Energy-efficient medium access control (MAC) protocols remain critical in resource-constrained Wireless Sensor Networks (WSNs) and IoT deployments, especially under mixed traffic patterns that combine event-driven and continuous monitoring operations. Traditional Time Division Multiple Access (TDMA)- and Bit Map Assisted (BMA)-based MAC protocols fail to adapt their duty cycles to spatiotemporal variations in sensor activity, resulting in unnecessary radio wake-ups and increased energy expenditure. To address this limitation, this paper proposes EEI-BMA, an AI-assisted, event-probability-aware MAC protocol that dynamically adjusts transmission scheduling using lightweight neural-network-based event prediction. The proposed framework incorporates per-node probability estimation, adaptive slot activation, and selective channel access to reduce transceiver activity while preserving sensing reliability. Simulation results obtained in the MATLAB environment show that EEI-BMA (Best Prediction) achieves 35--45% lower energy consumption than Traditional-TDMA, 22--30% savings compared with Energy-Aware TDMA, and 18--28% improvement over Traditional-BMA across varying node densities, packet sizes, event-generation probabilities, and continuous monitoring loads. Even with imperfect prediction, EEI-BMA consistently outperforms all baseline protocols, demonstrating strong robustness. The results confirm that prediction-guided MAC scheduling is a highly effective strategy for next-generation low-power WSNs and IoT systems.

翻译：暂无翻译