【作者】 掌明；

【导师】 王锁萍；

【作者基本信息】 南京邮电大学 ， 信息网络， 2011， 博士

【摘要】 由微处理器、传感器和无线通信接口组成的无线传感器网络采用全新的数据获取和处理技术,其在军事侦查、环境监测、医疗服务、工业监控等领域广阔的应用前景,引起了学术界和工业界的高度重视。由于传感器节点通常部署在无人值守或恶劣的环境中,能源无法补充或更换,因此协议必须节能;另外大规模的网络部署、自组织的管理方式、节点容易失效、无线信道的动态变化等无线传感器网络固有特性也会影响网络的整体性能。如何设计专门的算法和协议来有效地减少能量消耗和提高网络性能已经成为研究的热点问题。论文围绕节能这一主题,针对分簇的无线传感器网络中与节能和提高网络性能密切相关的算法和协议进行研究。　首先对MAC协议和能量消耗原因进行了分析和研究。基于竞争的MAC协议容易造成局部通信的大量冲突和增加空闲侦听,浪费节点的能量;基于TDMA的MAC协议降低了节点的能耗,但扩展性较差、信道利用率不高。针对以上问题,该文提出了基于分簇的能量高效混合MAC算法(CEEH-MAC),簇内采用基于TDMA机制的基本时隙调整系数来动态调整簇内节点的时隙大小,从而降低数据的传输时延;对不需要传输数据的节点不分配时隙,让其拥有较长的睡眠时间来节约能量;根据节点的剩余能量系数形成时隙分配顺序来减少状态转换的能耗;采用CSMA/CA机制实现邻居簇头通信。该方法从减少串音和空闲侦听两方面来降低能量消耗,通过动态分配时隙大小来减少时延和提高信道的利用率。仿真结果表明,在网络负载较大的情况下,CEEH-MAC算法节能效果明显,具有更小的平均时延和更长的网络生命周期。　其次,对现有的分簇算法进行了研究,为了克服在簇形成过程中角色转换的能耗、周期性簇头转换的计算能耗和确定下一轮簇头转换的时间等问题,论文提出了一种基于概率的能量平衡动态分簇算法(EBDCP),采用虚拟网格技术把簇分成若干个网格,每个网格中只有一个活动节点,其余节点处于睡眠状态;根据活动节点的剩余能量和发送数据包的次数来计算其成为下一轮簇头的概率,减少了计算和通信能耗,同时使得节点能耗具有良好的均衡性;通过动态门限值来确定簇头的动态转换时间。仿真结果说明,该算法节能效果良好,有效地平衡了节点的能量消耗,延长了网络的生命周期。　然后,研究了无线传感器网络的路由协议,针对分层路由协议中存在的能量消耗不均衡、传输时延较大和最优化路由选择等问题,论文提出了基于能量平衡的近似最小生成树路由算法(EBAMST),该算法基于分簇结构,综合考虑了节点的剩余能量、邻居节点数、传输距离等因素来计算成为下一跳的概率,定义了选择下一跳转发节点的衡量标准,簇头在选择路由时,从邻簇头中选择衡量标准最大的作为下一跳路由节点,通过避免让剩余能量较少的簇头大量转发数据来平衡能耗,从而延长网络生命周期,算法不需要复杂的计算和过多的存储空间来减少能量消耗,但得到的路由效果近似最小生成树路由算法。仿真结果表明该算法在能量消耗、平均时延和网络生命周期等方面具有良好的性能。　　最后,对无线传感器网络安全基础设施—密钥管理进行了研究,针对分簇式密钥管理中存在的计算量大、抗毁性弱和扩展性差等问题,论文提出了基于分簇的高效和安全的动态密钥管理算法(EESDKMC),簇头采用改进型的Blom矩阵动态生成簇基密钥,在簇内采用动态簇基密钥和哈希函数生成单向密钥实现簇头与活动节点之间的通信;在簇头之间采用两个相邻簇头密钥的公共部分结合随机密钥生成对密钥进行通信。该算法能根据网络拓扑的动态变化,实时地更新对密钥和单向密钥。仿真结果表明该算法在安全性、可扩展性、通信和计算开销方面具有良好的性能。更多还原

【Abstract】 Wireless sensor networks (WSNs) comprising a microprocessor, sensor and a wireless communication interface, using new data acquisition and processing technologies have gained attention in military, environment and medical fields. Since sensor nodes are usually deployed in unattended or harsh environments, where the power supply cannot be replaced easily, therefore the protocol must be energy-efficient. The inherent characteristics of WSNs such as large-scale network deployment, management self-organisation, resilience to node failure and dynamic wireless channel changes will also affect the overall network performance. How to design specialised algorithms and protocols to better reduce energy consumption and improve network performance has become an important issue. This thesis discusses the theme of WSN energy-efficiency,studies the algorithms and protocols closely related to the improvement of network performance and energy efficient clustering of WSNs.Firstly, analysis and research on MAC protocols and energy consumption has been performed. MAC protocol is likely to cause numerous conflicts in local communication and increase the amount of idle listening, leading to a waste of the nodes energy. TDMA based MAC protocol reduces the energy consumption of the node, but with poor scalability and low channel utilisation. To solve these problems, this thesis presents a cluster-based energy-efficient hybrid MAC algorithm based on Clustering(CEEH-MAC), using TDMA-based mechanism to dynamically adjust the slot size of the nodes in the cluster to reduce data transmission delay. To save energy, the cluster nodes that do not need data transmission do not allocate time slots, extending the sleep time of nodes and reducing the energy consumption of the state transition according to time slots formed by the node residual energy factor. Using a CSMA/CA mechanism to achieve neighbour cluster heads communication reduces energy consumption by reducing crosstalk and idle listening and decreasing the time delay and improves channel utilisation by dynamically allocating time slots. Simulation results indicate that the CEEH-MAC algorithm is energy efficient in the case of greater network load, with a smaller average network delay and longer lifetime.Secondly,we study existing clustering algorithms, in order to overcome the problem of energy consumption in role switching in the cluster formation process, energy consumption of the periodic cluster head conversion, determination of the time to sub-cluster head conversion on the next round, etc. This thesis proposes an energy balance dynamic clustering algorithm based on the probability(EBDCP). This divides the cluster into a number of grids using a virtual grid, in which each grid has only one active node and the remaining nodes are in a sleep state. Then the probability of the active node going into the next round cluster head is calculated according to its residual energy and the number of packets that are sent, reducing the computational and communication energy consumption. A dynamic threshold is used to achieve the dynamic generation of cluster heads in the network. The simulation demonstrates the algorithm improves the energy efficiency of WSNs effectively, balancing the energy consumption of nodes and extending the network lifetime.Then, we study the WSN routing protocol. According to the problems existing in the routing protocol like unbalanced energy consumption, transmission delay, optimal route selection; this thesis gives the energy balanced approximate minimum spanning routing (EBAMST). This algorithm is based on clustering structure, calculating the probability of being next hop by taking into consideration the surplus energy of the cluster head, the number of neighbouring nodes, the transmission distance and other factors, and constructs the lowest cost routing tree dynamically by the Prim. The simulation shows that this scheme has a good performance in energy consumption, average delay, network life cycle, etc.Finally, we study the key management of the WSN security infrastructure, which has some problems such as computation overhead, poor survivability and poor scalability. This thesis presents an energy efficient and secure dynamic key management based on cluster ( EESDKMC),adopting dynamic cluster-based key management and a one-way hash function to achieve the communication between the cluster head and the active nodes within the cluster. Using a modified version of the Blom matrix communication between cluster heads is accomplished. Based on dynamically changing the network topology, the algorithm can update the keys and one-way key instantly. Experimental results show that the algorithm has good performance in security, scalability, communication and computation overhead.更多还原