Streaming Stored Video Over Multi-Channel Wireless Systems: Scheduler Performance Evaluation

Abstract

We evaluate the queueing performance of various well-known downlink scheduling schemes in multi-channel wireless systems where input traffic is generated by streaming stored video. Multi-queue, multi-channel models can be used to analyze today's orthogonal frequency division multiple access (OFDMA) based fourth generation (4G) mobile wireless systems in which communication bandwidth is divided into many subchannels. Video streams exhibit long-range dependence, a characteristic which may have an impact on queue-length performance. Our work aims to bridge queueing models and simulation-based studies for rapid assessment of scheduler performance and design tradeoffs. We also consider finite user buffer sizes at the base station and demonstrate that traffic shaping before scheduling improves the quality of streamed video by reducing the amount of data loss.