A playout-buffer-sensitive time slot scheduling for integration of real-time and elastic traffic in wireless networks

Some channel-sensitive scheduling algorithms for mobile packet systems are designed to support real-time traffic. They schedule time slots depending on the channel state and queue length in the base station. Although queue length is correlated with the QoS of real-time applications, the essential performance measure of real-time applications is the playout buffer starvation probability. This study proposes a new scheduling algorithm, the playout-buffer-sensitive (PB) rule, which can efficiently support a mixture of real-time and elastic traffic. The PB rule achieves sufficient QoS for the real-time class as well as allocating higher throughput for the best effort class than the existing algorithms. We numerically show that the efficiency of the PB rule is much closer to the theoretical upper bound than the other algorithms.