Optimal signaling for MIMO interference networks

Abstract

In this thesis, we discuss the issue of overall mutual information maximization in MIMO interference networks. The interaction of links is modeled as a multi-link game in which each link maximizes its own objective function. The properties and performance of Nash equilibrium of the game are analyzed. We show that by imposing a limit on the number of independent data streams for each link, the performance of equilibrium can be substantially improved. A decentralized negotiation mechanism is proposed as an alternative to brute-force type of stream control. The problem of optimal power allocation is also considered. An extension of the negotiation mechanism is presented to incorporate the choice of power levels into the algorithm. Both the basic and extended negotiation processes converge to the optimal profile with arbitrarily high probability.