On Link Layer Prioritization for Wireless Communication

In this paper, we develop Prioritized Adaptive Code-Enhanced (PACE) link-layer protocol to achieve preferred data recovery order across connections, while maintaining stable and reliable data flow over a wireless network. We classify link-layer traffic arrivals into different priorities based on delay constraint and distortion associated with that traffic. We formulate the link-layer buffer as a multiclass M/G/1 priority queuing system and the decoding process by a nonhomogeneous Geometric density function. This formulation enables the determination of an optimal dynamic decoder scheduling for heterogeneous link-layer traffic. PACE employs rate-adaptive Low Density Parity Check (LDPC) codes for error recovery. We demonstrate experimentally that PACE reduces the throughput-delay cost by 20%-70% in comparison with the IEEE802.11 ARQ and Hybrid ARQ (HARQ) protocols. Further, it achieves 20%-60% throughput and 2-10dB PSNR improvements in channel bandwidth utilization and realtime video playback quality.