A joint polling and contention based opportunistic scheduling algorithm with fairness constraint

Abstract

Opportunistic schedulers improve the network capacity by allocating the network resources to users with good link qualities. A drawback of these schedulers is that they deprive the remaining users from receiving a fair share of the network resources, or violate their quality-of-service (QoS) requirements. Fair schedulers on the other hand impose a degree of fairness by allocating the network resources to the users in a weighted manner. However, majority of the existing fair algorithms do not impose the same degree of fairness in heterogenous networks. Furthermore, majority of these algorithms require full feedback from all users in order to make a scheduling decision. In this paper, we propose a scheduling algorithm that reduces the feedback overhead, and thus the scheduling delay, while maintaining a degree of fairness in heterogeneous wireless networks. To impose a degree of fairness, the scheduler prioritizes a set of starved users and allocates the network resource to the user (starved) with the best link quality. To reduce the feedback overhead, the scheduler requests users to feedback their link qualities in a joint polling and contention manner. Numerical results show that the proposed algorithm attains the highest degree of fairness while reducing the overall feedback load when compared to other well known fair and greedy algorithms. Furthermore, we show that a trade-off exits between the attainable degree of fairness and the scheduling delay, and hence, the scheduler should strike a balance between fairness and delay depending on the network priority.