Date: 10:00am, February 20, 2017
Location: Costa Commons (CEPSR 750)
Speaker: Dr. Martina Cardona Research Fellow, UCLA
Abstract: Information theory designs near-optimal relaying schemes assuming no complexity constraints. 5G systems are envisaged to deploy cooperative relaying at a large scale, using low-complexity schemes. In this talk, we discuss some approaches to achieve near-optimal rate performance over complexity-constrained relay networks. We consider a wireless network where the communication from a base station to a mobile user is assisted by N half-duplex relays. For such a network, a constant-gap capacity approximation involves an optimization over the 2^N possible listen/transmit relays configuration states. We leverage intrinsic structural properties of such networks and we show a surprising result: at most N + 1 states, out of the 2^N possible ones, suffice for a constant-gap capacity approximation for a class of half-duplex relay networks, which includes the practically relevant Gaussian noise case. We next design a polynomial-time algorithm that allows to compute the at most N + 1 active states sufficient for constant-gap capacity approximation for Gaussian half-duplex relay networks where the N relays are arranged in a line. The algorithm uses similarities between network states in half-duplex and edge coloring in a graph and achieves a rate that provides a closed-form expression for the approximate capacity.