Signal Processing Techniques for Coordinated Multipoint (CoMP)

Abstract

Coordinated multipoint (CoMP) is a multiple-input-multiple-output (MIMO) technique used in the fourth generation (4G) cellular networks in order to enhance edge user services. Coordination among different base stations (network nodes) is done through exchange of information among base stations over the backhaul of the network. In this dissertation, we model CoMP as an adaptive diffusion network, with the base stations as the network nodes. In adaptive diffusion networks, one of the main challenges is the large volume of data exchange among nodes needed to arrive at a collective decision. We first introduce an overview of the adaptation and learning process over a network through non-cooperative, centralized, and distributed manners. A comparison of the steady-state mean square deviation (MSD) performance for all strategies is provided. Two new algorithms for adaptive diffusion networks are proposed which offer a tradeoff between the mean-square error performance of the system and the volume of data exchanged among network nodes while preserving the network convergence rate. A study of the mean-square-stability of the network under the proposed algorithms is provided. Also, a study of the mean-error dynamic behavior of the network is carried out. A closed-form expression for the overall network steady-state mean-square-error is derived and verified against simulated data. The proposed algorithms is finally applied to a cellular network location estimation problem, and delivers good performance even under 75% reduction in data exchange volume.