- Cooperative Communication and Network Coding
- Wireless Communication Security Techniques
- Error Correcting Code Techniques
- Cellular Automata and Applications
- Caching and Content Delivery
- Advanced Data Storage Technologies
- Advanced MIMO Systems Optimization
- DNA and Biological Computing
- Chaos-based Image/Signal Encryption
- Advanced Wireless Network Optimization
- Advanced Wireless Communication Technologies
- Cryptography and Data Security
- Advanced Data Compression Techniques
- Energy Harvesting in Wireless Networks
- Stochastic Gradient Optimization Techniques
- Sparse and Compressive Sensing Techniques
- Smart Grid Security and Resilience
- Algorithms and Data Compression
- Distributed Sensor Networks and Detection Algorithms
- Advanced Wireless Communication Techniques
- Wireless Signal Modulation Classification
- Privacy-Preserving Technologies in Data
- Distributed systems and fault tolerance
- Image and Signal Denoising Methods
- Coding theory and cryptography
University of Toronto
2015-2024
Qualcomm (Canada)
2024
Institute of Electrical and Electronics Engineers
2017-2018
University of Illinois Chicago
2017-2018
University of Maryland, College Park
2017-2018
Monash University
2017-2018
New York University
2018
École Polytechnique Fédérale de Lausanne
2017
Massachusetts Institute of Technology
2004-2010
IIT@MIT
2006-2008
Cooperative diversity has been recently proposed as a way to form virtual antenna arrays that provide dramatic gains in slow fading wireless environments. However, most of the solutions require distributed space-time coding algorithms, careful design which is left for future investigation if there more than one cooperative relay. We propose novel scheme alleviates these problems and provides on order number relays network. Our first selects best relay from set M available then uses this...
The role of multiple antennas for secure communication is investigated within the framework Wyner's wiretap channel. We characterize secrecy capacity in terms generalized eigenvalues when sender and eavesdropper have antennas, intended receiver has a single antenna, channel matrices are fixed known to all terminals, show that beamforming strategy capacity-achieving. In addition, we study masked scheme radiates power isotropically directions it attains near-optimal performance high SNR...
The capacity of the Gaussian wiretap channel model is analyzed when there are multiple antennas at sender, intended receiver and eavesdropper. associated matrices fixed known to all terminals. A computable characterization secrecy established as saddle point solution a minimax problem. converse based on Sato-type argument used in other broadcast settings, coding theorem codebooks. At high signal-to-noise ratio (SNR), shown be attained by simultaneously diagonalizing via generalized singular...
Physical layer security which safeguards data confidentiality based on the information-theoretic approaches has received significant research interest recently. The key idea behind physical is to utilize intrinsic randomness of transmission channel guarantee in layer. evolution toward 5G wireless communications poses new challenges for research. This paper provides a latest survey various promising technologies, including coding, massive multiple-input multiple-output, millimeter wave...
We study a problem of broadcasting confidential messages to multiple receivers under an information-theoretic secrecy constraint. Two scenarios are considered: 1) all obtain common message; and 2) each receiver is independent message. Moreover, two models parallel channels fast-fading channels. For the case reversely degraded channels, one eavesdropper, arbitrary number legitimate receivers, we determine capacity for transmitting message, sum-capacity messages. assume that channel state...
In this paper, we investigate the problem of optimal content cache management for HTTP adaptive bit rate (ABR) streaming over wireless networks. Specifically, in media cloud, each is transcoded into a set files with diverse playback rates, and appropriate will be dynamically chosen response to channel conditions screen forms. Our design objective maximize quality experience (QoE) an individual end users, under limited storage budget. Deriving logarithmic QoE model from our experimental...
Wyner's wiretap channel is generalized to the case when sender, receiver and eavesdropper have multiple antennas. We consider two cases: deterministic fading case. In case, matrices of intended are fixed known all nodes. experience block sender has only receiver's state information (CSI) statistical knowledge eavesdropper's channel. For a scheme based on generalized-singular-value-decomposition (GSVD) proposed shown achieve secrecy capacity in high signal-to-noise-ratio (SNR) limit. When one...
A rechargeable battery may be used to partially protect the privacy of information contained in a household's electrical load profile. We represent system as finite state model make tractable computation rate leakage. Specifically, we use trellis algorithm estimate mutual between battery's input and output loads. show that stochastic policies can leak 26% less than so-called best-effort (that holds constant whenever possible). finally describe extension technique more realistic models system.
In Costa's dirty-paper channel, Gaussian random binning is able to eliminate the effect of interference which known at transmitter, and thus achieve capacity. We examine a generalization problem multiple access channel (MAC) setup, where structured (lattice-based) seems be necessary dirty-MAC, two additive signals are present, one each transmitter but none receiver. The achievable rates using vanish if both strong. contrast, it shown that lattice-strategies (“lattice precoding”) can positive...
Practical cooperative diversity protocols often rely on low-cost radios that treat multiple in-band signals as noise and thus require strictly orthogonal transmissions. We analyze the performance of a class opportunistic relaying employ simple packet level feedback It is shown diversity-multiplexing tradeoff proposed either matches or outperforms multi-input-single-output (MISO), zero-feedback performance. These gains indicate low complexity could be an appealing architecture for future user...
We study a wiretap channel model where the sender has <formula formulatype="inline" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex Notation="TeX">$M $</tex></formula> transmit antennas and there are two groups consisting of Notation="TeX">$J_{1}$</tex></formula> Notation="TeX">$J_{2}$</tex> </formula> receivers respectively. Each receiver single antenna. consider scenarios. First we compound — group 1 constitutes set legitimate receivers, all...
We propose a new class of error correction codes for low-delay streaming communication. consider an online setup where source packet arrives at the encoder every $M$ channel uses, and needs to be decoded with maximum delay $T$ packets. sliding-window erasure --- $\cC(N,B,W)$ which introduces either up $N$ erasures in arbitrary positions, or $B$ single burst, any window length $W$. When $M=1$, case source-arrival channel-transmission rates are equal, we MiDAS that achieve near optimal rate....
Smart-metering systems report electricity usage of a user to the utility provider on almost real-time basis. This could leak private information about provider. In this paper, we investigate use rechargeable battery in order provide privacy user. We assume that load sequence is first-order Markov process, satisfies ideal charge conservation, and measured using normalized mutual (leakage rate) between output. study optimal charging policy minimizes leakage rate among class policies satisfy...
This paper considers transmitting a sequence of messages (streaming messages) over packet erasure channel. In each time slot, the source constructs based on current and previous transmits packet, which may be erased when travels from to destination. Every message must recovered perfectly at destination subject fixed decoding delay. We assume that channel loss model introduces either one burst or multiple arbitrary erasures in any fixed-sized sliding window. Under this assumption, we fully...
We consider the problem of covert communication over a state-dependent channel, where transmitter has causal or noncausal knowledge channel states. Here, means that warden on should observe similar statistics when is sending message and it not. When sufficiently long secret key shared between receiver, we derive closed-form formulas for maximum achievable rate (covert capacity) discrete memoryless channels and, transmitter's channel-state information (CSI) noncausal, additive white Gaussian...
Given the ubiquity of deep neural networks, it is important that these models do not reveal information about sensitive data they have been trained on. In model inversion attacks, a malicious user attempts to recover private dataset used train supervised network. A successful attack should generate realistic and diverse samples accurately describe each classes in dataset. this work, we provide probabilistic interpretation formulate variational objective accounts for both diversity accuracy....
We describe a method to encode fingerprint biometrics securely for use, e.g., in encryption or access control. The system is secure because the stored data does not suffice recreate original biometric. Therefore, breach database security lead loss of biometric data. At same time suffices validate probe fingerprint. Our approach based on use distributed source coding techniques implemented with graph-based codes. present statistical model relationship between enrollment and (noisy)...
A two-transmitter Gaussian multiple access wiretap channel with antennas at each of the nodes is investigated. The matrices legitimate users are fixed and revealed to all terminals, whereas eavesdropper arbitrarily varying only known eavesdropper. secrecy degrees freedom (s.d.o.f.) region under a strong constraint characterized. transmission scheme that orthogonalizes transmit signals two intended receiver, uses single-user code for user, shown achieve s.d.o.f. region. converse involves...
We study low-delay error correction codes for streaming recovery over a class of packet-erasure channels that introduce both burst-erasures and isolated erasures. propose simple, yet effective whose parameters can be tuned to obtain tradeoff between the capability correct burst Our construction generalizes previously proposed which are only against establish an information theoretic upper bound on any code simultaneously erasures show our constructions meet in some special cases. discuss...
We study the secret-key capacity in a joint source-channel coding setup-the terminals are connected over discrete memoryless channel and have access to side information, modelled as pair of source sequences. As our main result, we establish upper lower bounds on capacity. In bound expression, equivocation terms components functionally additive even though scheme generates single by jointly taking into account equivocations. Our coincide, thus establishing capacity, when underlying wiretap...
We study the capacity of secret-key agreement over a wiretap channel with state parameters. The transmitter communicates to legitimate receiver and eavesdropper discrete memoryless sequence. generate shared secret key, that remains from eavesdropper. No public discussion is available. sequence known noncausally transmitter. derive lower upper bounds on capacity. bound involves constructing common reconstruction at terminals binning set sequences obtain secret-key. For special case Gaussian...
Many current and emerging applications require low-latency communication, including interactive voice video multiplayer gaming, multiperson augmented/virtual reality, various Internet of Things (IoT) applications. Forward error correction (FEC) codes for low-delay have several characteristics that distinguish them from traditional FEC. The encoding decoding operations must process a stream data packets in sequential fashion. Strict latency constraints limit the use long block lengths,...
The information-theoretic framework of Russo and J. Zou (2016) Xu Raginsky (2017) provides bounds on the generalization error a learning algorithm in terms mutual information between algorithm's output training sample. In this work, we study proposal, by Steinke Zakynthinou (2020), to reason about introducing super sample that contains as random subset computing conditional We first show these new based are tighter than those unconditional information. then introduce yet bounds, building...