Threshold-constrained GRU-based cognitive spectrum access algorithm for IoT

Abstract

To address the challenges of poor communication reliability, low efficiency, and high energy consumption encountered by internet of things (IoT) terminals in complex electromagnetic environments and high-concurrency scenarios, this paper proposes a threshold-constrained GRU-based cognitive spectrum access algorithm. The algorithm establishes a prediction-dominated and detection-assisted dynamic sensing strategy. It leverages the GRU model to mine temporal spectrum features, mitigating channel contention conflicts, thereby improving channel access success rate and reducing queuing delay. A threshold constraint unit is introduced to suppress blind access and redundant detection, substantially reducing the communication energy consumption incurred by spectrum detection and channel collision retransmissions. Simulation results demonstrate that the proposed algorithm achieves a synergistic optimization of communication reliability, effectiveness, and energy efficiency while maintaining lightweight deployment, and its overall performance significantly outperforms time division multiple access, carrier sense multiple access with collision avoidance and the DQN-based spectrum access algorithm. Furthermore, the proposed algorithm exhibits strong robustness against severe interference and random bursty traffic, providing a feasible solution for spectrum access of resource-constrained IoT terminals in complex dynamic scenarios.