【作者】 顾德；

【导师】 宋执环；

【作者基本信息】 浙江大学 ， 控制科学与工程， 2012， 博士

【摘要】 无线传感器网络(Wireless Sensor Networks, WSN)正在得到越来越多的关注和研究,其应用也在逐渐增多。在大规模应用中,均匀部署的WSN拓扑概貌主要由部署区域的几何形状决定。由于按事先规划进行部署的方案基本不予考虑,路由等问题必须在随机部署完成后进行自组织。为了能从大局上优化整体性能,必须首先对WSN的整体概况有所了解,其中拓扑瓶颈和边界是最重要的整体信息。本文通过对比发现,均匀部署的WSN的节点分布情况和组成物质的分子分布情况在微观上具有一定的相似性。微观特征相似的累积很有可能导致宏观特征相似。为此,本文提出了通过模拟物理过程寻找WSN拓扑边界和瓶颈的思想。首先寻找到了两种物理过程,能够使处在物体几何边界和瓶颈位置的分子群表现出特点。本文发现在封闭有边界的连续空间中对物体进行一定的刺激,通过观察热传导过程中某一时刻的温度场,汇聚到温度梯度为零但非局部极高温点的温度梯度线,可以确定为连续空间的瓶颈；通过观察物质扩散过程中某一时刻的浓度分布,等浓度线的中断处可以确定为连续空间的边界。为了辨识WSN的拓扑边界,本文提出了一种仅依赖节点间相邻关系的分布式算法。算法首先在WSN中模拟物质扩散过程,建立虚拟的物质浓度分布,然后通过比较WSN各个节点的虚拟物质浓度值,在WSN中建立起近似等浓度线,再通过判断近似等浓度线的中断点,最终辨识出WSN的边界节点。该算法的结果为下一步确定WSN瓶颈提供了必要的前提条件。为了识别WSN的拓扑瓶颈,在WSN拓扑边界已经确定的前提下,本文提出了一种仅依赖节点间相邻关系的分布式算法。首先在WSN中模拟热传导过程,建立虚拟的温度场,然后通过各节点在邻节点中选择虚拟温度值最高的作为父节点的步骤,在WSN中形成若干拓扑树,最后各节点通过判断其邻节点中是否有属于不同拓扑树节点,最终识别出WSN的瓶颈节点。本文所提出的算法,思想清晰,易于理解；算法前提简单,有广泛的应用的潜力。算法的输出结果,反映了WSN拓扑结构和WSN部署区域的全局概貌,能对其他WSN优化设计提供有用的信息。更多还原

【Abstract】 Wireless Sensor Networks (WSN) is becoming increasingly promising in practice, which arouses the global interest and research about it. The underlying geometric feature of the deployment field fundamentally affects the overall topology of the WSN in large scale applications. As the pre-deployment design and optimization are usually unpractical in random deployment scenarios, the global optimum of the WSN’s performance is achievable only if the topology dependent self-organizing process acquires the overview of the WSN, in which the bottleneck and the boundary are the most important.The distribution states are quite similar between the microscopic view over molecules and the uniformly random distributed WSN nodes. The similar microscopic features could very likely lead to a similar macroscopic characteristic.An idea is proposed that the boundary and the bottleneck of WSN could be recognized by simulating physical processes. Two phenomena are investigated first, as some features could distinguish the molecules at the bottleneck or the boundary and molecules at other positions. By observing the temperature field in heat conduction process after certain stimulation, the bottlenecks show up at the gradient lines that end on the non-extremal points where the gradient equal to 0. By observing a certain mass diffusion process, the boundaries could be identified at the points where the isoconcentration surfaces break.A distributed algorithm, which only requires the information about the adjacent relation, is proposed for detecting the boundary nodes of the WSN. After simulating the mass diffusion process in WSN, the virtual concentration field sets up and could be investigated. Then the topological boundaries of the WSN are recognized by identifying the break points of the virtual isoconcentration lines. The output of the algorithm also provides the vital information for recognizing the bottleneck.On the basis of boundary detection, a distributed algorithm, which only requires the neighboring information, is proposed for recognizing the nodes on bottleneck section of the WSN. After simulating the heat conduction process in WSN, a virtual temperature field sets up. When every node selects the hottest node in neighborhood as parent, some topological trees appear in the WSN. The topological bottleneck could be recognized by identifying the nodes in whose neighborhood there are nodes belonging to different trees.The idea of the proposed algorithms is clear and easy to understand. The assumption of the algorithms is rather weak, which suggests a potential of wide range applications. The output of the algorithms provides the overview of the WSN topology and the shape of the field, which could be very valuable for other algorithms dedicated to optimize the overall performance of WSN.更多还原