An Efficient Scheduler for Real and Non-Real Time Services Maximizing Satisfied Users in Wireless Networks

Traditional scheduling algorithms are focused in maximizing system throughput considering a grade of fairness. However, maximizing the system throughput does not necessarily result in maximizing the number of satisfied users (users with a packet delay below a threshold). Also, the throughput maximization could cause a low grade of fairness. In this paper, we propose a scheduling algorithm that maximizes the number of satisfied users for a mixture of real and non-real time services. The main idea of this algorithm is to find the optimum transmission probability for each user, based on their statistical channel variations and their required qualities of service (QoS). The algorithm dynamically adapts its parameters in order to obtain the optimum transmission probability for each user according to their channel variations and provides a simple congestion control mechanism to produce system stability. We consider a channel gain that reflects the effects of shadowing and multipath fading. The results show that with the proposed algorithm, the system capacity for real time services is increased 20% and for non-real time services is increased 40% over the M-LWDF scheduler.