Publications (97)65.39 Total impact

Article: Coded MapReduce
[Show abstract] [Hide abstract]
ABSTRACT: MapReduce is a commonly used framework for executing dataintensive jobs on distributed server clusters. We introduce a variant implementation of MapReduce, namely "Coded MapReduce", to substantially reduce the interserver communication load for the shuffling phase of MapReduce, and thus accelerating its execution. The proposed Coded MapReduce exploits the repetitive mapping of data blocks at different servers to create coding opportunities in the shuffling phase to exchange (key,value) pairs among servers much more efficiently. We demonstrate that Coded MapReduce can cut down the total interserver communication load by a multiplicative factor that grows linearly with the number of servers in the system and it achieves the minimum communication load within a constant multiplicative factor. We also analyze the tradeoff between the "computation load" and the "communication load" of Coded MapReduce. 
Article: Blind Index Coding
[Show abstract] [Hide abstract]
ABSTRACT: We introduce the "blind index coding" (BIC) problem, which generalizes the classic index coding problem by allowing the sender to have some uncertainty about the side information that is available at each receiver. This problem naturally arises in wireless networks, in which users obtain their side information through wireless channels with errors that may be unknown to the sender. For the proposed BIC problem, we develop a new general outer bound by first proving it for the 3user case, and then generalizing its construction to K users. The proof of the outer bound relies on development of a key lemma that uses a strong data processing inequality to account for the sender's uncertainty. We also propose a hybrid coding scheme that XORs random combinations of bits from a subset of messages with uncoded bits of other messages in order to blindly exploit side information, and illustrate its gain. We finally generalize the BIC problem to consider a wireless channel from sender to users, and observe that even in the case of two users the solution becomes nontrivial and a natural generalization of the hybrid coding scheme relying on XORing repetitions of uncoded bits strictly outperforms conventional schemes.  [Show abstract] [Hide abstract]
ABSTRACT: In this paper, we study the communication problem from rovers on Mars' surface to Marsorbiting satellites. We first justify that, to a good extent, the rovertoorbiter communication problem can be modelled as communication over a $2 \times 2$ Xchannel with the network topology varying over time. For such a fading Xchannel where transmitters are only aware of the timevarying topology but not the timevarying channel state (i.e., no CSIT), we establish the sum degreesoffreedom (DoF) by 1) proposing coding strategies that code across topologies, and 2) developing novel upper bounds on the sumDoF. Furthermore we demonstrate that the proposed scheme approximately achieves the ergodic sum capacity of the network. Using the proposed coding scheme, we numerically evaluate the ergodic rate gain over a timedivisionmultipleaccess (TDMA) scheme for Rayleigh and Rice fading channels. We also numerically demonstrate that with practical orbital parameters, a 9.6% DoF gain, as well as more than 11.6% throughput gain can be achieved for a rovertoorbiter communication network. 
Article: Asymptotic Justification of Bandlimited Interpolation of Graph signals for SemiSupervised Learning
[Show abstract] [Hide abstract]
ABSTRACT: Graphbased methods play an important role in unsupervised and semisupervised learning tasks by taking into account the underlying geometry of the data set. In this paper, we consider a statistical setting for semisupervised learning and provide a formal justification of the recently introduced framework of bandlimited interpolation of graph signals. Our analysis leads to the interpretation that, given enough labeled data, this method is very closely related to a constrained low density separation problem as the number of data points tends to infinity. We demonstrate the practical utility of our results through simple experiments. 
Article: Topological Interference Management with just Retransmission: What are the "Best" Topologies?
[Show abstract] [Hide abstract]
ABSTRACT: We study the problem of interference management in fast fading wireless networks, in which the transmitters are only aware of network topology. We consider a class of retransmissionbased schemes, where transmitters in the network are only allowed to resend their symbols in order to assist with the neutralization of interference at the receivers. We introduce a necessary and sufficient condition on the network topology, under which half symmetric degreesoffreedom (DoF) is achievable through the considered retransmissionbased schemes. This corresponds to the "best" topologies since half symmetric DoF is the highest possible value for the symmetric DoF in the presence of interference. We show that when the condition is satisfied, there always exists a set of carefully chosen transmitters in the network, such that by retransmission of their symbols at an appropriate time slot, we can neutralize all the interfering signals at the receivers. Quite surprisingly, we also show that for any given network topology, if we cannot achieve half symmetric DoF by retransmissionbased schemes, then there does not exist any linear scheme that can do so. We also consider a practical network scenario that models cell edge users in a heterogeneous network, and show that the characterized condition on the network topology occurs frequently. Furthermore, we numerically evaluate the achievable rates of the DoFoptimal retransmissionbased scheme in such network scenario, and show that its throughput gain is not restricted to the asymptotic DoF analysis.  [Show abstract] [Hide abstract]
ABSTRACT: In this work we study the impact of limited transmitter cooperation on interference management in twouser interference channel. In particular, we consider the twouser interference channel with delayed channel state information at the transmitters (delayed CSIT). We first present a model to capture and quantify the amount of cooperation between the transmitters. In this model we denote the fraction of shared messages that are intended for Rx1, Rx2 by ρ1,ρ2, respectively, and then, characterize the degrees of freedom (DoF) region as a function of ρ1, ρ2. As a result, the twouser interference channel and twouser multipleinput singleoutput (MISO) broadcast channel become special cases of no cooperation (ρ1 = ρ2 = 0) and full cooperation (ρ1 = ρ2 = 1) in our framework. Moreover, our result indicates that the maximum benefit of cooperation from the DoF perspective is achieved by sharing only half of the messages between the transmitters.  [Show abstract] [Hide abstract]
ABSTRACT: We consider the problem of determining rank loss conditions for a concatenation of fullrank matrices, such that each row of the composing matrices is scaled by a random coefficient. This problem has applications in wireless interference management and recommendation systems. We determine necessary and sufficient conditions for the design of each matrix, such that the random ensemble will almost surely lose rank by a certain amount. The result is proved by converting the problem to determining rank loss conditions for the union of some specific matroids, and then using tools from matroid and graph theories to derive the necessary and sufficient conditions. As an application, we discuss how this result can be applied to the problem of topological interference management, and characterize the linear symmetric degrees of freedom for a class of network topologies.  [Show abstract] [Hide abstract]
ABSTRACT: This monograph illustrates a novel approach, which is based on changing the focus to seek approximate solutions accompanied by universal guarantees on the gap to optimality, in order to enable progress on several key open problems in network information theory. We seek universal guarantees that are independent of problem parameters, but perhaps dependent on the problem structure. At the heart of this approach is the development of simple, deterministic models that capture the main features of information sources and communication channels, and are utilized to approximate more complex models. The program advocated in this monograph is to use first seek solutions for the simplified deterministic model and use the insights and the solution of the simplified model to connect it to the original problem. The goal of this deterministicapproximation approach is to obtain universal approximate characterizations of the original channel capacity region and source coding rate regions. The translation of the insights from the deterministic framework to the original problem might need nontrivial steps either in the coding scheme or in the outer bounds. The applications of this deterministicapproximation approach are demonstrated in four central problems, namely unicast/multicast relay networks, interference channels, multiple descriptions source coding, and joint sourcechannel coding over networks. For each of these problems, it is illustrated how the proposed approach can be utilized to approximate the solution and draw engineering insights. Throughout the monograph, many extensions and future directions are addressed, and several open problems are presented in each chapter. The monograph is concluded by illustrating other deterministic models that can be utilized to obtain tighter approximation results, and discussing some recent developments on utilization of deterministic models in multiflow multihop wireless networks. © 2015 A. S. Avestimehr, S. N. Diggavi, C. Tian and D. N. C. Tse. 
Article: Sampling Large Data on Graphs
[Show abstract] [Hide abstract]
ABSTRACT: We consider the problem of sampling from data defined on the nodes of a weighted graph, where the edge weights capture the data correlation structure. As shown recently, using spectral graph theory one can define a cutoff frequency for the bandlimited graph signals that can be reconstructed from a given set of samples (i.e., graph nodes). In this work, we show how this cutoff frequency can be computed exactly. Using this characterization, we provide efficient algorithms for finding the subset of nodes of a given size with the largest cutoff frequency and for finding the smallest subset of nodes with a given cutoff frequency. In addition, we study the performance of random uniform sampling when compared to the centralized optimal sampling provided by the proposed algorithms.  [Show abstract] [Hide abstract]
ABSTRACT: Computing optimal halfduplex schedules in Gaussian relay networks is a challenging problem due to the lack of an exact capacity characterization and the large number of transmitreceive configurations that must be considered. We approach the problem using a constantgap capacity approximation based on the cutset bound with independent encoding at the nodes. We formulate an optimization problem to obtain the cutset optimal halfduplex schedule and find that it is hard to solve in general. This is because it involves an exponential number of variables, since the number of ways to assign each node to either transmitter or receiver mode is exponential in the number of nodes. We present a general technique that takes advantage of specific structures in the topology of a given network and allows us to reduce the complexity of this problem. In certain classes of network topologies, our approach yields polynomial time algorithms for finding halfduplex schedules that achieve capacity within a constant gap. We use simulations to show running time improvements over alternative methods and compare the performance of various halfduplex scheduling approaches in different SNR regimes. 
Conference Paper: Alignandforward relaying for twohop erasure broadcast channels
[Show abstract] [Hide abstract]
ABSTRACT: We consider the problem of broadcast over wireless erasure networks. To understand the challenges and opportunities of these setups, we study a twohop erasure broadcast channel consisting of a single source, two relays, and two destinations desiring independent messages. In our network, no transmitter has channel state knowledge of erasures on outgoing links (i.e., no CSIT): The source has no knowledge of any channel state, each relay only has knowledge of the channel states of its incoming link, and destinations are provided with full channel knowledge. We propose a scheme, referred to as AlignandForward, that exploits the (unknown) common subspace of received signals at the relays, which results from the sourcetorelay broadcast, in order to minimize the dimension of the interference subspace at each destination. We show that AlignandForward outperforms available alternative schemes in terms of sumrate. We also present new outerbounds and demonstrate the optimality of AlignandForward in certain regimes.  [Show abstract] [Hide abstract]
ABSTRACT: We consider the problem of the twouser multipleinput singleoutput complex Gaussian Broadcast Channel where the transmitter has access to delayed knowledge of the channel state information. We characterize the capacity region of this channel to within a constant number of bits for all values of the transmit power. The proposed signaling strategy utilizes the delayed knowledge of the channel state information and the previously transmitted signals, in order to create a signal of common interest for both receivers. This signal is the quantized version of the summation of the previously transmitted signals. To guarantee the independence of quantization noise and signal, we extend the framework of lattice quantizers with dither, together with an interleaving step. For converse, we use the fact that the capacity region of this problem is upperbounded by the capacity region of a physically degraded broadcast channel with no channel state information where one receiver has two antennas. We then derive an outerbound on the capacity region of this degraded broadcast channel which in turn provides an outerbound on the capacity region of the twouser multipleinput singleoutput complex Gaussian broadcast channel with delayed knowledge of the channel state information. By careful examination, we show that the achievable rate region and the outerbound are within 1.81 bits/sec/Hz per user.  [Show abstract] [Hide abstract]
ABSTRACT: We study the degrees of freedom (DoF) of the multipleinput multipleoutput Xchannel (MIMO XC) with delayed channel state information at the transmitters (delayed CSIT), assuming linear coding strategies at the transmitters. We present two results: 1) the linear sum DoF for MIMO XC with general antenna configurations, and 2) the linear DoF region for MIMO XC with symmetric antennas. The converse for each result is based on developing a novel rankratio inequality that characterizes the maximum ratio between the dimensions of received linear subspaces at the two multipleantenna receivers. The achievability of the linear sum DoF is based on a threephase strategy, in which during the first two phases only the transmitter with fewer antennas exploits delayed CSIT in order to minimize the dimension of its signal at the unintended receiver. During Phase 3, both transmitters use delayed CSIT to send linear combinations of past transmissions such that each receiver receives a superposition of desired message data and known interference, thus simultaneously serving both receivers. We also derive other linear DoF outer bounds for the MIMO XC that, in addition to the outer bounds from the sum DoF converse and the proposed transmission strategy, allow us to characterize the linear DoF region for symmetric antenna configurations.  [Show abstract] [Hide abstract]
ABSTRACT: We consider the Gaussian wiretap channel where a transmitter wishes to communicate a secure message to a legitimate receiver in the presence of eavesdroppers, without the eavesdroppers being able to decode the secure message. We focus on a setting that the transmitter is blind with respect to the state of channels to eavesdroppers, and only has access to delayed channel state information (CSI) of the legitimate receiver, which is referred to as "blind wiretap channel with delayed CSIT". We then consider two scenarios: (i) the case where the secure communication is aided via a distributed jammer, (ii) the case where all nodes in the network are equipped with multiple antennas, referred to as blind MIMO wiretap channel with delayed CSIT. We completely characterize the secure Degrees of Freedom (SDoF) in both scenarios, when assuming linear coding strategies at the transmitter(s).  [Show abstract] [Hide abstract]
ABSTRACT: We characterize the capacity region of the twouser Binary Fading Interference Channel where the transmitters have no knowledge of the channel state information. We show that the entire capacity region is achieved by applying pointtopoint erasure codes with appropriate rates at each transmitter, and using either treatinterferenceaserasure or interferencedecoding at each receiver, based on the channel parameters. The result is obtained by developing a novel outerbound that has three main steps. We first create a contracted channel that has fewer states compared to the original channel, in order to make the analysis tractable. Using a Correlation Lemma, we then show that an outerbound on the capacity region of the contracted channel also serves as an outerbound for the original channel. Finally, using a Conditional Entropy Leakage Lemma, we derive our outerbound on the capacity region of the contracted channel, and show that it coincides with the achievable region by either treatinterferenceaserasure or interferencedecoding at each receiver. We also show that having access to delayed local knowledge of the channel state information, does not enlarge the capacity region.  [Show abstract] [Hide abstract]
ABSTRACT: We consider the problem of the twouser multipleinput singleoutput complex Gaussian Broadcast Channel where the transmitter has access to delayed knowledge of the channel state information. We characterize the capacity region of this channel to within a constant number of bits for all values of the transmit power. The proposed signaling strategy utilizes the delayed knowledge of the channel state information and the previously transmitted signals, in order to create a signal of common interest for both receivers. This signal is the quantized version of the summation of the previously transmitted signals. To guarantee the independence of quantization noise and signal, we extend the framework of lattice quantizers with dither, together with an interleaving step. For converse, we use the fact that the capacity region of this problem is upperbounded by the capacity region of a physically degraded broadcast channel with no channel state information where one receiver has two antennas. We then derive an outerbound on the capacity region of this degraded broadcast channel which in turn provides an outerbound on the capacity region of the twouser multipleinput singleoutput complex Gaussian broadcast channel with delayed knowledge of the channel state information. By careful examination, we show that the achievable rate region and the outerbound are within 1.81 bits/sec/Hz per user.  [Show abstract] [Hide abstract]
ABSTRACT: We study the degrees of freedom (DoF) of the multipleinput multipleoutput Xchannel (MIMO XC) with delayed channel state information at the transmitters (delayed CSIT), assuming linear coding strategies at the transmitters. We present two results: 1) the linear sum DoF for MIMO XC with general antenna configurations, and 2) the linear DoF region for MIMO XC with symmetric antennas. The converse for each result is based on developing a novel rankratio inequality that characterizes the maximum ratio between the dimensions of received linear subspaces at the two multipleantenna receivers. The achievability of the linear sum DoF is based on a threephase strategy, in which during the first two phases only the transmitter with fewer antennas exploits delayed CSIT in order to minimize the dimension of its signal at the unintended receiver. During Phase 3, both transmitters use delayed CSIT to send linear combinations of past transmissions such that each receiver receives a superposition of desired message data and known interference, thus simultaneously serving both receivers. We also derive other linear DoF outer bounds for the MIMO XC that, in addition to the outer bounds from the sum DoF converse and the proposed transmission strategy, allow us to characterize the linear DoF region for symmetric antenna configurations.  [Show abstract] [Hide abstract]
ABSTRACT: We present a new outer bound for the sum capacity of general multiunicast deterministic networks. Intuitively, this bound can be understood as applying the cutset bound to concatenated copies of the original network with a special restriction on the allowed transmit signal distributions. We first study applications to finitefield networks, where we obtain a general outerbound expression in terms of ranks of the transfer matrices. We then show that, even though our outer bound is for deterministic networks, a recent result relating the capacity of AWGN KxKxK networks and the capacity of a deterministic counterpart allows us to establish an outer bound to the DoF of KxKxK wireless networks with general connectivity. This bound is tight in the case of the "adjacentcell interference" topology, and yields graphtheoretic necessary and sufficient conditions for K DoF to be achievable in general topologies. 
Conference Paper: ITLinQ: A new approach for spectrum sharing
[Show abstract] [Hide abstract]
ABSTRACT: We consider the problem of spectrum sharing in wireless communication networks composed of multiple sourcedestination pairs. We define a novel concept of informationtheoretic independent sets (in short, ITIS) which indicates the sets of sourcedestination pairs in the network inside each of which, treating interference as noise is informationtheoretically optimal (to within a constant gap). Then, we propose a new spectrum sharing scheme called informationtheoretic link scheduling (in short, ITLinQ), which at each time schedules those users that form an ITIS. We first provide a performance guarantee of the ITLinQ scheme through characterizing the fraction of the capacity region that it can achieve in a network with source and destination nodes spread randomly in a fixed area. Moreover, we will show how to implement the ITLinQ scheme in a distributed fashion by using an initial 2phase signaling mechanism which provides the required channel state information at all the nodes. Finally, we compare the performance of the distributed ITLinQ scheme with similar stateoftheart spectrum sharing mechanisms, such as FlashLinQ, through numerical analysis and show that it can exhibit a sumrate gain of more than a 100%, while keeping the complexity at the same level.  [Show abstract] [Hide abstract]
ABSTRACT: The layered interference network is investigated with delayed channel state information (CSI) at all nodes. It is demonstrated how multihopping can be utilized to increase the achievable degrees of freedom (DoF). In particular, a multiphase transmission scheme is proposed for the $K$user $2K$hop interference network to systematically exploit the layered structure of the network and delayed CSI to achieve DoF values that scale with $K$. This result provides the first example of a network with distributed transmitters and delayed CSI whose DoF scales with the number of users.
Publication Stats
2k  Citations  
65.39  Total Impact Points  
Top Journals
Institutions

20132015

University of Southern California
 Department of Electrical Engineering
Los Ángeles, California, United States


20122014

University of California, Los Angeles
 Department of Electrical Engineering
Los Ángeles, California, United States


20092014

Cornell University
 Department of Electrical and Computer Engineering
Итак, New York, United States 
California Institute of Technology
 Department of Electrical Engineering
Pasadena, California, United States


2007

University of California, Berkeley
 Department of Electrical Engineering and Computer Sciences
Berkeley, California, United States
