Search

Federated learning (FL) allows multiple parties (distributed devices) to train a machine learning model without sharing raw data. How to effectively and efficiently utilize the resources on devices and the central server is a highly interesting yet challenging problem. In this paper, we propose an efficient split federated learning algorithm (ESFL) to take full advantage of the powerful computing capabilities at a central server under a split federated learning framework with heterogeneous end devices (EDs). By splitting the model into different submodels between the server and EDs, our approach jointly optimizes user-side workload and server-side computing resource allocation by considering users' heterogeneity. We formulate the whole optimization problem as a mixed-integer non-linear program, which is an NP-hard problem, and develop an iterative approach to obtain an approximate solution efficiently. Extensive simulations have been conducted to validate the significantly increased efficiency of our ESFL approach compared with standard federated learning, split learning, and splitfed learning.

This paper presents a privacy-preserving event detection scheme based on measurements made by a network of sensors. A diameter-like decision statistic made up of the marginal types of the measurements observed by the sensors is employed. The proposed detection scheme can achieve the best type-I error exponent as the type-II error rate is required to be negligible. Detection performance with finite-length observations is also demonstrated through a simulation example of spectrum sensing. Privacy protection is achieved by obfuscating the type data with random zero-modulo-sum numbers that are generated and distributed via the exchange of encrypted messages among the sensors. The privacy-preserving performance against ``honest but curious'' adversaries, including colluding sensors, the fusion center, and external eavesdroppers, is analyzed through a series of cryptographic games. It is shown that the probability that any probabilistic polynomial time adversary successfully estimates the sensors' measured types can not be much better than independent guessing, when there are at least two non-colluding sensors., Comment: 26 pages, 9 figures, submitted to IEEE Transactions on Information Theory

We consider a channel separation approach to counter the pilot attack in a massive MIMO system, where malicious users (MUs) perform pilot spoofing and jamming attack (PSJA) in uplink by sending symbols to the basestation (BS) during the channel estimation (CE) phase of the legitimate users (LUs). More specifically, the PSJA strategies employed by the MUs may include (i) sending the random symbols according to arbitrary stationary or non-stationary distributions that are unknown to the BS; (ii) sending the jamming symbols that are correlative to those of the LUs. We analyze the empirical distribution of the received pilot signals (ED-RPS) at the BS, and prove that its characteristic function (CF) asymptotically approaches to the product of the CFs of the desired signal (DS) and the noise, where the DS is the product of the channel matrix and the signal sequences sent by the LUs/MUs. These observations motivate a novel two-step blind channel separation method, wherein we first estimate the CF of DS from the ED-RPS and then extract the alphabet of the DS to separate the channels. Both analysis and simulation results show that the proposed method achieves good channel separation performance in massive MIMO systems.

This work constructs a discrete random variable that, when conditioned upon, ensures information stability of quasi-images. Using this construction, a new methodology is derived to obtain information theoretic necessary conditions directly from operational requirements. In particular, this methodology is used to derive new necessary conditions for keyed authentication over discrete memoryless channels and to establish the capacity region of the wiretap channel, subject to finite leakage and finite error, under two different secrecy metrics. These examples establish the usefulness of the proposed methodology., Comment: Submitted for possible publication

The fundamental lower size limit in wavelength selecting structures is explored with the aid of the Heisenberg uncertainty principle. The analysis shows that for a given wavelength selectivity resonating structures with optical feedback have the smallest dimensions. Experimental results obtained with integrated optics microring resonators (Q ~ 3.4 x 10^4) confirm the analysis. In addition, a discussion is given on the validity of the uncertainty principle in terms of hidden variables or restricted knowledge on the system in question., Comment: This is the first of a collection of four studies dealing with the weird properties of a photon, compiled in 2017 by Alfred Driessen. The original version of 2003 has not yet been published and the current version is essentially unchanged

This paper employs equal-image-size source partitioning techniques to derive the capacities of the general discrete memoryless wiretap channel (DM-WTC) under four different secrecy criteria. These criteria respectively specify requirements on the expected values and tail probabilities of the differences, in absolute value and in exponent, between the joint probability of the secret message and the eavesdropper's observation and the corresponding probability if they were independent. Some of these criteria reduce back to the standard leakage and variation distance constraints that have been previously considered in the literature. The capacities under these secrecy criteria are found to be different when non-vanishing error and secrecy tolerances are allowed. Based on these new results, we are able to conclude that the strong converse property generally holds for the DM-WTC only under the two secrecy criteria based on constraining the tail probabilities. Under the secrecy criteria based on the expected values, an interesting phase change phenomenon is observed as the tolerance values vary., Comment: Published in proceedings of ISIT2017, contains a correction in result statements

We consider an amplify-and-forward relay network composed of a source, two relays, and a destination. In this network, the two relays are untrusted in the sense that they may perform Byzantine attacks by forwarding altered symbols to the destination. Note that every symbol received by the destination may be altered, and hence no clean reference observation is available to the destination. For this network, we identify a large family of Byzantine attacks that can be detected in the physical layer. We further investigate how the channel conditions impact the detection against this family of attacks. In particular, we prove that all Byzantine attacks in this family can be detected with asymptotically small miss detection and false alarm probabilities by using a sufficiently large number of channel observations \emph{if and only if} the network satisfies a non-manipulability condition. No pre-shared secret or secret transmission is needed for the detection of these attacks, demonstrating the value of this physical-layer security technique for counteracting Byzantine attacks., Comment: 16 pages, 7 figures, accepted to appear on IEEE Transactions on Information Forensics and Security, July 2016

This paper introduces equal-image-size source partitioning, a new tool for analyzing channel and joint source-channel coding in a multi-terminal discrete memoryless channel environment. Equal-image-size source partitioning divides the source (combination of messages and codewords) into a sub-exponential number of subsets. Over each of these subsets, the exponential orders of the minimum image sizes of most messages are roughly equal to the same entropy term. This property gives us the strength of minimum image sizes and the flexibility of entropy terms. Using the method of equal-image-size source partitioning, we prove separate necessary conditions for the existence of average-error and maximum-error codes. These necessary conditions are much stronger than the standard Fano's inequality, and can be weakened to render versions of Fano's inequality that apply to codes with non-vanishing error probabilities. To demonstrate the power of this new tool, we employ the stronger average-error version of Fano's inequality to prove the strong converse for the discrete memoryless wiretap channel with decaying leakage, which heretofore has been an open problem., Comment: Submitted to IEEE trans info theory. Presented in part at the 2014 International Symposium on Information Theory (ISIT) in Honolulu, HI, USA. 29 pages, 1 example. Comments welcome. This version fixes a mistake in the application example and corrects a number of typos

In this paper we investigate the image size characterization problem. We show that any arbitrary source set may be decomposed into sets whose image size characterization is the same as its entropy characterization. We also show that the number of these sets required is small enough that one may consider that from a coding perspective the achievable entropy region and achievable exponent region are equal. This has an impact on many source networks and network problems whose solution heretofore could not have the image size characterization applied to them., Comment: 5 Pages, ISIT 14

A coding scheme based on irregular low-density parity-check (LDPC) codes is proposed to send secret messages from a source over the Gaussian wiretap channel to a destination in the presence of a wiretapper, with the restriction that the source can send only binary phase-shift keyed (BPSK) symbols. The secrecy performance of the proposed coding scheme is measured by the secret message rate through the wiretap channel as well as the equivocation rate about the message at the wiretapper. A code search procedure is suggested to obtain irregular LDPC codes that achieve good secrecy performance in such context., Comment: submitted to IEEE GLOBECOM 2011 - Communication Theory Symposium

The problem of secret sharing over the Gaussian wiretap channel is considered. A source and a destination intend to share secret information over a Gaussian channel in the presence of a wiretapper who observes the transmission through another Gaussian channel. Two constraints are imposed on the source-to-destination channel; namely, the source can transmit only binary phase shift keyed (BPSK) symbols, and symbol-by-symbol hard-decision quantization is applied to the received symbols of the destination. An error-free public channel is also available for the source and destination to exchange messages in order to help the secret sharing process. The wiretapper can perfectly observe all messages in the public channel. It is shown that a secret sharing scheme that employs a random ensemble of regular low density parity check (LDPC) codes can achieve the key capacity of the BPSK-constrained Gaussian wiretap channel asymptotically with increasing block length. To accommodate practical constraints of finite block length and limited decoding complexity, fixed irregular LDPC codes are also designed to replace the regular LDPC code ensemble in the proposed secret sharing scheme., Comment: submitted to IEEE Trans. on Information Forensics and Security, special issues on using the physical layer for securing the next generation of communication systems

Coding for the causal cognitive radio channel, with the cognitive source subjected to a half-duplex constraint, is studied. A discrete memoryless channel model incorporating the half-duplex constraint is presented, and a new achievable rate region is derived for this channel. It is proved that this rate region contains the previously known causal achievable rate region of \cite{Devroye06} for Gaussian channels., Comment: 5 pages, 2 figures, ISIT 2010

In this paper, a cooperative transmission design for a general multi-node half-duplex wireless relay network is presented. It is assumed that the nodes operate in half-duplex mode and that channel information is available at the nodes. The proposed design involves solving a convex flow optimization problem on a graph that models the relay network. A much simpler generalized-link selection protocol based on the above design is also presented. Both the proposed flow-optimized protocol and the generalized-link selection protocol are shown to achieve the optimal diversity-multiplexing tradeoff (DMT) for the relay network. Moreover, simulation results are presented to quantify the gap between the performances of the proposed protocols and that of a max-flow-min-cut type bound, in terms of outage probability., Comment: 31 pages, 10 figures. First version submitted to Transactions on Communications, Nov. 2007. This is the revised detailed version. Further updated on Nov. 16, 2009 with minor modifications. To appear

Secret sharing over the fast-fading MIMO wiretap channel is considered. A source and a destination try to share secret information over a fast-fading MIMO channel in the presence of a wiretapper who also makes channel observations that are different from but correlated to those made by the destination. An interactive authenticated unrestricted public channel is also available for use by the source and destination in the secret sharing process. This falls under the "channel-type model with wiretapper" considered by Ahlswede and Csiszar. A minor extension of their result (to continuous channel alphabets) is employed to evaluate the key capacity of the fast-fading MIMO wiretap channel. The effects of spatial dimensionality provided by the use of multiple antennas at the source, destination, and wiretapper are then investigated., Comment: Revision submitted to EURASIP Journal on Wireless Communications and Networking, Special Issue on Wireless Physical Layer Security, Sept. 2009. v.3: Fixes to proofs. Matthieu Bloch added as co-author for contributions to proofs

A discrete memoryless half-duplex relay channel is constructed from a broadcast channel from the source to the relay and destination and a multiple access channel from the source and relay to the destination. When the relay listens, the channel operates in the broadcast mode. The channel switches to the multiple access mode when the relay transmits. If the broadcast component channel is physically degraded, the half-duplex relay channel will also be referred to as physically degraded. The capacity of this degraded half-duplex relay channel is examined. It is shown that the block Markov coding suggested in the seminal paper by Cover and El Gamal can be modified to achieve capacity for the degraded half-duplex relay channel. In the code construction, the listen-transmit schedule of the relay is made to depend on the message to be sent and hence the schedule carries information itself. If the schedule is restricted to be deterministic, it is shown that the capacity can be achieved by a simple management of information flows across the broadcast and multiple access component channels., Comment: 10 pages, 2 figures

This paper describes an approach for half-duplex cooperative transmission in a classical three-node relay channel. Assuming availability of channel state information at nodes, the approach makes use of this information to optimize distinct flows through the direct link from the source to the destination and the path via the relay, respectively. It is shown that such a design can effectively harness diversity advantage of the relay channel in both high-rate and low-rate scenarios. When the rate requirement is low, the proposed design gives a second-order outage diversity performance approaching that of full-duplex relaying. When the rate requirement becomes asymptotically large, the design still gives a close-to-second-order outage diversity performance. The design also achieves the best diversity-multiplexing tradeoff possible for the relay channel. With optimal long-term power control over the fading relay channel, the proposed design achieves a delay-limited rate performance that is only 3.0dB (5.4dB) worse than the capacity performance of the additive white Gaussian channel in low- (high-) rate scenarios., Comment: Submitted to Transactions on Information Theory, December 2006. 34 pages, 4 figures. Correction, March 26, 2007. Revision, September 13, 2007: Revised text and provided additional proofs regarding diversity-multiplexing tradeoffs and delay-limited rates with optimal long-term power control