRCs), a class of storage codes in which each symbol can be recovered from a limited number of other symbols. More recently, there is an interest in cooperative locally recoverable codes, i.e., codes in which multiple symbols can be recovered from a small set of other code symbols.In this paper, we establish a relationship between coding schemes for the single-server PIR problem and LRCs. In particular, we show the following results: (i) PIR schemes designed for retrieving a single message are equivalent to classical LRCs; and (ii) PIR schemes for retrieving multiple messages are equivalent to cooperative LRCs. These equivalence results allow us to recover upper bounds on the download rate for PIR-SI schemes, and to obtain a novel rate upper bound on cooperative LRCs. We show results for both linear and non-linear codes.
  @INPROCEEDINGS{8988956,
  author = {S. {Kadhe} and A. {Heidarzadeh} and A. {Sprintson} and O. O. {Koyluoglu}},
  title = {On an equivalence between single-server PIR with side information and locally recoverable codes},
  booktitle={2019 IEEE Information Theory Workshop (ITW 2019)},
  address={Visby, Sweden},
  month={Aug.},
  year={2019}
  }
• S. Kadhe, O. O. Koyluoglu, and K. Ramchandran, "Gradient coding based on block designs for mitigating adversarial stragglers," in Proc. 2019 IEEE International Symposium on Information Theory (ISIT 2019), Paris, France, Jul. 2019.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  Distributed implementations of gradient-based methods, wherein a server distributes gradient computations across worker machines, suffer from slow running machines, called stragglers. Gradient coding is a coding-theoretic framework to mitigate stragglers by enabling the server to recover the gradient sum in the presence of stragglers. Approximate gradient codes are variants of gradient codes that reduce computation and storage overhead per worker by allowing the server to approximately reconstruct the gradient sum.In this work, our goal is to construct approximate gradient codes that are resilient to stragglers selected by a computationally unbounded adversary. Our motivation for constructing codes to mitigate adversarial stragglers stems from the challenge of tackling stragglers in massive-scale elastic and serverless systems, wherein it is difficult to statistically model stragglers. Towards this end, we propose a class of approximate gradient codes based on balanced incomplete block designs (BIBDs). We show that the approximation error for these codes depends only on the number of stragglers, and thus, adversarial straggler selection has no advantage over random selection. In addition, the proposed codes admit computationally efficient decoding at the server. Next, to characterize fundamental limits of adversarial straggling, we consider the notion of adversarial threshold - the smallest number of workers that an adversary must straggle to inflict certain approximation error. We compute a lower bound on the adversarial threshold, and show that codes based on symmetric BIBDs maximize this lower bound among a wide class of codes, making them excellent candidates for mitigating adversarial stragglers.
  @INPROCEEDINGS{8849690,
  author={S. {Kadhe} and O. O. {Koyluoglu} and K. {Ramchandran}},
  booktitle={2019 IEEE International Symposium on Information Theory (ISIT 2019)},
  title={Gradient Coding Based on Block Designs for Mitigating Adversarial Stragglers},
  year={2019},
  month={Jul.},
  address={Paris, France},
  doi={10.1109/ISIT.2019.8849690}
  }
• S. Kadhe, A. Heidarzadeh, A. Sprintson, and O. O. Koyluoglu, "On an equivalence between single-server PIR with side information and locally recoverable codes," in Proc. 2019 Information Theory and Applications Workshop (ITA 2019), San Diego, CA, Feb. 2019.
  [ITA 2019] [pdf preprint] [abstract] [BibTeX]
  Private Information Retrieval (PIR) problem has recently attracted a significant interest in the information-theory community. In this problem, a user wants to privately download one or more messages belonging to a database with copies stored on a single or multiple remote servers. In the single server scenario, the user must have prior side information, i.e., a subset of messages unknown to the server, to be able to privately retrieve the required messages in an efficient way. In the last decade, there has also been a significant interest in Locally Recoverable Codes (LRC), a class of storage codes in which each symbol can be recovered from a limited number of other symbols. More recently, there is an interest in cooperative locally recoverable codes, i.e., codes in which multiple symbols can be recovered from a small set of other code symbols. In this paper, we establish a relationship between scalarlinear schemes for the single-server PIR problem and scalarlinear LRCs. In particular, we show the following results: (i) PIR schemes designed for retrieving a single message are equivalent to classical LRCs; and (ii) PIR schemes for retrieving multiple messages are equivalent to cooperative LRCs. These equivalence results allow us to recover upper bounds on the download rate for PIR-SI schemes, and to obtain a novel rate upper bound on cooperative LRCs.
  @INPROCEEDINGS{Kadhe:equivalence19,
  author = {S. {Kadhe} and A. {Heidarzadeh} and A. {Sprintson} and O. O. {Koyluoglu}},
  title = {On an equivalence between single-server PIR with side information and locally recoverable codes},
  booktitle={2019 Information Theory and Applications Workshop (ITA 2019)},
  address={San Diego, CA},
  month={Feb.},
  year={2019}
  }
• Y. Chen, O. O. Koyluoglu, and A. J. Han Vinck, "Secrecy in communication networks: Being cooperative or competitive?," in Proc. 2018 Codassca Workshop on Collaborative Technologies and Data Science in Smart City Applications, Yerevan, Armenia, Sep. 2018.
  [Codassca] [pdf preprint] [abstract] [BibTeX]
  Communication networks have had a transformative impact on our society as they have revolutionized almost all aspects of human interaction. The explosive growth of data trafficc has led to an increased demand on improving the reliability, efficiency and security aspects of the systems. In this paper, we focus on the multiple access channel, a communication model where several transmitters communicate to a common receiver (e.g., a cellular telephone network) in the presence of an external eavesdropper. The goal is to explore the competitive yet cooperative relationship between the transmitters in order to obtain an efficient communication under a certain reliability and security guarantee.
  @INPROCEEDINGS{8277932,
  author={Y. {Chen} and O. O. {Koyluoglu} and A. J. H. {Vinck}},
  booktitle={Codassca Workshop on Collaborative Technologies and Data Science in Smart City Applications},
  title={Secrecy in Communication Networks: Being Cooperative or Competitive?},
  year={2018},
  month={Sep.},
  address={Yerevan, Armenia}
  }
• Y. Chen, O. O. Koyluoglu, and A. J. Han Vinck, "Joint secrecy over the K-transmitter multiple access channel," in Proc. 2017 IEEE Information Theory Workshop (ITW 2017), Kaohsiung, Taiwan, Nov. 2017.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  This paper studies the problem of secure communication over a K-transmitter multiple access channel in the presence of an external eavesdropper, subject to a joint secrecy constraint (i.e., information leakage rate from the collection of K messages to an eavesdropper is made vanishing). As a result, we establish the joint secrecy achievable rate region. To this end, our results build upon two techniques in addition to the standard information-theoretic methods. The first is a generalization of Chia-El Gamal's lemma on entropy bound for a set of codewords given partial information. The second is to utilize a compact representation of a list of sets that, together with properties of mutual information, leads to an efficient Fourier-Motzkin elimination. These two approaches could also be of independent interests in other contexts.
  @INPROCEEDINGS{8277932,
  author={Y. {Chen} and O. O. {Koyluoglu} and A. J. H. {Vinck}},
  booktitle={2017 IEEE Information Theory Workshop (ITW 2017)},
  title={Joint secrecy over the K-transmitter multiple access channel},
  year={2017},
  month={Nov.},
  address={Kaohsiung, Taiwan},
  doi={10.1109/ITW.2017.8277932}
  }
• B. Akgun, M. Krunz, and O. O. Koyluoglu, "Pilot contamination attacks in massive MIMO systems," in Proc. 2017 IEEE Conference on Communications and Network Security (CNS 2017), Las Vegas, NV, Oct. 2017.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  We consider a single-cell massive multiple-input multiple-output (MIMO) system in which a base station (BS) with a large number of antennas simultaneously transmits to K single-antenna users in the presence of an attacker. Massive MIMO systems often operate in a time division duplexing (TDD) fashion. The BS estimates the channel state information (CSI) at receivers based on their uplink pilot transmissions. Downlink transmission rates are highly dependent on these estimates, as the BS utilizes the CSI to exploit the beamforming gain offered by massive MIMO. However, this CSI estimation phase is vulnerable to malicious attacks. Specifically, an attacker can contaminate the uplink pilot sequences by generating identical pilot signals to those of legitimate users. We formulate a denial of service (DoS) attack in which the attacker aims to minimize the sum-rate of downlink transmissions by contaminating the uplink pilots. We also consider another attack model where the attacker generates jamming signals in both the CSI estimation and data transmission phases by exploiting in-band full-duplex techniques. We study these attacks under two power allocation strategies for downlink transmissions. Our analysis is conducted when the attacker knows or does not know the locations of the BS and users. When the attacker does not have perfect location information, stochastic optimization techniques are utilized to assess the impact of the attack. The formulated problems are solved using interior-point, Lagrangian minimization, and game-theoretic methods. We obtain a closed-form solution for a special case of the problem. Our results indicate that even though the attacker does not have the perfect location information, proposed pilot contamination attacks degrade the throughput of a massive MIMO system by more than 50%, and reduce fairness among users significantly. In addition, we show that increasing the number of pilot symbols does not prevent the proposed attacks, if the BS uniformly allocates powers for downlink transmissions.
  @INPROCEEDINGS{8228655,
  author={B. {Akgun} and M. {Krunz} and O. O. {Koyluoglu}},
  booktitle={2017 IEEE Conference on Communications and Network Security (CNS 2017)},
  title={Pilot contamination attacks in massive {MIMO} systems},
  year={2017},
  month={Oct.},
  address={Las Vegas, NV},
  doi={10.1109/CNS.2017.8228655}
  }
• I. Samy, O. O. Koyluoglu, and A. S. Rawat, "Efficient data access in hybrid cloud storage," in Proc. 55th Annual Allerton Conference on Communication, Control, and Computing (Allerton 2017), Monticello, IL, Oct. 2017.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  Hybrid cloud is a widely adopted framework where on-premise storage and/or compute resources are combined with public cloud system. This paper explores the storage aspect of this framework, which requires designing coding schemes that are aware of both local and global components of the available storage space. The coding schemes should provide efficient repair mechanisms for the data stored on the public cloud (global storage space) and utilize the local storage space to facilitate seamless access to the overall information stored on the hybrid cloud storage. This paper presents a mathematical model for hybrid cloud storage which takes all these requirements into account. The paper then extends the information flow graph approach to characterize the fundamental limits on access bandwidth of the system, i.e., the amount of data downloaded from the public cloud during the data reconstruction process. This paper also presents several explicit coding schemes that utilize the available local storage space to attain the fundamental limit on the access bandwidth. The setup where multiple clients with varying local storage spaces are supported by a single global storage space is also addressed.
  @INPROCEEDINGS{8262711,
  author={I. {Samy} and O. O. {Koyluoglu} and A. S. {Rawat}},
  booktitle={55th Annual Allerton Conference on Communication, Control, and Computing (Allerton 2017)},
  title={Efficient data access in hybrid cloud storage},
  year={2017},
  month={Oct.},
  address={Monticello, IL},
  doi={10.1109/ALLERTON.2017.8262711}
  }
• I. Samy, G. Calis, and O. O. Koyluoglu, "Secure regenerating codes for hybrid cloud storage systems," in Proc. 2017 IEEE International Symposium on Information Theory (ISIT 2017), Aachen, Germany, Jun. 2017.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  This paper considers a distributed storage system, where multiple storage nodes can be reconstructed simultaneously at a centralized location. This centralized multi-node repair (CMR) model is a generalization of regenerating codes that allow for bandwidth-efficient repair of a single failed node. This work focuses on the trade-off between the amount of data stored and repair bandwidth in this CMR model. In particular, repair bandwidth bounds are derived for the minimum storage multi-node repair (MSMR) and the minimum bandwidth multi-node repair (MBMR) operating points. The tightness of these bounds are analyzed via code constructions. The MSMR point is characterized through codes achieving this point under functional repair for general set of CMR parameters, as well as with codes enabling exact repair for certain CMR parameters. The MBMR point, on the other hand, is characterized with exact repair codes for all CMR parameters for systems that satisfy a certain entropy accumulation property.
  @INPROCEEDINGS{8006921,
  author={I. {Samy} and G. {Calis} and O. O. {Koyluoglu}},
  booktitle={2017 IEEE International Symposium on Information Theory (ISIT 2017)},
  title={Secure regenerating codes for hybrid cloud storage systems},
  year={2017},
  month={Jun.},
  address={Aachen, Germany},
  doi={10.1109/ISIT.2017.8006921}
  }
• D. Schwartz and O. O. Koyluoglu, "Neural noise improves path representation in a simulated network of grid, place, and time cells," in Proc. 2017 Computational and Systems Neuroscience Abstracts (Cosyne 2017), Salt Lake City, UT, Feb. 2017.
  [Cosyne 2017] [pdf preprint] [abstract] [BibTeX]
  Place cells in the hippocampus are active when an animal visits a certain location (referred to as a place field) within an environment, and remain silent otherwise. Grid cells in the medial entorhinal cortex (MEC) respond at multiple locations, and the firing fields follow a hexagonally symmetric periodic pattern. Time cells fire at successive moments in an interval of time with a well defined start and end. The joint activity of grid, place, and time cell populations forms a neural code for path representation (i.e. a temporally ordered sequence of visited locations). We study simulated networks of phenomenologically modeled noisy grid, place, and time cells connected via interneurons. The networks' synaptic connections are trained with an anti-Hebbian learning rule on activity evoked by stimulation with locations and times drawn from recordings of real rat motion during a spatial navigation task. In a previous work, we show that when grid and place cells collaborate in such a network, some errant activity is almost always corrected, and that the modular organization of grid cells can be exploited to improve de-noising and decoding accuracy. Here, we extend that work by including time cells so that temporal ordering of states of network activity can be decoded. We observe that low amplitude noise impairs specific decoding accuracy of space and time (i.e. accuracy of decoding either variable alone, for one moment), but - via biologically plausible de-noising operations implemented by interneurons - bolsters accuracy of decoding a full path. We generate testable predictions that may support the existence of de-noising communication between grid, place, and time cells.
  @INPROCEEDINGS{Schwartz:Neural17,
  author={Schwartz, D. and Koyluoglu, O. O.},
  title={Neural noise improves path representation in a simulated network of grid, place, and time cells},
  booktitle={2017 Computational and Systems Neuroscience Abstracts (Cosyne 2017)},
  address={Salt Lake City, UT},
  month={Feb.},
  year={2017}
  }
• S. Shivaramaiah, G. Calis, O. O. Koyluoglu, and L. Lazos, "Threshold-based file maintenance strategies for mobile cloud storage systems," in IEEE 2016 Global Communications Conference (Globecom 2016), Washington, DC, Dec. 2016.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  We study the data reliability problem for a community of devices forming a mobile cloud storage system. We consider the application of regenerating codes for maintaining a file within a geographically-limited area. Such codes require lower bandwidth to regenerate lost data fragments compared to file replication or reconstruction. We investigate threshold-based repair strategies where data repair is initiated after a threshold number of data fragments have been lost due to node mobility. We show that at a low departure-to-repair rate regime, a {\em lazy repair} strategy in which repairs are initiated after several nodes have left the system outperforms {\em eager repair} in which repairs are initiated after a single departure. This optimality is reversed when nodes are highly mobile. We further compare distributed and centralized repair strategies and derive the optimal repair threshold for minimizing the average repair cost per unit of time, as a function of underlying code parameters.
  @INPROCEEDINGS{Shivaramaiah:Threshold-based16,
  author = {Shivaramaiah, S. and Calis, G. and Koyluoglu, O. O. and Lazos, L.},
  title = {Threshold-based file maintenance strategies for mobile cloud storage systems},
  booktitle={Proc. IEEE 2016 Global Communications Conference (Globecom 2016)},
  address={Washington, DC},
  month={Dec.},
  year={2016}
  }
• M. Ragone, S. Gianelli, D. Schwartz, L. Su, O. O. Koyluoglu, and J. M. Fellous, "The role of hippocampal replay in a computational model of path learning," in Proc. Neuroscience 2016, San Diego, CA, Nov. 2016.
  [SfN] [pdf preprint] [abstract] [BibTeX]
  Hippocampal place cells play an important role in spatial navigation in rodents, but the exact mechanism by which networks of place cells store and process spatial information is poorly understood. ``Replay'' events, in which behaviorally relevant place cells are reactivated offline during sharp-wave ripples (SPWs), are likely critical to spatial information storage. However, the mechanisms by which cells are selected to fire within SPWs are unknown. In this study, we use a mobile robot and a biophysically realistic network model of place cells and interneurons with spike timing-dependent plasticity (STDP) to simulate hippocampal activity during spatial navigation. The robot is comparable in size to a rat and generates paths that are approximations of typical rat trajectories in spatial noise and velocity. These paths are used as inputs to the model and cause specific patterns of activation in the network, which, coupled with the STDP rule, generate a synaptic matrix that contains path-specific information. When SPWs are simulated in a trained network, its synaptic matrix lays constraints on which cells are reactivated, and may explain why some cells are selected for replay firing and others are not. The firing patterns of these cells during path traversal are then decoded in an ideal observer framework to estimate the path taken by the robot. We hypothesize that place cells that participate in trajectory replay events will convey more information about the path than non-replaying cells. We use the mean squared error (MSE) of the estimated location to test this hypothesis by comparing the MSE of decoding between different ensembles of place cells-all cells, replaying cells, and non-replaying cells. Recent work suggests that grid cells and place cells are spatially coherent during replay events. In a separate set of simulations, we investigate the effects of including modular or non-modular grid cell information to path decoding with the aforementioned place cell ensembles. We test the robustness of different code ensembles with nominal paths and paths that contain significant perturbations. Our study shows the extent to which replay is governed by the synaptic weights of the network as learning occurs across trials, and how replay may affect the encoding of a learned path in the face of perturbations.
  @INPROCEEDINGS{Ragone:role16,
  author = {Ragone, M. and Gianelli, S. and Schwartz, D. and Su, L. and Koyluoglu, O. O. and Fellous, J. M.},
  title = {The role of hippocampal replay in a computational model of path learning},
  booktitle={Neuroscience 2016},
  address={San Diego, CA},
  month={Nov.},
  year={2016}
  }
• D. Schwartz and O. O. Koyluoglu, "A hybrid code from grid and place cells," in Proc. Neuroscience 2016, San Diego, CA, Nov. 2016.
  [SfN] [pdf preprint] [abstract] [BibTeX]
  Place cells in the hippocampus are active when an animal visits a certain location (referred to as the place field) within an environment, and remain silent otherwise. Grid cells in the medial entorhinal cortex (mEC) respond at multiple locations, where the firing fields exhibit a hexagonally symmetric periodic pattern. Both cell types have a dorso-ventral organization with larger firing fields distributed towards the ventral end. In addition, grid cells are clustered within discrete modules, wherein cells share scale and rotational phase, but differ in spatial phase offset. The joint activity of grid and place cell populations, as a function of location, forms a neural code for space. Different neural networks are constructed by varying the numbers of cells, grid modules, and grid scaling ratios. An ensemble of codes, for a given set of parameters, is generated by randomly selecting grid cell phases and place cell tuning curves. For each code in the ensemble, codewords are generated by stimulating a network with a discrete set of locations. The resulting code's resilience to neural noise is measured by the minimum pairwise distance between codewords, d. A larger d implies a more noise tolerant representation of space. Normalized code rank, on the other hand, measures dimensionality of the code space. Code rate, the number of locations represented per neuron, measures resolution of location. Analysis of these codes produces the following: For a fixed number of place cells, grid cell parameters may be chosen to produce a code with nearly any desired rank. For a fixed set of grid cell population parameters, increasing the number of place cells increases rank to a maximum, after which, inclusion of additional place cells lowers rank. For a fixed code rate, there is a sharp tradeoff between rank and d, i.e. maximizing either minimizes the other, and lower rates yield more desirable tradeoffs. There is a similar tradeoff between d and code rate. These findings hold for any scaling ratio between consecutive grid modules, L. An intermediate value of L yields a better tradeoff between d and rate. Increasing the number of grid modules increases d, but does not impact rank. Finally, these coding theoretic observations are revisited by measuring the performances of biologically realizable error correction algorithms (e.g., winner take all) implemented by a network of place and grid cell populations, as well as interneurons, which implement de-noising operations. Simulations demonstrate that de-noising mechanisms analyzed here can significantly reduce mean squared error (MSE) of location decoding. Furthermore, the modular organization of grid cells can be exploited to improve MSE.
  @INPROCEEDINGS{Schwartz:hybrid16,
  author = {Schwartz, D. and Koyluoglu, O. O.},
  title = {A hybrid code from grid and place cells},
  booktitle={Neuroscience 2016},
  address={San Diego, CA},
  month={Nov.},
  year={2016}
  }
• Y. Chen, O. O. Koyluoglu, and A. Sezgin "Individual secrecy for the broadcast channel," in Proc. 2016 International Symposium on Information Theory and Its Applications (ISITA 2016), Monterey, CA, Oct. 2016.
  [ISITA] [pdf preprint] [abstract] [BibTeX]
  This paper studies the problem of secure communication over broadcast channels under the \emph{individual} secrecy constraints. That is, the transmitter wants to send two independent messages to two legitimate receivers in the presence of an eavesdropper, while keeping the eavesdropper ignorant of {\it each} message. A general achievable rate region is established by utilizing Marton's coding together with techniques such as rate splitting, Carleial-Hellman's secrecy coding, Wyner's secrecy coding and indirect decoding. Moreover, the individual secrecy capacity regions for some special cases are characterized, and an linear deterministic instance is exhibited to provide insights into the capacity regions under different secrecy constraints.
  @INPROCEEDINGS{Chen:Individual16,
  author = {Chen, Y. and Koyluoglu, O. O. and Sezgin, A.},
  title = {Individual secrecy for the broadcast channel},
  booktitle={2016 International Symposium on Information Theory and Its Applications (ISITA 2016)},
  address={Monterey, CA},
  month={Oct.},
  year={2016}
  }
• Y. Chen, O. O. Koyluoglu, and A. J. Han Vinck, "On secure communication over the multiple access channel," in Proc. 2016 International Symposium on Information Theory and Its Applications (ISITA 2016), Monterey, CA, Oct. 2016.
  [ISITA] [pdf preprint] [abstract] [BibTeX]
  This paper studies the problem of secure communication over a $2$-transmitter multiple access channel (MAC) in the presence of an external eavesdropper. Two different secrecy constraints are considered: 1) individual secrecy (i.e., information leakage rate from each message to the eavesdropper is made vanishing) and 2) joint secrecy (i.e., information leakage rate from both messages to the eavesdropper is made vanishing). As a general result, the respective achievable secrecy rate regions are established. The single-letter characterizations of both regions involve three auxiliary random variables, one for time sharing and two for channel prefixing. Numerical results are presented to demonstrate the impact of different secrecy constraints and the advantage of channel prefixing in enlarging the achievable (individual/joint) secrecy rate regions.
  @INPROCEEDINGS{Chen:secure16,
  author = {Chen, Y. and Koyluoglu, O. O. and Han Vinck, A. J. },
  title = {On secure communication over the multiple access channel},
  booktitle={2016 International Symposium on Information Theory and Its Applications (ISITA 2016)},
  address={Monterey, CA},
  month={Oct.},
  year={2016}
  }
• G. Calis and O. O. Koyluoglu, "On the maintenance of distributed storage systems with backup node for repair," in Proc. 2016 International Symposium on Information Theory and Its Applications (ISITA 2016), Monterey, CA, Oct. 2016.
  [ISITA] [pdf preprint] [abstract] [BibTeX]
  We consider a hierarchical DSS where the content is stored with coding in storage nodes and without coding in a backup node. We analyze the system where mobile nodes request the content from the storage nodes. Backup node is assumed to be accessible by storage nodes only in the case where repairs are required. Under this scenario, we derive the upper bound on the file size bound as well as establish critical points on the trade-off curve known as Minimum Storage Regenerating (MSR) and Minimum Bandwidth Regenerating (MBR) points. Next, we propose optimal code constructions by utilizing existing regenerating codes. Furthermore, we analyze the statistics of maintenance and data access costs under Poisson model for node failures and data requests. We derive expressions for expected values and variances of both such costs. Finally, numerical results are provided where we show that the most studied points in the literature (MSR and MBR) are not always optimal with respect to total cost. We also point out that variance of maintenance cost may be much important than variance of data access cost.
  @INPROCEEDINGS{Calis:Maintenance16,
  author = {Calis, G. and Koyluoglu, O. O.},
  title = {On the maintenance of distributed storage systems with backup node for repair},
  booktitle={2016 International Symposium on Information Theory and Its Applications (ISITA 2016)},
  address={Monterey, CA},
  month={Oct.},
  year={2016}
  }
• A. S. Rawat, O. O. Koyluoglu, and S. Vishwanath, "Centralized repair of multiple node failures," in Proc. 2016 IEEE International Symposium on Information Theory (ISIT 2016), Barcelona, Spain, Jul. 2016.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  This paper considers a distributed storage system, where multiple storage nodes can be reconstructed simultaneously at a centralized location. This centralized multi-node repair (CMR) model is a generalization of regenerating codes that allow for bandwidth-efficient repair of a single failed node. This work focuses on the trade-off between the amount of data stored and repair bandwidth in this CMR model. In particular, repair bandwidth bounds are derived for the minimum storage multi-node repair (MSMR) and the minimum bandwidth multi-node repair (MBMR) operating points. The tightness of these bounds are analyzed via code constructions. The MSMR point is characterized through codes achieving this point under functional repair for general set of CMR parameters, as well as with codes enabling exact repair for certain CMR parameters. The MBMR point, on the other hand, is characterized with exact repair codes for all CMR parameters for systems that satisfy a certain entropy accumulation property.
  @INPROCEEDINGS{Rawat:Centralized16,
  author = {Rawat, A. S. and Koyluoglu, O. O. and Vishwanath, S.},
  title = {Centralized repair of multiple node failures},
  booktitle={2016 IEEE International Symposium on Information Theory (ISIT 2016)},
  address={Barcelona, Spain},
  month={Jul.},
  year={2016}
  }
• A. S. Rawat, O. O. Koyluoglu, and S. Vishwanath, "Progress on high-rate MSR codes: Enabling arbitrary number of helper nodes," in Proc. 2016 Information Theory and Applications Workshop (ITA 2016), La Jolla, CA, Jan. 2016.
  [ITA 2016] [pdf preprint] [abstract] [BibTeX]
  This paper presents a construction for high-rate MDS codes that enable bandwidth-efficient repair of a single node. Such MDS codes are also referred to as the minimum storage regenerating (MSR) codes in the distributed storage literature. The construction presented in this paper generates MSR codes for all possible number of helper nodes d as d is a design parameter in the construction. Furthermore, the obtained MSR codes have polynomial sub-packetization (a.k.a. node size) α. The construction is built on the recent code proposed by Sasidharan et al. [1], which works only for d=n−1, i.e., where all the remaining nodes serve as the helper nodes for the bandwidth-efficient repair of a single node. The results of this paper broaden the set of parameters where the constructions of MSR codes were known earlier.
  @INPROCEEDINGS{Rawat:Progress16,
  author = {Rawat, A. S. and Koyluoglu, O. O. and Vishwanath, S.},
  title = {Progress on high-rate MSR codes: Enabling arbitrary number of helper nodes},
  booktitle={2016 Information Theory and Applications Workshop (ITA 2016)},
  address={La Jolla, CA},
  month={Jan.},
  year={2016}
  }
• B. Akgun, O. O. Koyluoglu, and M. Krunz, "Receiver-based friendly jamming with single-antenna full-duplex receivers in a multiuser broadcast channel," in Proc. 2015 IEEE Global Communications Conference (Globecom 2015), San Diego, CA, Dec. 2015.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  This paper considers a broadcast channel with a multi-antenna transmitter (Alice) sending two independent confidential data streams to two legitimate users (Bob and Charlie) in the presence of a passive eavesdropper (Eve). To enhance their secrecy rates, Bob and Charlie are assumed to be capable of self-interference suppression (SIS). Alice, on the other hand, uses MIMO precoding to generate the two confidential information signals along with its own (Tx-based) friendly jamming. The interfering signals at Bob and Charlie are removed by employing the zero-forcing technique. This, however, leaves ``vulnerability regions'' around Bob and Charlie, which can be exploited by a nearby eavesdropper. We address this problem by augmenting Tx-based friendly jamming with Rx-based friendly jamming, generated by Bob and Charlie. For the resulting broadcast channel, a secrecy encoding scheme is developed to construct the signals intended to Bob and Charlie. The corresponding achievable secrecy sum-rate is characterized, and an optimization problem is formulated. A special case of this problem is investigated. Simulation results show the effectiveness of utilizing (Tx- and/or Rx-based) jamming, and the impact of the degree of SIS on physical-layer security.
  @INPROCEEDINGS{Akgun:Receiver-based15,
  author={Akgun, B. and Koyluoglu, O. O. and Krunz, M.},
  title={Receiver-based friendly jamming with single-antenna full-duplex receivers in a multiuser broadcast channel},
  booktitle={2015 IEEE Global Communications Conference (Globecom 2015)},
  address={San Diego, CA},
  month={Dec.},
  year={2015}
  }
• Y. Chen, O. O. Koyluoglu, and A. Sezgin, "On the individual secrecy rate region for the broadcast channel with an external eavesdropper," in Proc. 2015 IEEE International Symposium on Information Theory (ISIT 2015), Hong Kong, China, Jun. 2015.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  This paper studies the problem of secure communication over broadcast channels under the lens of individual secrecy constraints (i.e., information leakage from each message to an eavesdropper is made vanishing). It is known that, for the communication over the degraded broadcast channels, the stronger receiver is able to decode the message of the weaker receiver. In the individual secrecy setting, the message for the weaker receiver can be further utilized to secure the partial message that is intended to the stronger receiver. With such a coding spirit, it is shown that more secret bits can be conveyed to the stronger receiver. In particular, for the corresponding Gaussian model, a constant gap (i.e., 0.5 bits within the individual secrecy capacity region) result is obtained. Overall, when compared with the joint secrecy constraint, the results allow for trading-off secrecy level and throughput in the system.
  @INPROCEEDINGS{Chen:individualsecrecy15,
  author={Chen, Y. and Koyluoglu, O. O. and Sezgin, A.},
  title={On the individual secrecy rate region for the broadcast channel with an external eavesdropper},
  booktitle={2015 IEEE International Symposium on Information Theory (ISIT 2015)},
  address={Hong Kong, China},
  month={Jun.},
  year={2015}
  }
• H. Si, O. O. Koyluoglu, and S. Vishwanath, "Achieving secrecy without any instantaneous CSI: Polar coding for fading wiretap channels," in Proc. 2015 IEEE International Symposium on Information Theory (ISIT 2015), Hong Kong, China, Jun. 2015.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  This paper presents a polar coding scheme for fading wiretap channels that achieves reliability as well as security without the knowledge of instantaneous channel state information at the transmitter. Specifically, a block fading model is considered for the wiretap channel that consists of a transmitter, a receiver, and an eavesdropper; and only the information regarding the statistics (i.e., distribution) of the channel state information is assumed at the transmitter. For this model, a coding scheme that hierarchically utilizes polar codes is presented in order to address channel state variation. In particular, on polarization of different binary symmetric channels over different fading blocks, each channel use (corresponding to a possibly different polarization) is modeled as an appropriate binary erasure channel over fading blocks. Polar codes are constructed for both coding over channel uses for each fading block and coding over fading blocks for certain channel uses. In order to guarantee security, message bits are transmitted such that they can be reliably decoded at the receiver, and random bits are introduced to exhaust the observations of the eavesdropper. It is shown that this coding scheme, without instantaneous channel state information at the transmitter, is secrecy capacity achieving for the corresponding fading binary symmetric wiretap channel.
  @INPROCEEDINGS{Si:Achieving15,
  author={Si, H. and Koyluoglu, O. O. and Vishwanath, S.},
  title={Achieving secrecy without any instantaneous CSI: Polar coding for fading wiretap channels},
  booktitle={2015 IEEE International Symposium on Information Theory (ISIT 2015)},
  address={Hong Kong, China},
  month={Jun.},
  year={2015}
  }
• Y. Chen, O. O. Koyluoglu, and A. Sezgin, "On the individual secrecy for Gaussian broadcast channels with receiver side information," in Proc. 2015 IEEE International Conference on Communications (ICC 2015), London, UK, Jun. 2015.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  This paper studies the individual secrecy capacity region of the broadcast channel with receiver side information. First, an achievable rate region is established for the discrete memoryless case by employing superposition coding. Further, it is extended to the corresponding Gaussian case, where the individual secrecy capacity region is characterized in case of a weak or strong eavesdropper (compared to two legitimate receivers). For the case left, inner and outer bounds are established and the individual secrecy capacity region is characterized for the low and high SNR regimes. Note that the last case is distinctive due to the individual secrecy constraint, in the sense that positive rate pair is still possible although the eavesdropper may have the advantage against at least one of the legitimate receivers over the channel, unlike the situation if the joint secrecy constraint is imposed.
  @INPROCEEDINGS{Chen:individual15,
  author={Chen, Y. and Koyluoglu, O. O. and Sezgin, A.},
  title={On the individual secrecy for Gaussian broadcast channels with receiver side information},
  booktitle={Proc. 2015 IEEE International Conference on Communications (ICC 2015)},
  address={London, UK,},
  month={Jun.},
  year={2015}
  }
• I. Aykin, O. O. Koyluoglu, and J.-M. Fellous, "Formation of dorso-ventral grid cell modules: The role of learning," in Proc. 2015 Computational and Systems Neuroscience Abstracts (Cosyne 2015), Salt Lake City, UT, Mar. 2015.
  [Cosyne 2015] [abstract] [BibTeX]
  Hippocampal place cells are active at specific locations, and grid cells in the medial entorhinal cortex show periodic grid-like firing as a function of location. Both types of cells are organized from dorsal to ventral levels in increasing spatial field sizes. The size gradient seems smooth for place cells, but modular for grid cells. Together, grid and place cells form a topographical map for navigational tasks. This map can be characterized by a weight matrix, where the entries correspond to synaptic connections between grid and place cells. Using a Hebbian plasticity rule with decay, these connections can be strengthened or weakened in accordance with the fields visited during behavior. Thus, a rat learning a path through specific locations will form a weight matrix that could act as a signature for the learned path.
  Starting with a homogeneous, smooth distribution of firing field gradients of place and grid cells, we study the structure of the weight matrix when following actual paths recorded from rodents. Two conditions were considered: random foraging and learning of paths between specific rewarded locations. The resulting weight matrices show significant differences: a) Weight matrix corresponding to foraging shows a smooth connectivity pattern throughout the synaptic population; b) The weight matrix of the path with reward locations shows a clear pattern consisting of sub-modules organized along the dorso-ventral axis. This result suggests that grid cells synaptically group themselves in modules on a learned path. c) This modularity does not appear in place-place or grid-place connections. d) These results are compatible with, and may partially explain, the electrophysiological results obtained experimentally.
  Overall, our plasticity-based framework is a novel computational model that suggests a mechanism for the formation of dorso-ventral modules in grid cells.
  @INPROCEEDINGS{Aykin:Formation15,
  author={Aykin, I. and Koyluoglu, O. O. and Fellous, J.-M.},
  title={Formation of dorso-ventral grid cell modules: The role of learning},
  booktitle={2015 Computational and Systems Neuroscience Abstracts (Cosyne 2015)},
  address={Salt Lake City, UT},
  month={Mar.},
  year={2015}
  }
• G. Calis and O. O. Koyluoglu, "Repairable block failure resilient codes," in Proc. 2014 IEEE International Symposium on Information Theory (ISIT 2014), Honolulu, HI, Jun. 2014.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  In large scale distributed storage systems (DSS) deployed in cloud computing, correlated failures resulting in simultaneous failure (or, unavailability) of blocks of nodes are common. In such scenarios, the stored data or a content of a failed node can only be reconstructed from the available live nodes belonging to available blocks. To analyze the resilience of the system against such block failures, this work introduces the framework of Block Failure Resilient (BFR) codes, wherein the data (e.g., file in DSS) can be decoded by reading out from a same number of codeword symbols (nodes) from each available blocks of the underlying codeword. Further, repairable BFR codes are introduced, wherein any codeword symbol in a failed block can be repaired by contacting to remaining blocks in the system. Motivated from regenerating codes, file size bounds for repairable BFR codes are derived, trade-off between per node storage and repair bandwidth is analyzed, and BFR-MSR and BFR-MBR points are derived. Explicit codes achieving these two operating points for a wide set of parameters are constructed by utilizing combinatorial designs, wherein the codewords of the underlying outer codes are distributed to BFR codeword symbols according to projective planes.
  @INPROCEEDINGS{Calis:Repairable14,
  author={Calis, G. and Koyluoglu, O. O.},
  title={Repairable block failure resilient codes},
  booktitle={Proc. 2014 IEEE International Symposium on Information Theory (ISIT 2014)},
  address={Honolulu, HI,},
  month={Jun.},
  year={2014}
  }
• Y. Chen, O. O. Koyluoglu, and A. Sezgin, "On achievable individual-secrecy rate region for broadcast channels with receiver side information," in Proc. 2014 IEEE International Symposium on Information Theory (ISIT 2014), Honolulu, HI, Jun. 2014.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  In this paper, we study the problem of secure communication over the broadcast channel with receiver side information, under the lens of individual secrecy constraints (i.e., information leakage from each message to an eavesdropper is made vanishing). Several coding schemes are proposed by extending known results in broadcast channels to this secrecy setting. In particular, individual secrecy provided via one-time pad signal is utilized in the coding schemes. As a result, we obtain an achievable rate region together with a characterization of the capacity region for special cases of either a weak or strong eavesdropper (compared to both legitimate receivers). Interestingly, the capacity region for the former corresponds to a line and the latter corresponds to a square with missing corners; a phenomenon occurring due to the coupling between user's rates. At the expense of having a weaker notion of security, positive secure transmission rates are always guaranteed, unlike the case of the joint secrecy constraint.
  @INPROCEEDINGS{Chen:achievable14,
  author={Chen, Y. and Koyluoglu, O. O. and Sezgin, A.},
  title={On achievable individual-secrecy rate region for broadcast channels with receiver side information},
  booktitle={Proc. 2014 IEEE International Symposium on Information Theory (ISIT 2014)},
  address={Honolulu, HI,},
  month={Jun.},
  year={2014}
  }
• H. Si, O. O. Koyluoglu, and S. Vishwanath, "Lossy compression of exponential and laplacian sources using expansion coding," in Proc. 2014 IEEE International Symposium on Information Theory (ISIT 2014), Honolulu, HI, Jun. 2014.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  A general method of source coding is proposed in this paper, which enables one to reduce the problem of compressing an analog (continuous-valued) source to a set of much simpler problems, compressing discrete sources. Specifically, the focus is on lossy compression of exponential and Laplace sources, which are represented as set of discrete variables through a finite alphabet expansion. Due to the decomposability property of such sources, the resulting random variables post expansion are independent and discrete. Thus, these variables can be considered as independent discrete source coding problems, and the original problem is reduced to coding over these sources with a total distortion constraint. Any feasible solution to this resulting optimization problem corresponds to an achievable rate distortion pair of the original continuous-valued source compression problem. Although finding the optimal solution for a given distortion is not a tractable task, we show that, via a heuristic choice, our expansion coding scheme still presents a good performance in the low distortion regime. Further, by adopting low-complexity codes designed for discrete source coding, the total coding complexity can be reduced for practical implementations.
  @INPROCEEDINGS{Si:Lossy14,
  author={Si, H. and Koyluoglu, O. O. and Vishwanath, S.},
  title={Lossy compression of exponential and laplacian sources using expansion coding},
  booktitle={Proc. 2014 IEEE International Symposium on Information Theory (ISIT 2014)},
  address={Honolulu, HI,},
  month={Jun.},
  year={2014}
  }
• A. S. Rawat, N. Silberstein, O. O. Koyluoglu, and S. Vishwanath, "Secure distributed storage systems: Local repair with minimum bandwidth regeneration," in Proc. 6th International Symposium on Communications, Control, and Signal Processing (ISCCSP'14), Athens, Greece, Apr. 2014.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  This paper addresses the issue of securing information stored on a distributed storage system from a passive eavesdropping attack. The security notion is perfect secrecy, i.e., the system is said to be secure only if the mutual information between the stored information and the observations at the adversary is zero. The paper summarizes state of the art on securing repair-efficient distributed storage systems. Then, storage systems that employ locally repairable codes with minimum bandwidth regenerating codes as local codes (MBR-LRCs) are investigated. A secure file size upper bound and a construction of secure MBR-LRCs are provided. These two are shown to match under special cases, establishing the secrecy capacity of these systems.
  @INPROCEEDINGS{Rawat:Secure14,
  author={Rawat, A. S. and Silberstein, N. and Koyluoglu, O. O. and Vishwanath, S.},
  title={Secure distributed storage systems: Local repair with minimum bandwidth regeneration},
  booktitle={Proc. 6th International Symposium on Communications, Control, and Signal Processing (ISCCSP'14)},
  address={Athens, Greece},
  month={Apr.},
  year={2014}
  }
• O. O. Koyluoglu, Y. Chen, and A. Sezgin, "Broadcast channel with receiver side information: Achieving individual secrecy," in Proc. 2014 International Zurich Seminar on Communications (IZS 2014), Zurich, Switzerland, Feb. 2014.
  [IZS] [pdf preprint] [abstract] [BibTeX]
  In this paper, we study the problem of secure communication over the broadcast channel with receiver side information, under the lens of individual secrecy constraints (i.e., information leakage from each message to an eavesdropper is made vanishing). Several coding schemes are proposed by extending known results in broadcast channels to this secrecy setting. In particular, individual secrecy provided via one-time pad signal is utilized in the coding schemes. As a preliminary result, we obtain a general achievable region together with a characterization of the capacity region for the case of a degraded eavesdropper.
  @INPROCEEDINGS{Koyluoglu:Broadcast14,
  author={Koyluoglu, O. O. and Chen, Y. and Sezgin, A.},
  title={Broadcast channel with receiver side information: Achieving individual secrecy},
  booktitle={Proc. 2014 International Zurich Seminar on Communications (IZS 2014)},
  address={Zurich, Switzerland},
  month={Feb.},
  year={2014}
  }
• H. Si, O. O. Koyluoglu, and S. Vishwanath, "Polar coding for fading channels," in Proc. 2013 IEEE Information Theory Workshop (ITW 2013), Seville, Spain, Sep. 2013.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  A polar coding scheme for fading channels is proposed in this paper. More specifically, the focus is on the Gaussian fading channel with a BPSK modulation, where the equivalent channel is modeled as a binary symmetric channel with varying cross-over probabilities. To deal with variable channel states, a coding scheme of hierarchically utilizing polar codes is proposed. In particular, by observing the polarization of different binary symmetric channels over different fading blocks, each channel use corresponding to a different polarization is modeled as a binary erasure channel such that polar codes could be adopted to encode over blocks. It is shown that the proposed coding scheme, without instantaneous channel state information at the transmitter, achieves the capacity of the corresponding fading binary symmetric channel.
  @INPROCEEDINGS{Si:Polar13,
  author={Si, H. and Koyluoglu, O. O. and Vishwanath, S.},
  title={Polar coding for fading channels},
  booktitle={Proc. 2013 IEEE Information Theory Workshop (ITW 2013)},
  address={Seville, Spain},
  month={Sep.},
  year={2013}
  }
• G. Kamath, N. Silberstein, N. Prakash, A. S. Rawat, V. Lalitha, O. O. Koyluoglu, P. V. Kumar, and S. Vishwanath, "Explicit MBR all-symbol locality codes," in Proc. 2013 IEEE International Symposium on Information Theory (ISIT 2013), Istanbul, Turkey, Jul. 2013.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  Node failures are inevitable in distributed storage systems (DSS). To enable efficient repair when faced with such failures, two main techniques are known: Regenerating codes, i.e., codes that minimize the total repair bandwidth; and codes with locality, which minimize the number of nodes participating in the repair process. This paper focuses on regenerating codes with locality, using pre-coding based on Gabidulin codes, and presents constructions that utilize minimum bandwidth regenerating (MBR) local codes. The constructions achieve maximum resilience (i.e., optimal minimum distance) and have maximum capacity (i.e., maximum rate). Finally, the same pre-coding mechanism can be combined with a subclass of fractional-repetition codes to enable maximum resilience and repair-by-transfer simultaneously.
  @INPROCEEDINGS{Kamath:Explicit13,
  author = {Kamath, G. and Silberstein, N. and Prakash, N. and Rawat, A. S. and Lalitha, V. and Koyluoglu, O. O. and Kumar, P. V. and Vishwanath, S.},
  title = {Explicit MBR all-symbol locality codes},
  booktitle={2013 IEEE International Symposium on Information Theory (ISIT 2013)},
  address={Istanbul, Turkey},
  month={Jul.},
  year={2013}
  }
• A. S. Rawat, O. O. Koyluoglu, N. Silberstein, and S. Vishwanath, "Secure locally repairable codes for distributed storage systems," in Proc. 2013 IEEE International Symposium on Information Theory (ISIT 2013), Istanbul, Turkey, Jul. 2013.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  This paper presents coding schemes for distributed storage systems (DSS) that are secure against eavesdroppers, while simultaneously enabling efficient node repair (regeneration). Towards this, novel upper bounds on secrecy capacity for minimum storage regenerating (MSR) codes and locally repairable codes (LRC) are derived. The eavesdropper model considered in this paper incorporates the ability to listen in on data downloaded during $\ell_2$ node repairs in addition to content stored on $\ell_1$ nodes. Finally, this paper presents coding schemes, based on precoding using Gabidulin codes, that achieve the upper bounds on secrecy capacity and characterize the secrecy capacity of DSS for various settings of system parameters.
  @INPROCEEDINGS{Rawat:Secure13,
  author = {Rawat, A. S. and Koyluoglu, O. O. and Silberstein, N. and Vishwanath, S.},
  title = {Secure locally repairable codes for distributed storage systems},
  booktitle={2013 IEEE International Symposium on Information Theory (ISIT 2013)},
  address={Istanbul, Turkey},
  month={Jul.},
  year={2013}
  }
• O. O. Koyluoglu, A. S. Rawat, and S. Vishwanath, "The secrecy capacity of minimum bandwidth cooperative regenerating codes," in Proc. 2013 IEEE International Symposium on Information Theory (ISIT 2013), Istanbul, Turkey, Jul. 2013.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  Regenerating codes enable trading off repair bandwidth for storage in distributed storage systems (DSS). Due to their distributed nature, these systems are intrinsically susceptible to attacks, and they may be susceptible to multiple node failures. This paper analyzes storage systems that employ cooperative regenerating codes that are robust to passive eavesdroppers, and proposes codes achieving the secrecy capacity for the minimum bandwidth cooperative regenerating point. The achievability results correspond to exact repair, and secure file size upper bounds are obtained using mincut analyses over a suitable secrecy graph representation of DSS. The main achievability argument is based on appropriate precoding of the data using MRD (Gabidulin) codes to eliminate any information leakage to the eavesdropper.
  @INPROCEEDINGS{Koyluoglu:Secrecy13,
  author = {Koyluoglu, O. O. and Rawat, A. S. and Vishwanath, S.},
  title = {The secrecy capacity of minimum bandwidth cooperative regenerating codes},
  booktitle={2013 IEEE International Symposium on Information Theory (ISIT 2013)},
  address={Istanbul, Turkey},
  month={Jul.},
  year={2013}
  }
• N. Silberstein, A. S. Rawat, O. O. Koyluoglu, and S. Vishwanath, "Optimal locally repairable codes via rank-metric codes," in Proc. 2013 IEEE International Symposium on Information Theory (ISIT 2013), Istanbul, Turkey, Jul. 2013.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  This paper presents a new explicit construction for locally repairable codes (LRCs) for distributed storage systems which possess all-symbol locality and the largest possible minimum distance, or equivalently, can tolerate the maximum number of node failures. This construction, based on maximum rank distance (MRD) Gabidulin codes, provides new optimal vector and scalar LRCs. In addition, the paper also discusses mechanisms by which codes obtained using this construction can be used to construct LRCs with efficient local repair of failed nodes by combination of LRCs with regenerating codes.
  @INPROCEEDINGS{Silberstein:Optimal13,
  author = {Silberstein, N. and Rawat, A. S. and Koyluoglu, O. O. and Vishwanath, S.},
  title = {Optimal locally repairable codes via rank-metric codes},
  booktitle={2013 IEEE International Symposium on Information Theory (ISIT 2013)},
  address={Istanbul, Turkey},
  month={Jul.},
  year={2013}
  }
• O. O. Koyluoglu and I. R. Fiete, "Information-theoretic limits on encoding over diverse populations," in 2013 Computational and Systems Neuroscience Abstracts (Cosyne 2013), Salt Lake City, UT, Feb. 2013.
  [Cosyne 2013] [abstract] [BibTeX]
  Population coding refers to a setting where a given stimulus is represented by the activities of a population of neurons. For instance, in orientation-tuned V1 neurons, each neuron fires near its preferred stimulus, with an activity profile given by the tuning curve. When combined with an estimator, these activities constitute a fully identified coding system in which the efficiency of the system is quantified by a measure of distortion (error) between the estimated stimulus and its actual value.
  Here, we use an information-theoretic approach to bound distortion (in a mean-square sense) for populations of neurons: a stimulus $s$ is first sent to an encoder, that computes a vector-valued function of the stimulus, and each entry of the vector is then represented in a separate population code. We assume the total number of neurons is fixed at $Q$, and that the Fisher information in each neural population scales linearly with the number of neurons in that population (as seen for unimodal tuning curves with Poisson spike variability, among various examples). We consider two scenarios: The encoder simply passes out $s$, to one population of $Q$ total neurons; or, it passes out elements of the $N$-d vector $\vec{x}(s)$, to $N$ populations of $M=Q/N$ neurons each. For these scenarios, we use joint source-channel coding theory to bound how the information-theoretically minimal distortion will scale with $M,N$. We show that breaking the neurons into $N$ populations can, with appropriate encoding, result in distortions that scale as $M^{-N}$, whereas directly representing the stimulus in a single population, by necessity, produces distortions that scale as $1/(MN)$.
  Our results show that diverse population encoding can result in potentially much lower distortion, and quantify how distortion scales with number of populations.
  @INPROCEEDINGS{Koyluoglu:Information-theoretic13,
  author={Koyluoglu, O. O. and Fiete, I. R.},
  title={Information-theoretic limits on encoding over diverse populations},
  booktitle={2013 Computational and Systems Neuroscience Abstracts (Cosyne 2013)},
  address={Salt Lake City, UT},
  month={Feb.},
  year={2013}
  }
• A. S. Rawat, N. Silberstein, O. O. Koyluoglu, and S. Vishwanath, "Optimal locally repairable codes with local minimum storage regeneration via rank-metric codes," in Proc. 2013 Information Theory and Applications Workshop (ITA 2013), San Diego, CA, Feb. 2013.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  This paper presents a new explicit construction for locally repairable codes (LRCs) for distributed storage systems. The codes possess all-symbols locality and maximal possible minimum distance, or equivalently, can tolerate the maximal number of node failures. This construction, based on maximum rank distance (MRD) Gabidulin codes, provides minimum distance optimal vector and scalar LRCs for a wide range of parameters. In addition, vector LRCs that allow for efficient local repair of failed nodes are considered. Towards this, the paper derives an upper bound on the amount of data that can be stored on DSS employing minimum distance optimal LRCs with given repair bandwidth, and presents codes which attain this bound by combining MRD and minimum storage regenerating (MSR) codes.
  @INPROCEEDINGS{Rawat:Optimal13,
  author={Rawat, A. S. and Silberstein, N. and Koyluoglu, O. O. and Vishwanath, S.},
  title={Optimal locally repairable codes with local minimum storage regeneration via rank-metric codes},
  booktitle={Proc. 2013 Information Theory and Applications Workshop (ITA 2013)},
  address={San Diego, CA},
  month={Feb.},
  year={2013}
  }
• Y. Yoo, O. O. Koyluoglu, S. Vishwanath, and I. R. Fiete, "Dynamic shift-map coding with side information at the decoder," in Proc. 50th Annual Allerton Conference on Communication, Control, and Computing (Allerton 2012), Monticello, IL, Oct. 2012.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  Shift-map codes have been studied as joint source- channel codes for continuous sources. These codes are useful in delay-limited scenarios and also provide better tolerance to deviations of the signal-to-noise ratio (SNR) from a target SNR, compared to separate source and channel coding. This paper defines a generalized family of shift-map codes that share a strong connection with redundant residue number systems (RRNS), and are henceforth called RRNS-map codes. In the proposed coding scheme, side information about the source allows the decoder to consider only a fraction of the codebook for decoding, with no change in the encoding process. With an appropriately designed RRNS-map code, in this fraction of the codebook, the codewords are much better separated than the original codebook. As a result, RRNS-map codes achieve the same distortion in the mean square error sense as conventional shift-map codes without side information, but significantly outperform shift-map codes when side information is provided to the decoder. This coding scheme is ideally suited to applications where a simple and fixed encoding scheme is desired at the encoder, while the decoder is given access to side information about the source.
  @INPROCEEDINGS{Yoo:Dynamic12,
  author={Yoo, Y. and Koyluoglu, O. O. and Vishwanath, S. and Fiete, I. R.},
  title={Dynamic shift-map coding with side information at the decoder},
  booktitle={Proc. 50th Annual Allerton Conference on Communication, Control, and Computing (Allerton 2012)},
  address={Monticello, IL},
  month={Oct.},
  year={2012}
  }
• O. O. Koyluoglu, K. Appaiah, H. Si, and S. Vishwanath, "Expansion coding: Achieving the capacity of an AEN channel," in Proc. 2012 IEEE International Symposium on Information Theory (ISIT 2012), Boston, MA, Jul. 2012.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  A general method of coding over expansions is proposed, which allows one to reduce the highly non-trivial problem of coding over continuous channels to a much simpler discrete ones. More specifically, the focus is on the additive exponential noise (AEN) channel, for which the (binary) expansion of the (exponential) noise random variable is considered. It is shown that each of the random variables in the expansion corresponds to independent Bernoulli random variables. Thus, each of the expansion levels (of the underlying channel) corresponds to a binary symmetric channel (BSC), and the coding problem is reduced to coding over these parallel channels while satisfying the channel input constraint. This optimization formulation is stated as the achievable rate result, for which a specific choice of input distribution is shown to achieve a rate which is arbitrarily close to the channel capacity in the high SNR regime. Remarkably, the scheme allows for low-complexity capacity-achieving codes for AEN channels, using the codes that are originally designed for BSCs. Extensions to different channel models and applications to other coding problems are discussed.
  @INPROCEEDINGS{Koyluoglu:Expansion12,
  author={Koyluoglu, O. O. and Appaiah, K. and Si, H. and Vishwanath, S.},
  title={Expansion coding: Achieving the capacity of an AEN channel},
  booktitle={2012 IEEE International Symposium on Information Theory (ISIT 2012)},
  address={Boston, MA},
  month={Jul.},
  year={2012}
  }
• M. Fadel, A. Hindy, A. El-Keyi, M. Nafie, O. O. Koyluoglu, and A. M. Tulino, "Resource allocation for throughput enhancement in cellular shared relay networks," in Proc. 35th IEEE Sarnoff Symposium (Sarnoff 2012), Newark, NJ, May 2012.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  The downlink frame of a cellular relay network is considered, where a shared MIMO decode-and-froward relaying is used to serve the users at the edge of the cell. The relay employs zero-forcing beamforming to manage the interference among the mobile stations (MSs) at the edge of the cell. A non-cooperative scheme is considered where there is no coordination between the base stations (BSs) and the relay station (RS), and a power control algorithm for the RS is developed that maximizes the rate of the relayed users. A cooperative setting which allows the coordination of a power allocation between BSs and RSs is also considered. For this setting, based on the proposed achievable scheme, an optimization formulation is derived to maximize the total throughput of the MSs subject to a constraint on the total power of the system. The problem is solved iteratively as a sequence of geometric programs. Simulation results are provided showing that a significant increase in the network throughput can be achieved via the proposed schemes compared to a conventional cellular system with no relays.
  @INPROCEEDINGS{Fadel:Resource12,
  author={Fadel, M. and Hindy, A. and El-Keyi, A. and Nafie, M. and Koyluoglu, O. O. and Tulino, A. M.},
  title={Resource allocation for throughput enhancement in cellular shared relay networks},
  booktitle={35th IEEE Sarnoff Symposium (Sarnoff 2012)},
  address={Newark, NJ},
  month={May},
  year={2012}
  }
• O. O. Koyluoglu and I. R. Fiete, "Information theoretic limits on performance in short-term memory tasks," in 2012 Computational and Systems Neuroscience Abstracts (Cosyne 2012), Salt Lake City, UT, Feb. 2012. (travel grant award)
  [Cosyne 2012] [abstract] [BibTeX]
  Short-term memory is limited. Early notions, that only a fixed number of items can be stored in memory, have been amended by recent experiments showing that memory systems can flexibly allocate memory resources across fewer or more items with a corresponding trade-off in recall accuracy. However, it remains unclear what factors limit working memory, and how it is best modeled to include parameters like item number, storage interval duration, and item complexity.
  Here, we build a normative, or information-theoretically ideal, model for the short-term memory of a set of analog variables. We treat all inputs presented to the memory system on an equal footing, as information. Noisy neural dynamics, that degrade memory over the storage interval, are interpreted as passing the information through a noisy channel. We first show why this mapping is appropriate. Next, we derive fundamental limitations on short-term memory performance using information theoretic results on optimal communication over noisy channels. We show how the resulting performance -- which depends on optimal encoding and decoding of the stored information for the given channel noise -- scales with item number, storage time, and the dynamic range of the inputs. We apply these results to a visual working memory task, and compare predictions for memory performance with psychophysical findings. The functional dependence of recall on the parameters listed above is well-predicted by the normative model, suggesting that the brain performs very good information encoding before the storage period, and decoding after. As we show, these results stand in clear contrast to the predictions made if the brain were directly storing the uncoded variables in continuous attractor networks matched to the dimension and coding range of the variables.
  @INPROCEEDINGS{Koyluoglu:Information12,
  author={Koyluoglu, O. O. and Fiete, I. R.},
  title={Information theoretic limits on performance in short-term memory tasks},
  booktitle={2012 Computational and Systems Neuroscience Abstracts (Cosyne 2012)},
  address={Salt Lake City, UT},
  month={Feb.},
  year={2012}
  }
• S. Vishwanath, O. O. Koyluoglu, H. Si, and K. Appaiah, "Coding over binary expansions," in Proc. 2012 Information Theory and Applications Workshop (ITA 2012), San Diego, CA, Feb. 2012.
  [ITA 2012] [abstract] [BibTeX]
  A general method of coding over expansions is proposed, which allows one to reduce the highly non-trivial problem of coding over continuous channels to much simpler discrete ones. More specifically, the focus is on the additive exponential noise (AEN) channel, for which the (binary) expansion of the (exponential) noise random variable is considered. It is shown that each of the random variables in the expansion corresponds to independent Bernoulli random variables (each having a mean given by a function of the corresponding level number and the mean of the underlying exponential random variable). This way, each of the expansion level (of the underlying channel) corresponds to a binary symmetric channel, and the coding problem is reduced to coding over these parallel channels.
  @INPROCEEDINGS{Vishwanath:Coding12,
  author={Vishwanath, S. and Koyluoglu, O. O. and Si, H. and Appaiah, K.},
  title={Coding over binary expansions},
  booktitle={Proc. 2012 Information Theory and Applications Workshop (ITA 2012)},
  address={San Diego, CA},
  month={Feb.},
  year={2012}
  }
• O. O. Koyluoglu, R. Soundararajan, and S. Vishwanath, "State amplification under masking constraints," in Proc. 49th Annual Allerton Conference on Communication, Control, and Computing (Allerton 2011), Monticello, IL, Sep. 2011.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  This paper considers a state dependent channel with one transmitter, Alice, and two receivers, Bob and Eve. The problem at hand is to effectively “amplify” the channel state sequence at Bob while “masking” it from Eve. The extent to which the state sequence cannot be masked from Eve is referred to as leakage, and the paper is aimed at characterizing the tradeoff-region between amplification and leakage rates for such a system. An achievable coding scheme is presented, wherein the transmitter enumerates the state sequence using two indices, and transmits one of the indices over the channel to facilitate the amplification process. For the case when Bob observes a “stronger” channel than Eve, the achievable coding scheme is enhanced with secure refinement. The optimal amplification-leakage rate difference, called as differential amplification capacity, is characterized for the degraded binary and the degraded Gaussian channels. For the degraded Gaussian model, extremal corner points of the tradeoff region are characterized. In addition, the gap between the outer bound and achievable rate-regions is determined, where it is shown that the gap is less than half a bit for a wide set of channel parameters.
  @INPROCEEDINGS{Koyluoglu:State11a,
  author={Koyluoglu, O. O. and Soundararajan, R. and Vishwanath, S.},
  title={State amplification under masking constraints},
  booktitle={Proc. 49th Annual Allerton Conference on Communication, Control, and Computing (Allerton 2011)},
  address={Monticello, IL},
  month={Sep.},
  year={2011}
  }
• K. Appaiah, O. O. Koyluoglu, and S. Vishwanath, "Polar alignment for interference networks," in Proc. 49th Annual Allerton Conference on Communication, Control, and Computing (Allerton 2011), Monticello, IL, Sep. 2011.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  Polar coding has originally been introduced as a capacity achieving low complexity code for binary input symmetric channels. Polar codes can be understood as transformations that replace a probabilistic channel with parallel deterministic counterparts. This paper builds on this interpretation of polar codes, using it to perform alignment over the resulting deterministic channels to obtain gains for interference networks. It is important to note here that polar codes are not chosen with encoding and decoding complexity in mind, which is just a fortuitous side-benefit, but with the aim of transforming the original channels into a class of deterministic parallel channels over which interference-alignment is well-understood. A degraded one-sided interference network is chosen as the illustrative example. Polar alignment is shown to increase the achievable sum rate over known random coding schemes. The paper concludes with a brief discussion of possible extensions.
  @INPROCEEDINGS{Appaiah:Polar11,
  author={Appaiah, K. and Koyluoglu, O. O. and Vishwanath, S.},
  title={Polar alignment for interference networks},
  booktitle={Proc. 49th Annual Allerton Conference on Communication, Control, and Computing (Allerton 2011)},
  address={Monticello, IL},
  month={Sep.},
  year={2011}
  }
• M. Shahmohammadi, O. O. Koyluoglu, T. Khattab, and H. El Gamal, "On the degrees of freedom of the cognitive broadcast channel," in Proc. 2011 IEEE International Symposium on Information Theory (ISIT 2011), Saint Petersburg, Russia, Jul. 2011.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  Cognitive broadcast channel, where two multiantenna transmitters communicate with their respective receivers, is considered. One of the transmitters is said to be cognitive (secondary) as it is assumed to know the messages of the other (primary) transmitter non-causally. The goal is to design cooperative schemes between the two transmitters, which impose only minimal changes to the primary broadcast channel (compared to the non-cognitive scenario). Towards this end, an achievable scheme is provided under which both intra cell and inter cell interferences at the primary receivers are aligned. The interference at the secondary receivers, on the other hand, is canceled by dirty paper coding. The corresponding achievable region and an outer bound region are provided in terms of the degrees of freedom (DoF) metric. Special cases shows the optimality of the proposed scheme in the high SNR regime for those cases. We also illustrate the advantage of the cognitive cooperation over the non-cognitive system by proving that the achieved sum DoF is strictly larger than the non-cognitive case.
  @INPROCEEDINGS{6033915,
  author={Shahmohammadi, M. and Koyluoglu, O. O. and Khattab, T. and {El Gamal}, H.},
  title={On the degrees of freedom of the cognitive broadcast channel},
  booktitle={Proc. 2011 IEEE International Symposium on Information Theory (ISIT 2011)},
  address={Saint Petersburg, Russia},
  month={Jul.},
  year={2011}
  }
• O. Gungor, O. O. Koyluoglu, H. El Gamal, and C. E. Koksal, "Proactive source coding," in Proc. 2011 IEEE International Symposium on Information Theory (ISIT 2011), Saint Petersburg, Russia, Jul. 2011.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  A coding problem, over a slotted system, is introduced where the sender has to transmit one out of several packets to the receiver, but learns the request only at the beginning of each slot with prior statistical information about which packet is needed at the receiver. There is an associated cost of sending bits at each slot, and the goal is to minimize the expected cost of the communication. A proactive coding scheme is proposed, where the source proactively communicates with the receiver before the receiver requests the message. This way, by designing a cost optimal side information at the receiver, the scheme is able to minimize the expected cost of the communication. Numerical results are provided demonstrating the gains obtained by proactive coding over the conventional coding technique.
  @INPROCEEDINGS{6033953,
  author={Gungor, O. and Koyluoglu, O. O. and {El Gamal}, H. and Koksal, C. E.},
  title={Proactive source coding},
  booktitle={Proc. 2011 IEEE International Symposium on Information Theory (ISIT 2011)},
  address={Saint Petersburg, Russia},
  month={Jul.},
  year={2011}
  }
• M. Shahmohammadi, O. O. Koyluoglu, T. Khattab, and H. El Gamal, "Joint interference cancellation and dirty paper coding for cognitive cellular networks," in Proc. 2011 IEEE Wireless Communications and Networking Conference (WCNC 2011), Cancun, Mexico, Mar. 2011.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  Downlink communication in a cellular network with a cognitive (secondary) cell is considered. In our model, the base station of the cognitive cell knows the messages of the other cell non-causally. We propose a new interference cancellation technique that zero forces the intra-cell interference in the primary cell by the help of the cognitive base station. In addition, as the primary messages are known at the cognitive base station, the interference caused by the primary base station on the secondary users are canceled using dirty paper coding (DPC). Moreover, we provide an outer bound on the achievable degrees of freedom (DoF) region and show that for some special cases the proposed signaling scheme is sum DoF optimal for the considered system when the cognitive cell operates at its maximum sum DoF. The benefit of the cognitive paradigm is also established using the derived outer-bound and showing that the achieved sum DoF is strictly larger than the case when cognitive message sharing is unavailable.
  @INPROCEEDINGS{5779461,
  author={Shahmohammadi, M. and Koyluoglu, O. O. and Khattab, T. and {El Gamal}, H.},
  title={Joint interference cancellation and dirty paper coding for cognitive cellular networks},
  booktitle={Proc. 2011 IEEE Wireless Communications and Networking Conference (WCNC 2011)},
  address={Cancun, Mexico},
  month={Mar.},
  year={2011}
  }
• O. O. Koyluoglu and H. El Gamal, "Polar coding for secure transmission and key agreement," in Proc. 2010 IEEE International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC 2010), Istanbul, Turkey, Sep. 2010.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  Wyner's work on wiretap channels and the recent works on information theoretic security are based on random codes. Achieving information theoretical security with practical coding schemes is of definite interest. In this note, the attempt is to overcome this elusive task by employing the polar coding technique of Arikan. It is shown that polar codes achieve nontrivial perfect secrecy rates for binary-input degraded wiretap channels while enjoying their low encoding-decoding complexity. In the special case of symmetric main and eavesdropper channels, this coding technique achieves the secrecy capacity. Next, fading erasure wiretap channels are considered and a secret key agreement scheme is proposed, which requires only the statistical knowledge of the eavesdropper channel state information (CSI). The enabling factor is the creation of advantage over Eve, by blindly using the proposed scheme over each fading block, which is then exploited with privacy amplification techniques to generate secret keys.
  @INPROCEEDINGS{5779461,
  author={Koyluoglu, O. O. and {El Gamal}, H.},
  title={Polar coding for secure transmission and key agreement},
  booktitle={Proc. 2010 IEEE International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC 2010)},
  address={Istanbul, Turkey},
  month={Sep.},
  year={2010}
  }
• E. Toher, O. O. Koyluoglu, and H. El Gamal, "Secrecy games over the cognitive channel," in Proc. 2010 IEEE International Symposium on Information Theory (ISIT 2010), Austin, TX, Jun. 2010.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  A secure communication game is considered for the cognitive channel with a confidential primary message, where the primary user is interested in maximizing its secure rate with lowest possible power consumption and the utility of the cognitive user is a weighted sum of the primary secrecy rate and the cognitive rate (corresponds to a spectrum law in favor of the legacy owners of the spectrum). An achievable rate region is derived for the channel with message splitting at the cognitive radio and noise forwarding. The game considers the case with no common message, but shows that even this limited scenario can still be beneficial. The established Nash Equilibrium (NE) shows that the cognitive user trades noise for bits. The results are also interesting in the sense that both users can benefit (by playing the distributed game) compared to their throughput resulting from the non-cooperative scenario.
  @INPROCEEDINGS{5513708,
  author={Toher, E. and Koyluoglu, O. O. and {El Gamal}, H.},
  title={Secrecy games over the cognitive channel},
  booktitle={Proc. 2010 IEEE International Symposium on Information Theory (ISIT 2010)},
  address={Austin, TX},
  month={Jun.},
  year={2010}
  }
• O. O. Koyluoglu, C. E. Koksal, and H. El Gamal, "On the effect of colluding eavesdroppers on secrecy capacity scaling," in Proc. European Wireless 2010 (EW 2010), Lucca, Italy, Apr. 2010.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  In a powerful secrecy attack, eavesdroppers can collude, i.e., they can share their observations. Securing information in such a scenario will be an even more challenging task compared to non-colluding case. We here analyze the effect of eavesdropper collusion on the achievable performance in both the path loss and ergodic multi-path fading models. We provide two results: 1) For the Poisson point process model in a random extended network, if the legitimate nodes have unit intensity ($\lambda=1$) and the colluding eavesdroppers have an intensity of $\lambda_e = O((log n)^{-(2+p)})$ for any $p>0$, almost all of the nodes can achieve a secure rate of $\Omega(1/(\sqrt{n}))$ and 2) In the K-user Gaussian interference channel with E external colluding eavesdroppers, a secure degrees of freedom of $\eta=[1/2-E/K]^+$ per frequency-time slot is achievable for each user in the ergodic setting (in the absence of the eavesdropper channel state information).
  @INPROCEEDINGS{5483463,
  author={Koyluoglu, O. O. and Koksal, C. E. and {El Gamal}, H.},
  title={On the effect of colluding eavesdroppers on secrecy capacity scaling},
  booktitle={Proc. European Wireless 2010 (EW 2010)},
  address={Lucca, Italy},
  month={Apr.},
  year={2010}
  }
• O. O. Koyluoglu, C. E. Koksal, and H. El Gamal, "On secrecy capacity scaling in wireless networks," in Proc. 2010 Information Theory and Applications Workshop (ITA 2010), La Jolla, CA, Feb. 2010.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  We study a random extended network, where the legitimate and eavesdropper nodes are assumed to be placed according to Poisson point processes in a square region of area n. It is shown that, when the legitimate nodes have unit intensity, $\lambda=1$, and the eavesdroppers have an intensity of $\lambda_e=O((log n)^{-2})$, almost all of the nodes achieve a perfectly secure rate of $\Omega{1/\sqrt{n}}$. The achievability argument is based on a novel multi-hop forwarding scheme where randomization is added in every hop to ensure maximal ambiguity at the eavesdropper(s). Remarkable, under these assumptions, securing the transmissions of nodes does not entail a loss in the per-node throughput in terms of scaling.
  @INPROCEEDINGS{5454119,
  author={Koyluoglu, O. O. and Koksal, C. E. and {El Gamal}, H.},
  title={On secrecy capacity scaling in wireless networks},
  booktitle={Proc. 2010 Information Theory and Applications Workshop (ITA 2010)},
  address={La Jolla, CA},
  month={Feb.},
  year={2010}
  }
• A. El Gamal, O. O. Koyluoglu, M. Youssef, and H. El Gamal, "New achievable secrecy rate regions for the two way wiretap channel," in Proc. 2010 IEEE Information Theory Workshop (ITW 2010), Cairo, Egypt, Jan. 2010.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  This work develops new achievable rate regions for the two way wiretap channel. In our setup, Alice and Bob wish to exchange messages securely in the presence of a passive eavesdropper Eve. In the full-duplex scenario, our achievability argument relies on allowing the two users to jointly optimize their channel prefixing distributions, such that the new channel conditions are favorable compared to that of Eve. Random binning and private key sharing over the channel are then used to exploit the secrecy advantage available in the equivalent cascade channel and to distribute the available secrecy rate among users. For the half-duplex case, we introduce the idea of randomized scheduling and establish the significant gain it offers in terms of the achievable secrecy sum-rate. We also quantify the gains that can be leveraged from the proposed schemes in the modulo-2 and Gaussian channels via numerical results in certain selected scenarios.
  @INPROCEEDINGS{5503136,
  author={El Gamal, A. and Koyluoglu, O. O. and Youssef, M. and {El Gamal}, H.},
  title={New achievable secrecy rate regions for the two way wiretap channel},
  booktitle={Proc. 2010 IEEE Information Theory Workshop (ITW 2010)},
  address={Cairo, Egypt},
  month={Jan.},
  year={2010}
  }
• K. Khalil, M. Youssef, O. O. Koyluoglu, and H. El Gamal, "On the delay limited secrecy capacity of fading channels," in Proc. 2009 IEEE International Symposium on Information Theory (ISIT 2009), Seoul, Korea, Jun. 2009.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  In this paper, the delay limited secrecy capacity of the flat fading channel is investigated under two different assumptions on the available transmitter channel state information (CSI). The first scenario assumes perfect prior knowledge of both the main and eavesdropper channel gains. Here, upper and lower bounds on the secure delay limited capacity are derived and shown to be tight in the high signal-to-noise ratio (SNR) regime (for a wide class of channel distributions). In the second scenario, only the main channel CSI is assumed to be available at the transmitter. Remarkably, under this assumption, we establish the achievability of non-zero secure rate (for a wide class of channel distributions) under a strict delay constraint. In the two cases, our achievability arguments are based on a novel two-stage approach that overcomes the secrecy outage phenomenon observed in earlier works.
  @INPROCEEDINGS{5205955,
  author={Khalil, K. and Youssef, M. and Koyluoglu, O. O. and {El Gamal}, H.},
  title={On the delay limited secrecy capacity of fading channels},
  booktitle={Proc. 2009 IEEE International Symposium on Information Theory (ISIT 2009)},
  address={Seoul, Korea},
  month={Jun.},
  year={2009}
  }
• O. O. Koyluoglu, M. Shahmohammadi, and H. El Gamal, "A new achievable rate region for the discrete memoryless X channel," in Proc. 2009 IEEE International Symposium on Information Theory (ISIT 2009), Seoul, Korea, Jun. 2009.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  We consider the discrete memoryless X channel, a communication model with two transmitters and two receivers in which every transmitter has a message for every receiver. We propose an achievable scheme, based on the message splitting and binning techniques, which results in the best inner bound on the capacity region of the X channel to date.
  @INPROCEEDINGS{5206034,
  author={Koyluoglu, O. O. and Shahmohammadi, M. and {El Gamal}, H.},
  title={A new achievable rate region for the discrete memoryless X channel},
  booktitle={Proc. 2009 IEEE International Symposium on Information Theory (ISIT 2009)},
  address={Seoul, Korea},
  month={Jun.},
  year={2009}
  }
• O. O. Koyluoglu and H. El Gamal, "On the secrecy rate region for the interference channel," in Proc. 2008 IEEE International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC 2008), Cannes, France, Sep. 2008.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  This paper studies interference channels with security constraints. The existence of an external eavesdropper in a two-user interference channel is assumed, where the network users would like to secure their messages from the external eavesdropper. The cooperative binning and channel prefixing scheme is proposed for this system model which allows users to cooperatively add randomness to the channel in order to degrade the observations of the external eavesdropper. This scheme allows users to add randomness to the channel in two ways: 1) Users cooperate in their design of the binning codebooks, and 2) Users cooperatively exploit the channel prefixing technique. As an example, the channel prefixing technique is exploited in the Gaussian case to transmit a superposition signal consisting of binning codewords and independently generated noise samples. Gains obtained form the cooperative binning and channel prefixing scheme compared to the single user scenario reveals the positive effect of interference in increasing the network security. Remarkably, interference can be exploited to cooperatively add randomness into the network in order to enhance the security.
  @INPROCEEDINGS{4699954,
  author={Koyluoglu, O. O. and {El Gamal}, H.},
  title={On the secrecy rate region for the interference channel},
  booktitle={Proc. 2008 IEEE International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC 2008)},
  address={Cannes, France},
  month={Sep.},
  year={2008}
  }
• O. O. Koyluoglu, H. El Gamal, L. Lai, and H. V. Poor, "On the secure degrees of freedom in the K-user Gaussian interference channel," in Proc. 2008 IEEE International Symposium on Information Theory (ISIT 2008), Toronto, ON, Canada, Jul. 2008.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  This paper studies the K-user Gaussian interference channel with secrecy constraints. Two distinct network models, namely the interference channel with confidential messages and the one with an external eavesdropper, are analyzed. Using interference alignment along with secrecy pre-coding at each transmitter, it is shown that each user in the network can achieve non-zero secure degrees of freedoms (DoFs) in both scenarios. In particular, the proposed coding scheme achieves K-2/2K-2 secure DoFs for each user in the interference channel with confidential messages model, and K-2/2K secure DoFs in the case of an external eavesdropper. The fundamental difference between the two scenarios stems from the lack of channel state information (CSI) about the external eavesdropper. Remarkably, the results establish the positive impact of interference on the secrecy capacity of wireless networks.
  @INPROCEEDINGS{4595013,
  author={Koyluoglu, O. O. and {El Gamal}, H. and Lifeng Lai and Poor, H. V.},
  title={On the secure degrees of freedom in the K-user Gaussian interference channel},
  booktitle={Proc. 2008 IEEE International Symposium on Information Theory (ISIT 2008)},
  address={Toronto, ON, Canada},
  month={Jul.},
  year={2008}
  }
• O. O. Koyluoglu and H. El Gamal, "On the utility of frequency reuse in cognitive radio channels," in Proc. 2007 IEEE International Symposium on Information Theory (ISIT 2007), Nice, France, Jun. 2007.
  [IEEE Xplore] [pdf preprint] [abstract] [BibTeX]
  We consider the generalized cognitive radio channel where the secondary user is allowed to reuse the frequency during the active periods of the primary user, as long as the primary rate remains the same. In this setting, the optimal power allocation policy with a single antenna secondary transmitter (and receiver) is explored. Interestingly, we show that the offered gain resulting from the frequency reuse during the active periods of the spectrum disappears in both the low and high signal-to-noise ratio (SNR) regimes. This drawback, however, is shown to disappear with multi-antenna nodes by using simple zero-forcing strategies at both ends of the secondary channel.
  @INPROCEEDINGS{4557540,
  author={Koyluoglu, O. O. and {El Gamal}, H.},
  title={On the utility of frequency reuse in cognitive radio channels},
  booktitle={Proc. 2007 IEEE International Symposium on Information Theory (ISIT 2007)},
  address={Nice, France},
  month={Jun.},
  year={2007}
  }

PhD Dissertation

• O. O. Koyluoglu, "Wireless physical layer security: An information theoretic approach," PhD dissertation, The Ohio State University, Dec. 2010.
  [OhioLINK] [abstract] [BibTeX]
  We are in the midst of wireless revolution, and increasing demand continues for wireless applications. This explosive growth, of wireless communications and services, inevitably renders security into a challenging quality of service constraint that must be accounted for in the network design. The state of the art methods in combating the security threats are usually founded on cryptographic approaches. These techniques typically assume limited computational resources at adversaries, are usually derived from unproven assumptions, and most of the time do not offer a measurable security notion. Information theoretic security, on the other hand, eliminates the aforementioned limitations of the cryptographic techniques at the physical layer of communication systems.
  In this thesis, we concentrate on both the theoretical and the practical aspects of physical layer security. We first start by analyzing elemental interference networks, in particular, two-user channels with an adversary. The problem here is to characterize the fundamental limits on secure transmission rates. Towards this end, we devise coding schemes, forming inner bounds to the capacity region, and compare the achievable rates with outer bounds. This analysis is useful to explore microscopic gains that can be leveraged by the different coding schemes, and our analysis shows that the inherent interference can in fact be exploited to cooperatively confuse the eavesdropping node.
  We secondly focus on large networks. As the users interfere, the problem here is to identify the optimal ways of distributing network resources among users under the secrecy constraints. Initially, we consider a network with finite number of nodes in the high signal to noise ratio regime, where the noise impairing each receiver is deemphasized compared to the signal strengths. In this asymptotic scenario, we propose efficient interference alignment techniques along with secrecy pre-coding allowing each user to achieve positive secure degrees of freedom. Next, we investigate the secrecy capacity scaling, i.e., the secure rate per user as the number of users gets large. We propose a novel multi-hop forwarding scheme, and, utilizing tools from the percolation theory, we show that each user can achieve the optimal secure rate scaling under a mild condition on the eavesdropper density.
  The final attempt is on the design of practical coding schemes for physical layer security. Specifically, we show that polar codes achieve non-trivial perfect secrecy rates for a fairly large set of channels with a low encoding and decoding complexity. We also propose a secret key agreement scheme over fading channels. This technique essentially combines physical layer coding schemes with cryptographic methods, and, establishes key agreement by only requiring the statistical knowledge of the eavesdropper channel state information.
  These results are encouraging in the sense that a) interference can be exploited to enhance security, b) not only small sized networks, but also large networks can achieve information theoretic security, c) there exists practical coding schemes achieving secrecy for a fairly large set of channels, and d) physical layer security can be implemented together with higher layer cryptographic protocols, and hence, allows for cross-layer security solutions.
  @PHDTHESIS{Koyluoglu:Wireless10,
  author={Koyluoglu, O. O.},
  title={Wireless physical layer security: An information theoretic approach},
  school={The Ohio State University},
  year={2010}
  }