Memory-Based Optimal Response Sensor Coverage Optimization Algorithm

To address the limitations of conventional sensor coverage optimisation algorithms in traditional sensor networks—namely their reliance on predefined iteration counts and lack of self-organisation capabilities during random deployment in specialised scenarios—this paper proposes a Memory-Based Best Response (MBRC) sensor coverage optimisation algorithm. This approach models each sensor as an autonomous agent with decision-making capabilities. By constructing a potential game framework, it innovatively incorporates a memory mechanism to achieve synergistic optimisation of global and local information. Theoretical analysis demonstrates that when the memory lengthm≥ 2, the algorithm converges to the global Nash equilibrium with probability 1. Simulation experiments reveal that MBRC exhibits superior performance compared to other algorithms proposed in this paper across networks of varying scales and node counts: atn= 20, coverage reaches 95–100%, with increased memory length yielding enhanced algorithmic stability (reduced coefficient of variation). Compared with simulated annealing, genetic algorithms, and two reinforcement learning approaches, MBRC demonstrates distinct advantages in both convergence and solution quality. Particularly in large-scale networks, its average utility significantly surpasses traditional algorithms. This research offers novel insights for autonomous optimisation in distributed sensor networks, presenting broad application prospects in environmental monitoring, intelligent security systems, and related domains.