Consensus-Based Distributed MIMO Decoding Using Semidefinite Relaxation

Abstract

A distributed algorithm is developed for decoding a message broadcasted from a wireless multi-antenna access point to a network of single-antenna (sensor) nodes. Sensors exchange local messages to reach consensus on the transmitted message. Different from recent distributed detectors where the amount of information exchanges increases exponentially with the alphabet size (i.e., the number of hypotheses tested), the novel consensus-based approach introduced here relies on semi-definite relaxation techniques and can afford inter-sensor exchanges of polynomial order. The resultant near-optimum convexified problem is solved in a distributed fashion using the alternating direction method of multipliers. No constraint is imposed on the network topology so long as it remains fully connected. Preliminary simulations demonstrate the merits of the novel distributed detection algorithm.