Abstracts

Monday 11 May 2009

Plenary Talk: David Tse, "The Two-User Gaussian Interference Channel"

Interference is a central phenomenon in wireless communications. A basic information theoretic model to study optimal interference management is the two-user Gaussian interference channel. In this talk, we first summarize recent progress in understanding the capacity of this channel. Using these results as a foundation, we then discuss how feedback and cooperation can be used to increase interference-limited capacity.

Morning Session: The Interference Channel

Vincent Poor, "Interference and Secrecy"

The physical properties of the wireless medium - and particularly its open nature - make wireless communications especially susceptible to eavesdropping. As a counterpoint to this problem, the physical layer can also potentially be exploited to provide security in wireless networks, a possibility that has attracted considerable attention in recent years. One notable conclusion of recent work in this area is that fading can be helpful in providing secrecy in wireless networks, just as it is in providing additional capacity or reliability via MIMO systems. [See, e.g., Liang, Poor & Shamai, "Secure Communications Over Fading Channels," IEEE Transactions on Information Theory 54 (6) June 2008.] This talk will address the use of the other principal physical impairment of wireless - namely interference - as an aid in providing secret communications over wireless links. In particular, after a brief review of recent work on physical layer security in wireless networks, several situations in which interference can be beneficial to network security will be described. These situations include the multi-user Gaussian interference channel, in which interference alignment and secrecy pre-coding can be used to provide non-zero secure degrees of freedom to each user, and the wire-tap channel with a helping interferer, in which an interferer can increase the secrecy capacity of the channel even without knowledge of confidential messages. It is interesting to note that interference is also a consequence of the open nature of the wireless medium; however, unlike eavesdropping, which is made possible by the broadcast aspect of the open medium, interference stems from the superposition property of the medium. Thus, the use of interference to provide secrecy exploits one consequence of an open medium to offset the negative effects of another consequence of openness.

Venu Veeravalli, "Sum Capacity of Gaussian MIMO Interference Channels in the Low Interference Regime"

Treating interference as noise has recently been shown to be sum capacity achieving for the two-user single-input single-output (SISO) Gaussian interference channel in a low interference regime. In this talk, we characterize the low interference regime for the two-user multiple-input multiple-output (MIMO) Gaussian interference channel. For the special cases of the multiple-input single-output (MISO) and single-input multiple-output (SIMO) Gaussian interference channels, we provide explicit conditions on the channel parameters that define low interference regimes for these channels.

Pramod Viswanath, "Cooperative Interference Management: An Information Theoretic Study"

In a traditional interference channel sources only transmit and destinations only listen. In practice, radios both transmit and receive. By allowing sources to listen and destinations to transmit as well, we engender potential cooperation between the radios. We characterize the sum capacity of this channel (with either source or destination cooperation) to a constant number of bits. Novel communications schemes that handle cooperative interference management arise. An intriguing reciprocity result is also uncovered.

Sriram Vishwanath, "Lattice Alignment through Symbol Alignment for Interference Networks"

This paper uses alignment over finite fields (so called "symbol alignment") to enable lattice alignment in interference networks. The main result of this paper is as follows: That an information theoretic capacity problem can be reduced to an algebraic/linear systems problem using nested lattice coding. We show that, for a large class of interference channels, lattices can provide an achievable rate region that is larger than that achieved by Gaussian coding schemes.

Amir Khandani, "Randomized Resource Allocation in Decentralized Networks"

This talk (joint work with K. Moshksar and A. Bayesteh) addresses the problem of resource (time/bandwidth) allocation in decentralized networks to manage the impact of the multi-user interference. In addition to the additive Gaussian noise and fading, two other random factors that can heavily affect the strategy and performance of resource allocation are accounted for: i) randomness in the number of active users, ii) randomness in the amount of overlap among coding blocks (asynchronous users). Subject to these uncertainties and considering interference as additive noise, it is shown that a randomized resource allocation can result in significant improvements in performance measures such as multiplexing gain, outage probability and fairness as compared to the conventional methods of interference avoidance. This conclusion is valid in many cases of interest even if the performance of interference avoidance methods is enhanced by sensing resource occupancy followed by an opportunistic allocation.

Afternoon Session: Cooperative Networks & Relays

Gerhard Kramer, "Communicating Probability Distributions and Relaying"

Moe Win, "Cooperative Localization-Aware Networks"

Salman Avestimeher, "Approaching the Capacity of the Multi-Pair Bidirectional Relay Network via a Divide and Conquer Strategy"

Ashutosh Sabharwal, "Mismatched Channel Information in Interference Networks"

Babak Hassibi, "The Entropy Power Inequality and the Gaussian Interference Channel"

Poster Session

Peter Rost, Gerhard Fettweis, J. Nicholas Laneman, "Broadcast and Interference in Relay-Assisted Next-Generation Cellular Systems"

We discuss a cellular network, where cooperative base-station transmission and relaying are combined. Inter-cell cooperation is able to mitigate interference at cell-edges while relays are able to improve the frequency reuse. We evaluate the system using its throughput and cost-benefit tradeoff.

Stefan Krone and Gerhard Fettweis, "Aspects of Optimal Quantization in Communications"

Digital communications systems do typically require analog-to-digital conversion at the receiver. The data rate that can be achieved at maximum is not only affected by the sampling frequency and quantization resolution of the analog-to-digital conversion but also by the actual quantization characteristic. This work discusses the design of optimal quantization characteristics allowing for maximum data rate from an information-theoretic perspective.

Umar H. Rizvi, Ferkan Yilmaz, Mohamed-Slim Alouini, Gerard J. M. Janssen and Jos H. Weber, "Performance of RF Level Diversity Combining with Quantized Channel Estimates over Rayleigh Fading"

Ferkan Yilmaz and Mohamed-Slim Alouini, "Product of Shifted Gamma Variates and Outage Capacity of Multicarrier Systems"

The probability density function and the cumulative distribution function of the product of shifted Gamma variates are obtained in terms of the generalized Fox's H function. Using these new results, the exact outage capacity of multi carrier transmission through a Nakagam-m fading channel is presented. Moreover, it is shown that analytical and simulation results are in perfect agreement.

Tareq Y. Al-Naffouri and Babak Hassibi, "On the Distribution of Indefinite Quadratic Forms in Gaussian Random Variables"

In this poster, we propose a transparent approach to evaluating the CDF of indefinite quadratic forms in Gaussian random variables and ratios of such forms. This quantity appears in the analysis of different receivers in communication systems and in normalized adaptive filters in signal processing. Instead of trying to find the pdf of this quantity as is the case in many papers in literature, we focus on finding the CDF. The basic trick that we implement is to replace inequalities that appear in the CDF calculations with the unit step function and replacing the latter with its Fourier transform. This produces a multi-dimensional integral that can be evaluated using complex integration. We show how approach extends to nonzero mean Gaussian real/complex vectors and to the joint distribution of indefinite quadratic forms.

Amir Laufer and Yeheskel Bar-Ness, "Full Rate Space Time Codes for Large Number of Transmitting Antennas with Linear Complexity and Decoding and High Performance"

Tuesday 12 May 2009

Plenary Talk: Bertrand Hochwald, "The Role of MIMO in 4G Mobile Wireless Systems and Beyond"

Fourth-generation mobile wireless multiuser systems such as WiMax are now being deployed with multiple-antennas at both the transmitter and receiver. These multiple-input-multiple-output (MIMO) deployments are modest in the number of antennas, typically two at the basestation and two at the mobile. We have had some time now to evaluate the effectiveness of MIMO in live mobile networks, and what have we learned? We can say with some confidence that MIMO is here to stay. The promise of high spectral efficiencies often dominates MIMO discussions, but there are other perhaps non-obvious reasons why MIMO is important for future wireless systems. I will discuss what we have learned so far using field data from our deployments, and speculate on the role of MIMO in systems to come.

Morning Session: Multiuser MIMO

Nihar Jindal, Marious Kountouris, Jeff Andrews,"The Transmission Capacity of Multiuser MIMO in Ad Hoc Networks"

Multiuser MIMO capacity is well-understood for the uplink and downlink, but not for distributed, or ad hoc, networks. The primary difficulty in such cases is accurately but tractably modeling the interference, which we overcome using stochastic geometry. First, we show that with N transmit and N receive antennas per node -- such symmetry is typical in an ad hoc network since all nodes send and receive -- the transmission capacity grows superlinearly with N and is achieved with dirty paper coding. To our knowledge this is the first capacity result showing better than O(N) scaling in any multiuser MIMO setting. Second, we show that much of this gain is lost if linear precoding is employed en lieu of DPC. The loss is more rapid than in downlink or uplink scenarios because of the large dynamic range of interference in an ad hoc network (due to the near-far problem). Finally, we demonstrate that receive array processing is very important and can recover much of the loss caused by linear precoding. We prove that even with a single active transmit antenna per node the transmission capacity still scales linearly with N, and that this is achieved by optimally splitting the receive array resources between interference cancellation and diversity.

Chan-Byoung Chae, "Multiuser/Multi-cell MIMO Transmission with Coordinated Beamforming"

Multiuser MIMO (multiple-input multiple-output) uses multiple antennas to support transmission to and reception from multiple users in wireless cellular systems. While it is likely that mobile units in cellular systems will have multiple receive antennas, most prior work on the downlink assumed only one receive antenna at the mobile station. Under this assumption, linear transmit beamforming solutions are easy to compute since they are a simple function of the mobile stations' channel state information. The problem is more complicated when the mobile stations have multiple receive antennas, e.g. as expected in the next deployment of 3GPP LTE-Advanced and IEEE802.16m. Coordinated beamforming, where the transmit and receive beamforming vectors are designed jointly through an iterative algorithm, is one solution to this problem. Unfortunately, the iterative coordinated beamforming algorithms do not always converge and further require complete channel state information. In this talk, a closed form coordinated beamforming solution is proposed for two users. Extensions to three users and to multi-cell environments with clustered broadcast and full broadcast are also discussed. The proposed algorithms provide good sum-rate performance yet only require linear transmission and reception.

Raymond Knopp, "Practical Considerations for MU-MIMO in Centralized and Distributed Wireless Networks"

Howard Huang, Sivarama Venkatesan, Angel Lozano, Reinaldo Valenzuela, "Network MIMO for Indoor Wireless Applications"

Network MIMO is a family of techniques whereby each user in a wireless system is served through all the access points within its range of influence. By tightly coordinating the transmission and reception of signals at multiple access points, Network MIMO transcends the limits on spectral efficiency due to intercell interference. Taking prior information-theoretic analyses of Network MIMO to the next level, this paper quantifies the spec-tral efficiency gains obtainable under realistic propagation and operational conditions. Our study relies on detailed simulations and, for specificity, is conducted within the framework of the IEEE 802.16e Mobile WiMAX system. All the relevant physical-layer functionalities of Mobile WiMAX are accurately replicated. Furthermore, to facilitate the coordination be-tween access points, we postulate an indoor deployment organized around a Gigabit-ethernet backhaul. The results confirm that Network MIMO stands to provide a multiple-fold increase in spectral efficiency under such conditions.

Vincent Lau, "Delay-Sensitive Spatial Division Multiple Access (SDMA) Systems"

In this talk, we shall focus on the delay-optimal power allocation for buffered SDMA systems in wireless fading channels. This problem is challenging as it involves both the information theory (to capture the PHY dynamics) and the queueing theory (to capture the delay dynamics) and brute force approach does not lead to any viable solutions. We exploit the birth-death dynamics of the buffered SDMA systems and stochastic decomposition approach to derive the delay optimal power adaptation scheme in a SDMA system. Unlike the conventional CSI-only power control solution, the delay-optimal power control is a function of both the CSI and QSI and has the multi-level water-filling structure in which the QSI determines the water-level and the CSI determines the power allocation across the SDMA users. This new approach overcomes the complexity issue mentioned above and allow us to obtain closed-form performance expressions so as to obtain the following first order insights.

Afternoon Session: Compressed Sensing

Mark Davenport, "Compressive Radio Receivers"

Compressive sensing (CS) is an alternative to Shannon/Nyquist sampling for acquisition of sparse or compressible signals that can be well approximated by just K less than N elements from an N-dimensional basis. Instead of taking periodic samples, we measure inner products with M less than N random vectors and then recover the signal via a sparsity-seeking optimization or greedy algorithm. Using CS, robust signal recovery is possible from just M=O(K log(N/K)) measurements. The implications are promising for many applications and enable the design of new kinds of analog-to-digital converters, cameras and imaging systems, and sensor networks. In this talk, we will overview our work on compressive radio receivers that efficiently acquire sparse radio signals from ultra wide bandwidths. To enable these receivers, we have developed new theory for signal detection, filtering and interference cancelation, and beamforming in the compressive domain.

Bhaskar Rao, "Bayesian Methods for Sparse Signal Recovery "

In this talk, we discuss Bayesian methods for computing sparse solutions to a under determined linear system of equations. We discuss sparsity inducing priors and utilizing a super-Gaussian prior, the MAP formulation is employed to derive a regularized least squares cost function. The FOCUSS algorithm for minimizing the above mentioned cost function is presented and its properties discussed. Next we discuss a empirical Bayesian approach such as sparse Bayesian learning (SBL), which use a parameterized prior to encourage sparsity via evidence maximization. We discuss several properties of the SBL approach and contrast it with the MAP approach.

Waheed U. Bajwa, Akbar Sayeed, and Robert Nowak, "Compressed Channel Sensing: Theory and Implications"

Numerous experimental studies in the recent past have shown that physical wireless channels encountered in practice tend to exhibit sparse structures at high signal space dimension in the sense that majority of the channel degrees of freedom end up being either zero or nearly zero when operating at large bandwidths and packet durations and/or with large plurality of antennas. However, traditional training-based channel estimation methods relying on linear reconstruction schemes at the receiver seem incapable of exploiting the inherent low-dimensionality of such sparse channels, thereby leading to overutilization of the key communication resources of energy and bandwidth. Recently, a number of researchers have tried to address this problem and proposed training signals and reconstruction strategies that are tailored to the anticipated characteristics of sparse multipath channels. But much of the emphasis in these studies has been directed towards establishing the feasibility of the proposed sparse-channel estimation methods numerically rather than analytically. In contrast, by leveraging key ideas from the theory of compressed sensing, we have recently proposed new training-based estimation methods for various classes of sparse single- and multiple-antenna channels that are provably more effective than their traditional counterparts. A common theme underlying all our training-based methods is the use of sparsity-inducing, mixed-norm-optimization-based reconstruction methods at the receiver (such as, Dantzig selector, lasso) that have arisen out of recent advances in the theory of sparse signal representation, more commonly studied under the rubric of compressed sensing these days. In the spirit of compressed sensing, we term this particular approach to estimating sparse multipath channels as compressed channel sensing (CCS)---the analogy here being that CCS requires far fewer communication resources to estimate sparse channels than do the traditional training-based methods. The goal of this paper is to provide a unified summary of the key ideas underlying the theory of CCS and discuss the implications of explicitly accounting for sparsity in channel estimation methods.

Phil Schniter, "Sparse Reconstruction as Noncoherent Decoding"

In sparse reconstruction, a key component of compressive sensing, the goal is to estimate a sparse coefficient vector from noisy linear measurements of the vector. By formulating the sparse reconstruction problem as a joint detection/estimation problem, where the detection concerns the indices of non-zero coefficients and estimation concerns the values of those coefficients, we establish links to noncoherent decoding, where detection concerns the indices of codewords and estimation concerns the channel fading coefficients. Leveraging these links, we apply pairwise-error analysis tools from the noncoherent decoding literature to the sparse reconstruction problem in order to obtain error bounds, as a function of SNR and restricted-isometry constant, that are tight at high SNR.

Vahid Tarokh, "SPARLS: A Low Complexity Recursive L1-Regularized Least Squares Algorithm"

The RLS algorithm is one of the most applied algorithms in signal processing applications, in particular in wireless communications. In this presentation, we discuss SPARLS, a low complexity recursive L_1 regularized Least Square Algorithm for identification, detection, compensation or equalization of sparse signals, and channels of interest. The SPARLS algorithm provides significant gains in performance and is orders of magnitude less complex than RLS. It is also significantly more robust. This is a joint work with Behtash Babadi and Nicholas Kolouptsidis.

Wednesday 13 May 2009

Plenary Talk: Vahid Tarokh, "Near Theoretical Lower bound Sparse Representation and Compressive Sampling"

In this talk, we will consider the problem of sparse representation of signals and the associated problem of compressive sampling. By making a connection to, and using lessons learned from communications theory, we develop universal lower bounds on sparse representations of signals, and study the minimum number of samples required to be able to reconstruct sparse signals. We then produce sampling methods that perform within 2 dB from the theoretical lower bounds.

This is a joint work with Mehmet Akcakaya, Behtash Babadi and Jinsoo Park.

Morning Session: Cognitive Radio

Syed Jafar, "Degrees of Freedom of Cognitive Interference Networks"

Osvaldo Simeone, "Spectrum Leasing Via Cooperation"

Wenyi Zhang, "Efficient Mechanism of Receive Side Information Feedback and its Application in Cognitive Radio"

Qing Zhao, "Quickest Change Detection for Cognitive Radio Systems"