Proactive cooperative scheduling and buffer management for multimedia networks

Abstract

Is it possible, and to what extent feasible, for link schedulers and buffer managers in multimedia networks with propagation delays to relieve effectively congestion by means of feedback signaling? We show that a proactive cooperative control algorithm for distributed resource scheduling and buffer management obviates the problem of propagation delays. The algorithm is based on traffic predictions of correlated input traffic streams into network nodes. Nodes projecting to be under stress signal their neighboring upstream nodes for the latter to readjust their flow rates, whileguaranteeing quality of service as negotiated at call setup. Detailed implementations of the adaptive, feedback control mechanisms of distributed asynchronous algorithms that project future overload conditions are presented. The design allows the underlying single-node link scheduling and buffer management algorithm to be run in either a cooperative or a noncooperative mode of operation.