Reinforcement-Learning-Enabled Partial Confident Information Coverage for IoT-Based Bridge Structural Health Monitoring

Internet-of-Things (IoT)-based bridge structural health monitoring (BSHM) has recently attracted considerable attention from both academic and industrial communities of civil engineering and computer science. In conjunction with researchers from civil engineering and computer science, this article studied a fundamental problem motivated from practical IoT-based BSHM: how to effectively prolong network lifetime while guaranteeing desired coverage. Integrating a promising reinforcement learning model named learning automata (LA) with confident information coverage (CIC) model, this article presented an energy-efficient sensor scheduling strategy for partial CIC coverage in IoT-based BSHM system to guarantee network coverage and prolong network lifetime. The proposed scheme fully exploits cooperation among deployed nodes and alternatively schedules the wake/sleep status of nodes while satisfying network connectivity and partial coverage ratio. Especially, the proposed scheme takes full advantage of the LA model to adaptively learn the optimal sensor scheduling strategy and significantly extend network lifetime. A series of comparison simulations using real data sets collected by a practical BSHM system strongly verify the effectiveness and energy efficiency of the proposed algorithm. To the best of our knowledge, this is the first study on how to combine the reinforcement learning mechanism with partial coverage for maximizing the network lifetime of the IoT-based BSHM.

Automated summary

This article addresses the problem of effectively prolonging the network lifetime of IoT-based bridge structural health monitoring systems while ensuring the desired coverage, proposing an energy-efficient sensor scheduling strategy and utilizing the LA model for adaptive learning of optimal scheduling strategy. The proposed scheme demonstrates significant improvements in network lifetime and energy efficiency through simulations using real data sets collected by a practical BSHM system.