【作者】 易军；

【导师】 石为人；

【作者基本信息】 重庆大学 ， 控制理论与控制工程， 2010， 博士

【摘要】 无线传感器网络(WSNs)是控制学科的前沿研究方向,其理论与技术发展极大的受到应用驱动。近年来,大量应用场景需要WSNs在具有事件监测能力的同时,还具有任务执行、事件控制功能。为了满足应用需求,典型的WSNs衍生出一种由大量传感器节点与少数执行器节点组成的新型网络结构,学术界称之为无线传感器/执行器网络(WSANs)。无线传感器/执行器网络是一种新型信息获取和处理技术。在无线传感器/执行器网络中,传感器节点需要与执行器节点进行大量的紧密协作,共同完成对环境的监测和对事件的处理,因此出现了一种新的网络现象:传感器节点-执行器节点(SA)协作和执行器节点-执行器节点(AA)协作。如何通过节点间的协同合作,共同完成对环境事件的处理,同时保证网络的实时性和能耗均衡,是无线传感器/执行器网络研究面临的一大挑战。本文面向无线传感器/执行器网络所涉及的理论与技术,针对传统的无线传感器网络相关协议和算法无法完全满足无线传感器/执行器网络新的需求,以节点之间的协同合作为研究手段,以提高网络实时性和能耗均衡为目的,构建SA协作和AA协作模型,基于最优化理论、群体智能优化、图论等计算方法展开理论与算法研究。主要研究成果包括如下四个方面:①针对现有协作算法大多只对下一跳中继节点的选择进行研究,没有考虑网络能耗均衡问题,提出一种基于SA协作模型的分簇算法(CASA)。算法从全网的能耗和时延影响的角度,建立基于SA协作的能耗模型,综合考虑时延、连通度等约束条件,以网络能量优化为目标,构造非线性优化函数,求解网络理想执行器节点个数和传感器节点传输半径等网络分簇所需参数,并在此基础上完成节点的部署和成簇。网络平均时延和生存期等仿真结果表明,相比典型算法,该策略能够在满足一定连通度的前提下,优化网络部署,增强网络实时性和能量均衡性。②针对AA协作中,事件频发区域内节点能耗过快问题,基于分簇结构,提出一种AA实时协作框架(RC)。通过将任务分解成若干任务单元,应用招投标理论、信息论,建立基于熵权的执行代价评价标准,根据网络中执行器节点的实际情况分派任务元,使得多个任务元可以并发执行,缩短执行时间;同时利用多个执行器节点间的协作,将执行能耗分摊到邻居执行器节点,延长网络生存期。仿真、分析结果表明,相比典型算法,RC框架同时兼顾了任务执行的实时性和网络能耗均衡,较好地解决了事件频发区域内节点能耗过快问题。③针对多个有执行顺序限制条件的任务同时在多个执行器节点执行的任务分派问题,提出一种基于AA协作的单目标任务分派算法(SOTS)。建立以能量为约束条件,任务最大完成时间为目标函数的优化模型,并提出结合NEH方法与微粒群优化算法的混合优化算法进行目标函数求解。该算法将微粒群优化算法加以改造,用于任务分派的全局搜索;同时利用NEH方法增加算法局部搜索能力,确保解空间的充分搜索。仿真、分析结果表明,相比典型算法,SOTS搜索速度快,收敛性好,适合解决同时发生多个复杂事件的场合。④针对多个有执行顺序限制条件的任务同时在多个执行器节点执行的任务分派问题,以最大完成时间、能耗均衡指标、存储成本为优化目标,提出一种基于AA协作的多目标任务分派算法(MOTS)。利用理想点法将多目标优化规范化为单目标优化,通过执行器节点角色确定降低问题复杂程度,提出结合基于自适应学习策略的多邻域搜索策略与微粒群优化算法的混合优化算法进行目标函数求解,确保解空间的充分搜索。仿真、分析结果表明,MOTS算法搜索速度快,收敛性好,各执行器节点的实时性、能耗等性能指标均优于比较算法,为解决同时发生多个复杂事件提供了新的思路。更多还原

【Abstract】 The wireless sensor network (WSNs), whose theory and technology development are mostly driven by application, is an emerging research direction in the area of control. In recent years, there are a large number of applications that require the coordination between sensors and higher capability devices to support not only environmental monitoring but also the proper execution of specific tasks. As a result, wireless sensor and actor networks (WSANs) which is integrated by a large quantity of sensor nodes and a few number of actor nodes has been proposed as a new extension of WSNs.WSANs is a new type of information acquisition and processing technology. In WSANs, coordination mechanisms are required among sensors and actors to gather information about the physical world and then perform appropriate actions upon the environment. In particular, new networking phenomena called sensor-actor (SA) and actor-actor (AA) coordination may occur. Currently research in WSANs face a serious challenge that is how to provide distributed information sensing and processing by means of coordination among nodes, the requirement of real-time and energy-balancing characteristic can be ensure at the same time.Most of typical protocols and algorithms for WSNs may not be well-suited for the unique features and application requirements of WSANs. Facing these limitations, based on optimization theory, swarm intelligence optimization, graph theory and other calculating methods, the coordination mechanism of SA and AA is carried out in this thesis to meet the requirement of real-time and energy-balancing characteristic. The important research results are as follows:①Most of the existing collaborative algorithms had been proposed for the next relay node without considering network energy balance. Considering that in WSANs the cluster size, the cluster head number and the residual energy of the node are key indicators of energy-efficient clustering algorithms, the article proposes the CASA, a clustering algorithm based on sensor-actor coordination model, in order to make the whole network energy consumption more balanced. The algorithm establishes the energy consumption model to obtain the optimized number of cluster heads which determine the cluster size. The sensor-actor communication energy consumption is modeled as a nonlinear program to obtain optimal transmission range of sensors and number of actors. Some methods are used to deploy actors and form cluster of heterogeneous sensor and actor network. Considering a few application requirements such as low-latency, connectivity and energy-efficient, performances of the proposed approaches are validated through simulations.②Based on clustering networks strut, a real-time actor-actor coordination framework is proposed to solve“hot zone”problem. In this article, Tasks are partitioned into different task units and the cost of taking action is computed among different actors by using auction method. The actor-actor coordination is formulated as a balanced or non-balanced task assignment optimization problem to achieve more energy-balance. In addition, different task units are executed in parallel to enhance higher real-time response. The result of simulation shows that the algorithm could provide more balance in energy consumption and higher real-time performance.③A single-objective task scheduling approach based on actor-actor coordination for WSANs is proposed to solve the execution problem of ordered execution tasks collaboratively among actors. The purpose of approach is minimizing the maximum response time in the actuators subject to residual energy constraints and schedule execution period of each task operation within given time. The algorithm is based on the principle of particle swarm optimization (PSO), an evolutionary computation technique, which is simple and has fewer adjustable parameters and possesses high search efficiency. Nawaz-Enscore-Ham (NEH) as a local search algorithm employs certain probability to avoid becoming trapped in a local optimum and has been proved to be effective for a variety of situations. Simulation results have shown that the proposed hybrid approach is of high convergence speed and good performance between task response time and balancing the energy dissipation among actors.④In view of the actor-actor task coordination in WSANs, a multi-objective task scheduling approach is proposed. Considering the maximum response time, energy-balanced metric and storage cost, the task assignment among actuators is formulated as a multi-objective optimization problem. A modified ideal point algorithm is used to solve the dimension problem caused by different targets. By translating the multi-object optimization problem into a single-object one, the near-optimum execution period of each task operation would be scheduled in our approach. The algorithm is based on the principle of particle swarm optimization (PSO), an evolutionary computation technique, which is simple and have fewer adjustable parameters and possesses high search efficiency. Multi-neighboring experience as a local search algorithm employs certain probability to avoid becoming trapped in a local optimum and has been proved to be effective for a variety of situations. Simulation results have shown that the proposed algorithm is effective in terms of three performances.更多还原