群体智能及在无线传感器网络中的应用

【作者】 耶刚强；

【导师】 梁彦； 潘泉；

【作者基本信息】 西北工业大学 ， 控制理论与控制工程， 2007， 硕士

【摘要】 无线传感器网络(Wireless Sensor Networks，WSN)是由一组稠密布置、能量有限、随机撒布的传感器构成的无线自组织网络，其主要功能是智能感知外界环境，并将处理后的信息提供给用户。当随机把节点撒向感兴趣区域，就要使这些节点按照要求工作。在有限能耗下如何组织协调这些节点并使数据能够顺利畅通的传到汇聚节点，关键是要通过这些海量节点的互连，互通，互操作来完成任务。所以解决网络自组织和路由是确保WSN有效工作的基础。 群体智能(Swarm Intelligence，SI)作为一种新兴演化计算技术，已成为研究热点，它与人工生命，特别是进化策略和遗传算法有着极为特殊的联系，完成的理论和应用研究证明群智能方法是一种能够有效解决大多数全局优化问题的新方法。更为重要的是，群体智能的潜在并行性和分布式特点为处理大量的以数扼库形式存在的数据提供了技术保证。由于WSN通过节点的调度自组织来完成任务，其节点相当于群体智能中的个体，如何协调调度这些节点需要一种有效的智能算法来完成，而群体智能算法正适合这种需求。所以引入群体智能算法用于解决无线传感器中的自组织和路由问题，使各节点能够互连，互通，互操作，并使数据有效地进行传输。 本文主要研究了群体智能算法和无线传感器网络中的路由和自组织问题。主要贡献如下： 1．研究了群体智能算法，并首次提出了基于粒子数生灭的动态粒子数微粒群算法。不但提高了种群的探索性能，而且降低了计算量。 2．研究了WSN的路由协议。根据蚁群算法和遗传算法的特点并充分考虑到WSN路由不但要提高数据的传输效率而且要降低能耗的要求，提出了基于蚁群-遗传的WSN路由算法。利用蚂蚁算法分布式和遗传算法集中式的特点，在降低能耗的基础上提高了路由效率。 3．研究了WSN的自组织问题。提出了基于智能的WSN自组织算法，使网络节点能自适应的睡眠唤醒，提高节点利用率，降低节点能耗提高网络寿命。更多还原

【Abstract】 With the advances of technologies in micro-electro-mechanics and wireless communication, it becomes feasible to deploy a large-scale wireless sensor network with thousands of tiny and inexpensive sensor nodes scattered over a vast field so that information of interest can be obtained, processed, transmitted, and fused automatically via node collaboration. Such wireless sensor networks have a wide-range of potential applications, both military and civil, including target tracking, surveillance, and environment control and security management.Swarm intelligence (SI) as a computational intelligence is attentioned by more and more researches. There exist the closed relations among artifical intelligence, evolution algorithm and genetic algorithm. It is widely accepted that swarm intelligence can solve many optimization problems. Our aim focus on SI and its applications in wireless sensor networks, including sensor wakeup control and routing. The main contributions are as fellows:1. Research on the swarm intelligence and propose the dynamic population size based particle swarm optimization. This algorithm not only improves the search ability but also decreases the computation cost.2. Research on the routing protocol of wireless sensor networks, an Ant Colony-Genetic Routing Algorithm is proposed for routing optimization design, in which the communication messages sent by nodes for searching the optimal route are treated as ants with limited life-span. Through the ants’ movement back and forth between source nodes and sink nodes, multiple candidate routing paths can be distributedly obtained. Each candidate path is then considered as a gene sequence and through the selection, crossover and mutation operations on them, the optimal routing path is determined at sink node. The results show that ACGRA can save energy cost and increase the life-span. Besides this, the reliability and adaptation of the network is also improved.3. Research on self-organization of wireless sensor networks. We proposed Swarm更多还原