基于强化学习的无线传感网栅栏可信信息覆盖优化理论与方法

Barrier coverage is one of the most decisive and critical issues in wireless sensor networks. Most of the current related works on barrier coverage are based on the idealistic disc coverage model and assume the priori information including the sensing scenarios and network topology, which are not suitable and applicable for practical applications. Motivated from the critical cyber-enabled border safety surveillance application, this project focuses on the barrier coverage based on the reinforcement learning, learning automata and confident information coverage model. The major investigations of this project are as follows: (1) We study how to efficiently construct single-modal and multi-modal confident information coverage barrier paths based on the learning automata theory and model, which can save the network deployment cost and energy consumption; (2) For maximizing the network lifetime, we devise single-modal and multi-modal barrier confident information coverage sensor sleeping scheduling strategies by making full use of the designed efficient learning, rewarding and penalty mechanisms; (3) In order to improve the coverage performance, we propose a dynamic irregular cell learning automata model and investigate how to detect and mend the single-modal and multi-modal barrier confident information coverage gaps; (4) We will evaluate and cross-verify the performance of the proposed models and algorithms by experimental simulations and a practical network prototype. The study of this project can significantly balance the energy consumption, prolong the network lifetime, optimize the network barrier confident information coverage performance, and improve the network quality of service. The research of this project will remarkably extend the current theory and methodology framework of barrier coverage, and also will obtain significant development for the barrier coverage optimization theories and practical applications.

栅栏覆盖问题是无线传感网研究中的核心问题之一。现有研究大多采用过于理想化的圆盘覆盖模型且假设具有感测环境和网络的先验知识，这与实际应用场景尚有差距。本项目面向边界安全监测关键应用需求，基于可信信息覆盖模型，采用强化学习机制下的学习自动机理论，研究无线传感网栅栏覆盖问题。研究内容包括：(1)基于学习自动机理论，研究单模和多模可信信息覆盖栅栏路径构建方法，节约网络成本；(2)构建合理学习和奖惩机制，设计单模和多模栅栏可信信息覆盖节点休眠调度策略，最大化网络寿命；(3)建立动态不规则元胞学习自动机模型，提出单模和多模栅栏可信信息覆盖空隙检测和修补算法，提升覆盖性能；(4)通过仿真及真实平台对所提模型和方法的性能进行交叉验证。通过本项目的研究，可均衡能量消耗，延长网络寿命，优化栅栏覆盖性能，提升网络服务质量。本项目的研究可拓展现有无线传感网栅栏覆盖理论和方法体系，具有重要理论价值和实践意义。

栅栏覆盖问题作为无线传感网研究中的核心问题之一。现有研究大多采用过于理想化的圆盘覆盖模型且假设具有感测环境和网络的先验知识，这与实际应用场景尚有差距。本项目面向边界安全监测关键应用需求，基于可信信息覆盖模型，采用了强化学习机制下的学习自动机理论，较好地解决了无线传感网栅栏覆盖问题。研究内容包括：(1)基于学习自动机理论，研究了单模和多模可信信息覆盖栅栏路径构建方法，有效地节约了网络成本；(2)构建合理学习和奖惩机制，设计了单模和多模栅栏可信信息覆盖节点休眠调度策略，从而最大化网络寿命；(3)建立动态不规则元胞学习自动机模型，提出了单模和多模栅栏可信信息覆盖空隙检测和修补算法，显著地提升了覆盖性能；(4)通过仿真及真实平台对所提模型和方法的性能进行了交叉验证。通过本项目的研究，实现了均衡能量消耗、延长网络寿命、优化栅栏覆盖性能、提升网络服务质量，较好地完成了预期研究目标。本项目的研究已拓展了现有无线传感网栅栏覆盖理论和方法体系，具有重要理论价值和实践意义。