Stochastic traffic engineering for real-time applications over wireless networks

Abstract

In this work, we focus on the Stochastic Traffic Engineering (STE) problem arising from the support of QoS-demanding real-time media-streaming applications over fading and congestion affected TCP-friendly/IP multiantenna wireless pipes. First, after recasting the tackled STE problem in the form of a suitable cross-layer nonlinear stochastic optimization problem, we develop a traffic analysis of the overall underlying multiple-input multiple-output (MIMO) wireless pipe that points out the relative effects of both fading-induced errors and congestion-induced packet losses on the goodput offered by the resulting end-to-end connection. Second, we develop an optimal cross-layer resource management policy that allows a joint scheduling of the media encoding rate (i.e., playin rate), transmit energy and delivery rate (i.e., playout rate) of each end-to-end connection active over the considered access network. Salient features of the presented joint scheduling policy are that: (i) it is self-adaptive; (ii) it is able to provide hard (i.e., deterministic) QoS guarantees, in terms of hard limited playout delay and playout rate-jitter; and (iii) it explicitly accounts for the performance interaction of the protocols implemented at all layers of the considered stack.