Distributed power allocation for network MIMO with a Bayesian game-theoretic approach

Network multiple input multiple output (MIMO), in which each user is served by all the transmitters within its range of influence, is a promising technology for higher spectral efficiency. However, such performance gain comes at the expense of significant signaling overhead due to the sharing of channel state information (CSI) and transmission data. This paper proposes a truly distributed power allocation scheme for network MIMO, where “truly” means no CSI related information exchange in any form is required among transmitters. Such property is quite desirable for network MIMO but does not hold for most of the existing schemes in the literature. With the proposed scheme, each transmitter determines the power allocation with local CSI only. The cooperation is achieved by utilizing a Bayesian game-theoretic model, where the cooperating transmitters are modeled as players which share a common payoff function. A symmetric Bayesian Nash equilibrium is selected as the operating point, which can be determined individually by each transmitter performing an identical optimization problem separately. Simulation result shows that the distributed algorithm gives almost the same average sum rate with the centralized solution which requires global CSI.