【作者】 俸皓；

【导师】 罗蕾；

【作者基本信息】 电子科技大学 ， 计算机系统结构， 2017， 博士

【摘要】 随着无线传感器网络应用规模的不断扩大,由于其高度面向特定应用、节点资源严重受限等内在特点,导致了原有网络体系结构阻碍了其自身的进一步发展。近年来随着软件定义网络范式的兴起,研究者将软件定义网络模型和无线传感器网络模型相结合,产生了软件定义无线传感器网络这一新兴的网络体系结构,增强了无线传感器网络的可管理性、可配置性和可复用性,从而为新的无线传感器网络应用模式提供了更加强有力的平台支持。新型网络体系结构的提出,使得原有的计算模式、优化方法不再适用,如何根据软件定义无线传感器网络的特点,设计出新的优化模型,解决无线传感器网络中的固有问题,变得尤为重要。本文在结合软件定义无线传感器网络模型的情况下,充分考虑了传感器网络的特点,针对传感器网络中的任务分配及节点调度、多目标定位、移动性管理等问题展开了研究,主要工作与创新点如下:1)建立了软件定义无线传感器网络中适合大规模节点情况下的任务分配及节点调度模型。该模型引入了合作感知框架,充分考虑了节点的资源约束、任务的可调度性因素以及任务对于感知质量的需求,以最小化网络能耗和实现节点能耗的负载均衡为目标,提出了相应的任务分配与节点调度全局优化算法。同时,考虑到无线传感器网络的动态特性,针对网络运行中的变化情况,设计了相应的局部优化算法,使得调度策略具有更加良好的实时性能。2)建立了适合软件定义无线传感器网络的基于压缩感知的多目标定位模型。针对该模型,设计了基于离散人工蜂群算法的位置重构方法,该算法具有所需参数少、计算复杂度低、收敛速度快的特点。该模型减少了定位过程中所需要传递数据的大小,减轻了网络通信的压力,并考虑了噪声的影响,具有一定的实用价值。3)为保证网络的可用性和降低网络能耗,建立了软件定义无线传感器网络中多移动sink数据采集的模型。该模型将移动节点停留时间等数据采集的相关因素直接作为移动节点路径规划的决策依据,提出了一种更为通用的基于多移动节点的多目标数据采集策略,并将此问题建模为一种时变多旅行商问题模型。设计了一种适合在控制器上运行的、针对问题特点的混合遗传算法来求解多个移动节点的规划路径,并对设计的算法给出了收敛性证明。4)为进一步缩短移动sink节点的路径,降低数据采集的延时,本文考虑了节点的通信范围所带来的影响,将路径规划问题建模为具有邻域的多旅行商模型。针对这一混合优化模型,设计了高效的近似算法进行求解,提高了路径规划的实时性,并给出了算法结果的上界及其理论证明。更多还原

【Abstract】 With the continuous development of the applications of wireless sensor network,since WSN ’ s inherent characteristics such as extreme applications specific,limited node resources,the original network architecture hindered the further development of its own.In recent years,with the emergence of software defined network paradigm,researchers combined software defined network model and the model of the wireless sensor network to produce a new network architecture,which is software defined wireless sensor networks.The software defined wireless sensor network enhanced wireless sensor networks manageability,configurability and reusability,and provides a more powerful platform that support to the wireless sensor networks new application.The new network architecture makes the original computing model and optimization method no longer conform to the development of wireless sensor networks.How to design the new optimization models to solve the inherent problems in wireless sensor networks according to the characteristics of software defined wireless sensor network becomes particularly important.Considering the characteristics of sensor networks and based on the model of software defined wireless sensor networks,the task allocation and nodes scheduling,multi target localization and mobility management in wireless sensor networks are studied in this dissertation.The main research contents and contributions of this dissertation include:1)Build a task allocation and nodes scheduling model suitable for large-scale nodes in software defined wireless sensor networks.The model introduces a cooperative sensing framework,which takes full account of the resource constraints of sensors,the schedulability of the tasks,and the task’s demand of the sensing quality.In order to minimize the energy consumption of network and realize the load balancing of nodes energy consumption,a reasonable global optimization algorithm for task allocation and nodes scheduling is proposed.At the same time,considering the dynamic characteristics of wireless sensor networks,according to the changes in the network’s operation,the corresponding local optimization algorithm is designed,so that the scheduling strategy has better real-time performance.2)Build a multi target localization model based on compressed sensing for software defined wireless sensor networks and design a position reconstruction algorithm based on discrete artificial bee colony to solve the model.The algorithm has the characteristics of less required parameters,low computational complexity and fast convergence speed.The model reduces the size of data needed to transfer and mitigate the pressure of network communication.Meanwhile,the influence of noise is taken into account,it has certain practical value.3)Build a multi mobile sink data collecting model in software defined wireless sensor networks in order to ensure the availability of the network and reduce the energy consumption of the network.In this model several concerned factors such as resident time of mobile nodes and data transfer rate are to be considered as decisive disciplines for the plan path of mobile node,and proposed a more general multi-objective data colleting strategies based on multi-mobile nodes;this proposed data collecting technique is abstracted as a model of time variable multiple traveling salesman problem.Belonging to a discrete optimal problem,the proposed model is solved by with a proposed hybrid genetic algorithm which is suitable for the controller to determine the paths of the multi-mobile nodes.The convergence analysis of the proposed algorithm is given.4)In order to further shorten the path of mobile sink nodes and reduce the delay of data collecting,the influence of the communication range of nodes is considered,and the path planning problem is modeled as a multi traveling salesman with neighborhood problem.To solve this hybrid optimization model,an efficient approximate algorithm is proposed.The real-time performance of the path planning is improved.Finally,the dissertation provides the upper bound of the algorithm and its the oretical proof.更多还原