【作者】 魏博；

【导师】 胡捍英；

【作者基本信息】 解放军信息工程大学 ， 通信与信息系统， 2009， 博士

【摘要】 无线传感器网络(WSN)是当前信息技术领域的一个热点研究方向,由于传感器节点只配备有限的电源,在某些特定的应用场合(比如火场、战场等)下,更换电源是近乎不可能的,这使得传感器节点的寿命在很大程度上依赖于电源的寿命。因此,如何降低传感器节点的能源消耗以延长无线传感器网络的寿命是无线传感器网络设计重点考虑的问题之一。本文主要针对无线传感器网络的节能机制,进行了深入的研究,论文的主要工作及贡献如下：1.提出了一种基于伪群首的分群节能算法—ECPC (Energy-efficient Clustering Algorithm Based on Pseudo Cluster-head)。算法采用中继传输模式,克服了经典分群节能LEACH算法中使用单跳通信带来的能耗不均问题,以及远离sink的传感器节点由于能量消耗过快而过早失效的缺陷。把“伪群首”的概念引入分群算法,将群首与sink节点之间的单跳通信变为多跳通信,从而均衡了节点的能量消耗,避免了长距离传输数据造成的能量消耗。与现有的多跳分群算法相比,ECPC算法为群首进行多跳转发设计了一个高能效的路由算法。仿真结果证明,ECPC算法无论是在网络寿命还是在网络中发送的数据量都方面都要优于LEACH算法以及基于LEACH的改进算法-MLEACH。2.提出了一种群规模受限的无线传感器网络节能算法-SCA(Size Constrained Clustering Algorithm for Wireless Sensor Networks)。算法通过限制群规模的方法,克服了无线传感器网络中在多跳数据传输模式下,由于距离sink节点远近不同,导致的节点能耗不均的hot-spot问题。在SCA算法中,使用限制群规模的方法,减少了群内成员节点数量,降低了群内数据量,从而达到了节省能量的目的；才外,还研究了如何能够通过使用较少的跳数,减少距离群首较远节点的能量消耗。与现有的多跳分群算法具有以下优势：采用非均匀分群的方法,减小了群的规模,从而减少了网络内的数据冲突并均衡了节点的能量消耗。仿真结果显示,SCA算法在网络生存时间方面的性能要优于LEACH算法。3.提出了一种基于极小支配集的无线传感器网络能效算法—ECBD(Energy-efficient Clustering algorithm Based on Minimum Dominating set—ECBD)。算法使用信号接收强度RSSI来选择支配节点,克服了单纯依靠节点度选择群首而带来的吞服量下降等问题。使用RSSI,只需要知道节点大概的物理位置,同时可以减少了节点间通信开销,可以在分群机制中使用最少的能量消耗来维持节点间的转发和通信,从而使支配节点与群内其他节点通信消耗最少的能量,减少整个网络的能量消耗。使用支配集进行分群算法设计,就是将问题建模为寻找群首节点的最优化问题,论文中证明了分布式的ECBD算法的算法复杂度为O(log n log△)。算法中使用节点的本地连接信息、邻节点连通度以及剩余能量数来决定哪个节点应该成为群首节点。仿真结果显示,无论是在分群规模,分群时间还是在单位发送消息的能耗ECBD算法均优于HEED算法。4.在ECBD算法的基础上提出了两种改进算法：将一个节点可以加入距离它k跳的群中,称为kECBD算法。在kECBD算法中,利用多跳通信减少网络中的群首数量,减少了节点的能耗,且在ECBD算法的条件下,群首的选取更为灵活,覆盖相同数量的节点所需的群首数量更少；引入一个运行扩展机制,在算法运行的每一轮中提供更大的群首选择范围,从而可以在较少的轮数内完成分群,称为iECBD。仿真证明,这两种算法的网络性能均优于HEED算法。5.提出了一种基于电路启发式的无线传感器网络节能协议—CHEP。CHEP协议在已知全网信息的情况下,通过集中式的算法,研究了网络路由的最优化问题。协议从无线传感器网络的全网拓扑结构的角度出发,用电路的概念模拟整个无线传感器网络。如果一个传感器节点向距离其较远的节点发送数据,那么在电路中对应的就是该节点通过一个很大的电阻传输电流,这个节点的电压就要提升,提升的电压阻止其他节点向它传输数据,这样就可以保证节点不会远距离的传输数据,平衡了节点的能量损耗,电路网络模拟结果具有平衡的性质。仿真结果显示该协议能够平衡各传感器的能耗,从而达到延长整个网络寿命的目的。在本论文阐述协议思想,并针对网络生存时间以及节点能量分布进行了仿真。此外还针对典型的无线传感器网络模型,同样基于己知全网信息条件下,定义了“健康”网络模型,通过该定义提出了“利益圈”的概念。提出了一种基于特定无线传感器网络模型的能量优化路由算法思想：分步层进算法,并在此基础之上推出了全局能量最优的能耗上限估算,将寻找最优路由归纳为数学问题,并进行了求解。更多还原

【Abstract】 As a new kind of data collection technique, wireless sensor network (WSN) has become one of the hotspot in present information technology field. Wireless sensor node is a tiny device with limited power, and change battery is almost impossible in some applications. This makes the lifetime of nodes mostly depend on the battery, so reduced power consumption to extend the lifetime of network is the primary problem in the research.This thesis carries out a deep research mainly on energy efficient of wireless sensor networks. The main research and contributions are as follows:1. Energy-efficient Clustering Algorithm Based on Pseudo Cluster-head is proposed. In this paper we propose a novel routing algorithm, ECPC, which copes with limited transmission range of each node and brings in a pseudo cluster head to improve the energy efficiency of LEACH. Our solution copes with a fatal drawback of LEACH-each node may have limited transmission range due to energy and hardware limitations. So, data collection at cluster head have to be multi-hop relayed to sink. Further more, simulation results demonstrates our idea. The overall transmitted data and sensor network life time of ECPC is better than that of modified LEACH and much better than original LEACH.2. A Size-constrained Clustering Algorithm (SCA) for Wireless Sensor Network is proposed. By constraining cluster size, the proposed algorithm solves the Hot-Spot problem in Multi-hop Wireless Sensor Networks, which is caused by the unbalanced load of network nodes resulted from their various distances to the base station. By reducing the number of member nodes as well as the volume of data of a cluster, less energy is consumed. It also investigates the problem of reducing the energy cost of the nodes, which are far from a cluster head, by reducing the number of hops. Compared with previous multi-hop clustering algorithms, the proposed SCA algorithm has the following advantage:the unbalanced clustering method reduces the size of a cluster; therefore it reduces the data confliction in the network which helps balance the energy costs of nodes. Simulation shows that the SCA algorithm is superior to the LEACH algorithm in network lifetime.3. An Energy-efficient Clustering algorithm Based on Minimum Dominating Set (ECBD) is presented. This algorithm selects support nodes by received signal strength indicator (RSSI), which solves the decreased throughput problem of the algorithms which select cluster heads simply by node degrees. Using RSSI, only the approximate precise location of a node is required, which can save the communication cost and use the minimal energy cost to maintain the transferring and communication between nodes. Thus, a node costs much less energy to communicate with other nodes in a cluster, and in this way the total energy cost of the entire network is reduced. This algorithm is intrinsically an optimization problem on finding cluster heads, its computational cost is O(log n log△). It selects a cluster head by utilizing the local link information, the link degree to the neighbors, and the remaining energy of a node. Simulation suggests that the ECBD algorithm is superior to the HEED algorithm not only in cluster size, but also in clustering time and in the energy cost per sent messages.4. Two improved algorithms are proposed based on the ECBD algorithm:1) the kECBD algorithm, which allows a node to be added into the cluster in its k-hops. The kECBD algorithm takes advantage of multi-hop communication to reduce the number of cluster head, which reduces the energy cost of a node. Compared with ECBD, it is more flexible in selecting cluster heads, and requires less cluster heads to cover the same number of nodes.2) the iECBD algorithm, which incorporates a running extension mechanism to provide more candidate cluster heads in each round, so that clustering can be finished in less runs. Simulation shows that the performance of these two algorithms is better than that of the HEED algorithm.5. A Circuit Heuristic-based Wireless Sensor Network Energy Saving Policy (CHEP) is discussed. In the case we assumed that the information of the entire network is available, network routing optimization is tackled by a centralized algorithm. Starting from the network topology, this policy models the entire wireless sensor network as a circuit. The circuit network simulation has the characteristic of balancing. If a sensor node sends messages to a node faraway, it corresponds to transmitting large electronic current via a large resistance in a circuit, which prevents a node to transfer long-distance message and so that the energy costs of the nodes are balanced. Simulation suggests that this policy can balance the energy cost of the sensor nodes and hence prolongs the lifespan of the entire network. This thesis presented the idea of this policy and provided simulations on the network life time and node energy distribution.Moreover, in the case that the entire network is observed, considering a typical wireless sensor network model, this thesis defined the concept of a "healthy" network, and proposed the idea of "sphere of interest". It proposed the step-by-step idea of network routing optimization based on the energy of a typical wireless sensor network, and furthermore proposed a method to estimate the upper bound of the global optimization of energy cost. It generated the mathematical problem of finding the optimal route and provided a solution.更多还原