Optimal One-Shot Scheduling for MIMO Networks

A MIMO network is a wireless network made up of individual MIMO links.The problem we consider is to maximize throughput in a multihopMIMO network with interference suppression. Our problem formulationaccounts for variable rates on the MIMO links, which depend on thechannel conditions of the link, and the manner in which thediversity-multiplexing trade-off is handled. We present an ILPformulation of the MIMO one-shot scheduling problem with variable rates, whichis the first {em exact} formulation of a MIMO network optimizationproblem that accounts for full interference suppression capabilities of MIMO links.We use CPLEX to evaluate the optimal solution based on theILP formulation for wireless networks with up to 32 concurrently transmittinglinks. We also modify a heuristic algorithm from a related MIMO schedulingproblem to work in our problem setting. Results show that the heuristic canscale to networks with 80 or more concurrent links, but is 10-20% from optimalin terms of throughput. We show that the heuristic scheduler is not able tofully exploit the diversity-multiplexing-interference suppression tradeoff,which is inherent in the problem. This shows that there is substantial roomfor developing improved scheduling algorithms for MIMO networks andprovides some insight into promising directions to explore.