Delay-Optimal Buffer-Aware Probabilistic Scheduling with Adaptive Transmission

TL;DR

This paper develops a buffer-aware probabilistic scheduling framework for cross-layer systems with random data arrivals, optimizing delay-power tradeoff using Markov reward processes and threshold-based strategies.

Contribution

It introduces a novel probabilistic scheduling approach, characterizes the delay-power tradeoff as piecewise-linear, and provides an efficient algorithm for optimal threshold-based strategies.

Findings

Optimal delay-power tradeoff is piecewise-linear.

Threshold-based strategies achieve the optimal tradeoff.

Algorithm results match linear programming optimization.

Abstract

Cross-layer scheduling is a promising way to improve Quality of Service (QoS) given a power constraint. In this paper, we investigate the system with random data arrival and adaptive transmission. Probabilistic scheduling strategies aware of the buffer state are applied to generalize conventional deterministic scheduling. Based on this, the average delay and power consumption are analysed by Markov reward process. The optimal delay-power tradeoff curve is the Pareto frontier of the feasible delay-power region. It is proved that the optimal delay-power tradeoff is piecewise-linear, whose vertices are obtained by deterministic strategies. Moreover, the corresponding strategies of the optimal tradeoff curve are threshold-based, hence can be obtained by a proposed effective algorithm. On the other hand, we formulate a linear programming to minimize the average delay given a fixed power constraint. By varying the power constraint, the optimal delay-power tradeoff curve can also be obtained. It is demonstrated that the algorithm result and the optimization result match each other, and are further validated by Monte-Carlo simulation.