Probabilistic Admission Control with Dual-Window Channel Occupancy Statistics for Statistical Priority-Based Multiple...

Statistical priority-based multiple access (SPMA) provides differentiated channel access for multi-priority traffic by comparing the channel occupancy statistic (COS) with hierarchical thresholds. However, existing threshold-setting methods ignore the inherent fluctuation in the COS caused by random traffic arrivals, so that a large fraction of time slots are incorrectly blocked and the channel performance degrades sharply under overload. To address this issue, this paper proposes DW-PA-SPMA, a dual-window probabilistic admission SPMA protocol. Its core idea is to move traffic management from post-hoc threshold interception at the transmission time slot to probabilistic admission at the packet generation instant: based on a capacity budget that relates the available channel capacity to the traffic demand of each priority, excess packets are truncated at the source following the principle of "highest priority first, remaining capacity for lower priorities", so that no queue buffers or channel contention resources are wasted. A dual-window statistic separates instantaneous fluctuation from the steady-state trend, and a dual-EWMA difference method estimates the load variation time scale online, allowing the long window length to adapt to traffic variations without prior load information. Simulation results under a six-segment time-varying load show that the protocol keeps the highest-priority end-to-end completion rate above the 90% target from light load to full load and remains stable under 2.0× overload, whereas existing threshold-based protocols degrade sharply in the heavy-load segment, demonstrating the advantages of the proposed protocol in dynamic traffic scenarios.