【作者】 高英明；

【导师】 王立鼎； 金仁成；

【作者基本信息】 大连理工大学 ， 机械电子工程， 2009， 博士

【摘要】 无线传感器网络节点由于具有模块化、自组织等特性而能够被高密度、大规模地部署,因此,无线传感器网络能感知所处环境的细微变化,增强了人们对于广域信息的掌控能力。无线传感器网络在空天、军事、工业、农业、交通、环保、防护及家庭等领域具有十分广阔的应用空间。然而,以目前的技术水平来看,无线传感器网络距离低成本、大规模、长时间工作等指标的要求仍存在较大差距。这就要求无线传感器网络在性能上(如能量均衡和定位精度等)有新的提高,也需要理论上有所突破。本文考虑到技术的发展趋势,采用32位微控制器和ZigBee技术自主开发了新型无线传感器网络演示系统。对开发过程中涉及到的能量管理技术问题进行了讨论;基于随机优化方法对无线传感器网络能量均衡问题进行了研究。分析了32位微控制器LPC2138的低功耗特性和能量管理功能,基于LPC2138研制了3款具有动态能量管理功能(DPM)的无线传感器网络节点。采用事件驱动方式,应用传感器网络节点Ⅲ完成了与所研项目其余2个子系统的接口设计,满足了项目对于无线传感器网络系统连续工作时间的指标要求。基于ZigBee技术开发了新的通信协议栈。改进了非时隙CSMA/CA算法,解决了确认帧碰撞问题;简化了AODV路由算法,使其具备了一定的能量高效性;针对网状拓扑结构的无线传感器网络能量管理问题,提出以自较准方式同步节点的工作状态,以退避方式补偿命令发布延时对节点工作状态同步性的影响。观察到无线传感器网络节点最低功耗干涉现象。即传感器网络节点最低功耗并不是微控制器LPC2138与其它组成器件最低功耗的简单叠加,LPC2138的引脚配置会影响自身和与其相连器件的功耗。这为进一步降低节点的功耗提供了思路。针对传感器节点发射/接收功率相差不大的情况,修正了随机电池模型关于电荷单元的定义,引入节点的按需唤醒机制,使简单的单脉冲随机电池模型能够研究无线传感器网络的能量均衡问题。定义了无线传感器网络基本单元,该单元包括1个信源节点、多个路由节点和1个信宿节点。基于马尔可夫决策过程理论,建立了均衡基本单元节点能量的数学模型。模型用冗余路由节点电池的当前名义容量表示系统的状态,以电池的恢复概率计算系统的状态转移概率,以电池恢复的电荷数最大确定最优值函数。为减少运算量,基于线性规划理论给出一种便于求解的方法。更多还原

【Abstract】 Sensor nodes of wireless sensor networks can be deployed with high density and large-scale due to its characteristics of low cost and self-organization,which make wireless sensor networks reflect subtle changes of environment and then enhance humman capability of acquiring and controlling information.So the application scenarios of wireless sensor networks are wide,such as military,industry,agriculture,transporation,family and environment protection,et al.However,there is great gap between the present technology level and the final destination,for example,low cost,low power and long work time of sensor node.The pratical utilization target needs performance improvement and theory breakthrough concerning wireless sensor networks.However,there is great gap between the present technology level and the final destination,for example,low cost,low power and long work time of sensor node.The pratical utilization target needs performance improvement and theory breakthrough concerning wireless sensor networks.Considering development direction of technology,this paper develops a novel wireless sensor networks demonstration platform based on 32-bit micro-controller and ZigBee.Energy management technologies adpoted in the development process are discussed.Energy balance theory issues of wireless sensor networks are researched based on stochastic optimization.Three kinds of sensor node with dynamic power management(DPM) ability are developed using 32-bit micro-controller LPC2138,low power consumption performance and energy management arthitecture of LPC2138 are analyzed.Through event-driven method,the interface design with other two sub-system of researched project is accomplished using sensor nodeⅢ.The design satisfies evaluation index on continuous work time of project.A novel protocol stack based on ZigBee is developed.Un-slotted CSMA/CA is improved in order to avoid collision of acknowledge frame.Ad-hoc On-Demand Distance Vector(AODV) routing algorithm is simplified,which makes AODV energy efficient. According to mesh topology of wireless sensor networks,this paper puts forward self -calibration method to synchronize work state of sensor nodes,instructions sending delay are compensated using back-off mechanism.The minimal power consumption interference phenomena of sensor node are observed. Sum of components minimal power consumption does not equal to the whole sensor node miminal power consumption,pins configuration of LPC2138 affect power consumption of itselt and other components connected with LPC2138.It provides design guideline to decrease power consumption of sensor node.According to the condition that transmitting/receiving power consumption of sensor node nearly equal,this paper revises definition of charge unit defined by stochastic battery model,introduces waking-up meehanisim according to needs of sensor node,so that single pulsed stochastic battery model can be used to research energy balance issues of wireless sensor networks.Basic unit consisting one source node,several intermediate node and one destination node of wireless sensor networks is defined.Mathematical model used to balance energy consumption of basic unit and based on markov decision process is presented.The mathematical denotes system state through current battery nominal capacity,computes probability of state transition through battery recovery probability,choose decision criterion through recovery charge unit number being maximal.The results can be considered as reference to develop more efficient and optmal algorithm.更多还原