Efficient Performance Evaluation of Wireless Networks with Varying Channel Conditions

Abstract

This paper investigates the performance of opportunistic schedulers in wireless networks. A base station communicates over fading channels with multiple mobile nodes, each experiencing varying and not necessarily identical wireless channel conditions. An opportunistic scheduler optimises performance by accounting for both buffer size as well as channel conditions when allocating the transmitter energy among its users. The present study provides the necessary analytical tools to assess performance of opportunistic schedulers both fast and accurately, thereby allowing for fast evaluation and comparison of scheduling algorithms. The scheduler is modelled as a Markovian queueing system with multiple finite queues in a random environment. Already for a limited number of users and limited buffer capacities, the size of the state space of the Markov model makes the direct calculation of the steady-state probability vector nearly impossible. Therefore, we rely on Maclaurin series expansions so as to study the scheduler under light traffic conditions as well as in overload. The computational complexity for calculating the first $N$ terms in the series expansions is $O(NM^2S)$, where $M$ is the size of the state space of the exogenous channel process and $S$ is the size of the state space of the entire Markov chain.