【作者】 解文斌；

【导师】 陈永光；

【作者基本信息】 国防科学技术大学 ， 信息与通信工程， 2009， 博士

【摘要】 无线传感器网络提供了一种新的信息感知和采集方式,是未来普适计算的重要技术基础。随着技术的成熟和市场的推广,无线传感器网络将应用于越来越多的领域,其中最主要的一个应用就是监测。本文的研究工作是以战场监测为主要应用背景,从无线传感器网络的流量分布特性、负载均衡、在线数据收集、拓扑控制、移动数据收集和安全性等多个方面展开系统研究,为建立基于无线传感器网络的战场监测系统提供了基本的理论和技术基础。本文取得的创新性成果在于:(1)从微观角度研究了拓扑传输结构对于传感器网络数据传输的影响,提出了多对一的离散网络空间下任意节点负载密度的定义和分布式算法。节点负载密度可以准确描述离散空间下传感器网络的流量负载分布,并由此估计出各个节点的能耗速率差异。给出了准负载均衡的条件,通过改变拓扑传输结构能够实现绝大多数节点的负载均衡。提出了一种多Sink自适应部署算法,能够把整个传感器网络近似均匀地划分为多个子网。(2)提出了一种低延迟和高能效的在线路由算法ODMLR,结合了最短路径路由策略和最大剩余能量策略的优点,能够根据节点的剩余能量及其在分层网络拓扑中的位置动态选择最优的路由策略。ODMLR是一种分布式局部化算法,只需要根据局部信息动态决定下一跳,适合于拓扑动态变化的无线传感器网络,即使不知道查询请求和事件源的先验知识,也可以实现低延迟和高能效的自适应在线路由。(3)提出了一种新的拓扑感知的最小连通支配集(MCDS)启发式算法——TACDS,并证明了其正确性。提出了一种节点转发因子的新概念,能够准确描述节点对于整个网络拓扑的影响力。以局部转发因子作为选择支配节点的优先级标准,减小了支配节点选择的盲目性,TACDS算法能够根据2跳内的局部拓扑信息快速构造出较小的CDS,从而得到基于该支配集的虚拟骨干网。仿真结果表明该算法优于其他的分布式CDS算法,可以更好地近似MCDS。(4)提出了一种移动Sink的自适应移动策略。基于TACDS算法,由传感器网络通过自组织方式自动解决Sink驻留节点集的选择问题。利用遗传算法确定最优的遍历路径,降低了不可控制的随机移动所导致的高延迟和低效率,在低延迟的同时保证了对于传感器网络的全覆盖。通过简单扩展到多移动Sink机制,可以成倍降低数据延迟,同时还提供较高的数据成功发送率。对于存在区域自治子网的传感器网络监测系统,提出了一种基于混合计算模型的传感器网络数据收集机制。(5)研究了移动数据收集中的安全性问题,提出了一种最小特权的双向认证协议。在双向认证过程中同步建立共享的会话密钥,能够保证移动实体和目标节点之间的安全数据传输,并容忍移动实体和传感器节点被俘获所造成的影响。移动实体和目标节点只需要接收一次对方发送的消息,就可以验证对方的合法性。由基站限制移动实体的特权,只赋予移动实体完成数据收集任务所需的最小特权。利用单向散列函数保证数据收集任务的完整性。更多还原

【Abstract】 Wireless Sensor Networks (WSNs) provide a new paradigm for sensing and gathering information from various environments. They are the important technology infrastructure of future ubiquitous computing. With the mature technology and market promotion, WSNs will be used in many and diverse applications. One of the most important applications is monitoring. Taking the battlefield surveillance as the main background, the dissertation lays the theory and technology foundation for establishing WSNs-based battlefield surveillance system, on the basis of the systematical research on traffic distribution, load balancing, online data gathering, topology control, mobile data gathering and security of wireless sensor networks.The primary contributions of this dissertation include:1. The influence of topology structure on data transmission in WSNs is researched from the microscopic scale. The definition of load density of any node in many-to-one discrete network space and the distributed algorithm of load density are proposed. With load density, the load distribution of wireless sensor networks can be described accurately in discrete space. Then, the difference in energy consumption speed of sensor nodes can be estimated. A load sub-balancing condition is present. The load balancing of most nodes can be realized by this condition. An adaptive deployment algorithm for multiple sinks is proposed. The whole sensor network can be divided equably with it.2. A low latency and energy efficient online routing algorithm (ODMLR) is proposed. The algorithm combined the virtue of shortest path routing and maximum residual energy strategy. Each node selects optimal routing strategy dynamically according to its residual energy and situation in topology structure. ODMLR is a distributed and localized algorithm, which selecting next hop node only using local information. ODMLR can effectively extends the network lifetime and produces a small network latency, even if the network topology varies dynamically, without knowledge of future query arrivals and data sources.3. A new topology-aware heuristic algorithm on minimum connected dominating sets (MCDS), TACDS is proposed in this dissertation and which correctness is proved. A new concept of forwarding factor is present, which can describe the importance of nodes to the network topology accurately. By taking advantage of the topology characteristic of nodes, the algorithm can reduce the blindness in the process of selecting dominating nodes, and form a smaller CDS based on 2-hop local information, consequently obtain a virtue backbone network with the CDS. The simulation results show that the algorithm is prior to other distributed CDS algorithms, and more close to minimum CDS. 4. An adaptive movement strategy for mobile sinks in wireless sensor networks is proposed. Mobile sinks can traverse the sensor network in a timely and efficient way by it. Based on TACDS algorithm, the sensor network can determine the sink’s sojourn points autonomously. By using a genetic algorithm, the sinks can get the optimal data gathering tour, reduce the delay and inefficiency resulted by uncontrolled sink mobility, while ensure the complete coverage for the network. The mechanism can easily extend to multiple mobile sinks to reduce the data latency greatly and provide a higher data delivery ratio. A hybrid computing paradigm based data gathering mechanism in distributed autonomy sensor networks is presented.5. A least privilege mutual authentication key agreement protocol is present for wireless sensor networks using a mobile entity. A session key is created in the process of mutual authentication. The protocol provides mutual authentication and secure data transmission between the mobile entity and target nodes, and can tolerate the mobile entity and sensor nodes compromises. The mobile entity and target nodes can be authenticated each other, only need to receive a message from the other side. Based on the principle of least privilege, the base station only grants the mobile entities the least privilege required to accomplish their tasks. The protocol secures the integrity of data gathering tasks by one-way hash function.更多还原