【作者】 王倩；

【导师】 张石；

【作者基本信息】 东北大学 ， 电子与通信工程（专业学位）， 2016， 硕士

【摘要】 随着信息技术的不断发展,无线传感器网络被应用在诸多领域。无线传感器网络的覆盖是一切应用得以顺利发展的前提,对其覆盖控制的研究优化将进一步提高无线传感器的网络性能,扩大其应用范围,加快其发展速度。覆盖控制是判断网络服务质量的重要依据。其主要是在保证网络服务质量的同时,通过一些技术或者是协议优化,达到网络覆盖率的最大化,并且尽量降低能量损耗,达到延长网络生命周期的效果。覆盖控制算法的合适与否不仅影响网络获取信息的能力,也影响网络的生命周期、服务质量等。因此,对覆盖控制的研究具有非常重要的意义。本文利用遗传算法对于组合优化问题的独特优势,将遗传算法作为雏形进行改进应用于无线传感器网络覆盖控制。首先,针对简单的二元感知覆盖模型中产生的覆盖冗余,提出加权多种群遗传算法和约束多种群遗传算法控制无线传感器网络覆盖,采用最大化覆盖率和节点休眠率作为目标函数,将多目标函数问题转化为单目标函数进行优化,利用多种群的协同并行进化维持种群多样性,避免遗传算法陷入早熟。仿真结果表明,所提的多种群遗传算法具有有效性及稳定性,求解性能较加权遗传算法和约束遗传算法均有所改善。其次,针对复杂的概率感知覆盖模型中产生的覆盖冗余,提出基于改进NSGA-Ⅱ算法控制无线传感器网络覆盖,将能耗系数作为目标函数之一,采取多目标遗传算法进行覆盖控制优化,引入删除算子避免陷入局部最优解,采取循环交叉算子保留最优个体,自适应调整交叉概率和变异概率防止算法早期陷入局部最优或进化后期停滞。进行仿真实验后,改进NSGA-Ⅱ可以有效求解Pareto最优解,其次对比该算法应用于二元感知覆盖和概率感知覆盖的效果,概率感知覆盖模型的网络覆盖率较低,且需要较长的求解时间,最后,对比未改进的NSGA-Ⅱ算法,改进后可以有效避免算法陷入早熟收敛并可以降低能耗系数,延长网络生命周期。最后,对本文所做工作进行总结,提出研究发展方向。更多还原

【Abstract】 With the emergence of a number of new technologies for the characteristics of the sensing and intelligent,the information technology switches internet communication into internet of things(IoT)communication,and promote the development of the networking,which is an important part of wireless sensor networks.Coverage control is an important indicator of quality of service of network.It is mainly that ensuring the quality of service of network,at the same time,reducing the energy loss and extending the network lifetime through a number of technology or protocol optimization.Coverage control algorithm not only affects network sensing information,but also affects the life cycle of network and quality of service and so on.Therefore,it is meaningful to study coverage control.This paper used the unique advantage of genetic algorithm for combinatorial optimization problems,improving genetic algorithm to applied in the wireless sensor networks coverage control.First of all,for the coverage redundancy of simple Boolean sensing model,this paper proposed a weighted multi-population genetic algorithm and constrained multi-population genetic algorithm for wireless sensor network coverage control,which used a node sleep scheduling,set the maximum coverage and node dormancy rate as the objective function,transformed multi-objective function optimization as a single objective function optimization,maintained the diversity of the population and avoided the premature convergence of genetic algorithm with the co-evolution in various groups.The simulation results show that,the proposed algorithm have validity and stability,and the performance of the algorithms are improved than the genetic algorithm and the weighted constraint genetic algorithm.Secondly,aiming at the coverage redundancy of complex probability sensing model,this paper proposed an energy efficient cover set selection with improvement NSGA-II,which introduced energy consumption coefficient as the objective function,took a multi-objective genetic algorithm to optimize the selection of set covering,avoided falling into local optimal solution with delete operator,retained the best individual with cycle crossover operator,prevented the algorithm from the later evolution stagnation with the adaptive crossover probability and mutation probability.Finally,this paper made a summary of the work and put forward the direction of research.更多还原