Energy Efficient Quality of Service Traffic Scheduler for MIMO Downlink SVD Channels

In this paper we focus on minimizing the long-term average power consumption of a single transmitter providing Quality of Service (QoS) enabled traffic to a single receiver. Both the transmitting and receiving stations are equipped with multiple antennas. First, we present a general {Kx M} system model where K is the number of independently buffered QoS streams and M is the number of parallel channels available through MIMO SVD eigenmode transmission. Through application of the constrained Markov decision process (MDP) framework combined with a novel MAC layer rate assignment scheme, a randomized per-buffer scheduling policy is obtained. The designed policy exploits queue state information to schedule traffic while meeting throughput, delay and loss constraints. Packets scheduled for transmission during each frame are mapped across the set eigenmode channels subject to available channel resources and the set of channel eigenvalues. Simulation results are provided for several scenarios. System drawbacks, limitations and extensions are also discussed.