Impact of Random Packet Arrival on the Performance of Video Streaming over OFDMA Downlink Systems

Streaming multiple video sessions over wireless networks is challenging due to the scarcity of wireless resources and the channel variations. Orthogonal frequency-division multiple access is an appealing solution as it can flexibly schedule multiple users according to a target optimization objective. Channel-aware scheduling (CS), the traditional way of ensuring spectrum-efficient delivery of packets, may not yield the best video performance as the content of a video packet entails a different level of contribution to the decoded quality. Joint channel- and media-aware scheduling (JCMS), a new paradigm, can offer a remarkable performance enhancement by enabling packet prioritization both within intra- and inter-user video flows. These gains, however, are purely derived under the full-buffer scenario. In reality, packets arrive stochastically at the scheduler and its impact on the multiuser video performance is left open. This work aims to bridge this gap by investigating the impact of stochastic packet arrival on the performance of CS and JCMS schemes. Simulation results reveal that a lightweight CS scheme can even outperform two much complicated JCMS schemes.