【作者】 沙超；

【导师】 王汝传；

【作者基本信息】 南京邮电大学 ， 信息网络， 2011， 博士

【摘要】 对目标环境进行全面多媒体监控的需求和图像传感器、CMOS(Complementary Metal Oxide Semiconductor,互补金属氧化物半导体)相机以及麦克风等能够无所不在地从环境中捕获多媒体内容的廉价微型硬件孕育了无线多媒体传感器网络(Wirless Multimedia Sensor Networks,WMSNs)的产生和快速发展,作为一种新型的传感器网络,近年来,已经引起了学术界和工业界的高度关注。WMSNs由装备图像传感器、摄像机、麦克风以及其它传感器,可以产生多媒体信息的传感器节点组成,这些具有计算、存储和通信能力的多媒体传感器节点通过自组织方式形成分布式感知网络,具备协作地感知、检索、处理和传输网络覆盖区域内音频、视频、静态图像、数值数据等多媒体信息的能力。大部分传统无线传感器网络(Wireless Sensor Networks,WSNs)测量温度、湿度、光强、目标位置等数值物理现象,而WMSNs与之最大不同在于更多关注于音、视频、静态图像等大数据量、大信息量媒体信息的采集和处理,各个功能层次上都需要新的解决关键问题的技术和方法。同传统的无线传感器网络类似,能耗问题是无线多媒体传感器网络中最受关注的基础问题之一,甚至比传统的无线传感器网络受关注的程度更高。一般而言,节点都是依靠电池供电的设备,而多媒体应用则需要网络具备更高的数据传输速率和更强的处理能力,并将产生庞大的数据量。为有效降低能量消耗,我们在全面分析无线多媒体传感器网络能耗来源与产生原因的基础上,针对多媒体信息传感、信息处理和信息传输的特点,提出一种基于层间协作的无线多媒体传感器网络节能体系,并在此基础上,分别研究无线多媒体传感器网络节能定位方法、节能覆盖与控制方案、休眠调度机制以及面向能量优化的信息处理与数据传输技术。并最终实现一种无线多媒体传感器网络与多类型网络融合接入系统。论文工作和贡献主要包括以下几个方面:(1)提出了基于LQI(Link Quality Instruction,链路质量指示)差值的节能定位方法,通过邻居信标节点的交互,获取相应的路径损耗因子,并根据信标节点和待定位节点间的不同链路质量值之差,利用极大似然估计实现低复杂度高精度定位。同时,构建了一种基于类螺旋状信标移动模型的单信标移动定位方案,降低了定位过程中的能耗开销,并利用移动信标实现多媒体传感节点的方向角度定位。(2)在多媒体传感器网络定向感知模型的基础上,研究其节能覆盖与优化方案。分别提出基于贪婪算法和Voronoi图的多媒体节点传感方向调整方案。此外,为有效降低无线多媒体传感器网络能耗,提出一种基于多目标遗传优化的节能覆盖方法。以节点不同开启顺序建立个体基因组,采用双交叉和最优基因组变异保留并不断产生优秀个体,在保证网络覆盖度要求的同时,开启尽可能少的节点。(3)给出了多媒体节点能耗预测模型,并在此基础上提出一种基于生命期划分的节能数据传输策略。通过能耗判定建立多层簇,并在保证能耗均衡的基础上划分节点生命期,同时利用节点工作状态轮转,进一步延长了网络生存时间。此外,还设计并实现了基于相对信息熵的多媒体信息融合方法和基于相邻节点重叠区域判定的WMSNs休眠调度方案。(4)提出一种多媒体传感器网络节能数据采样与传输方法。通过可变采样粒度的周期性传感,降低采样能耗。同时,基于色块匹配,进一步降低了数据传输开销。此外,对节点数据压缩代价进行评估,使其以最小的能量代价传输数据,并根据多媒体事件流规律,优化网络生存时间。最后,提出一种多媒体传感器网络中的多路径传输方法。综合考虑传输时间、链路剩余能量及传输速率等因素,通过元数据包的探寻,建立三类路径。并利用自适应采样时间间隔调整,降低了网络拥塞的可能性。(5)构建了一个基于无线多媒体传感器网络的多网融合体系。研制实现了低功耗的无线图像、音频与视频节点与无线多媒体传感器网络网关设备,并在此基础上,设计并实现了面向多类型基础设施网络的WMSNs接入系统,以实现对网络的能耗管理。更多还原

【Abstract】 The demand for completely monitoring the target environment with multimedia information and the availability of low-cost, small-scale hardware such as imaging sensors, CMOS cameras, and microphones, all of which can ubiquitously capture multimedia content form the field, have fostered the emergence and rapid development of Wireless Multimedia Sensor Networks (WMSNs). As a new and emerging type of sensor networks, WMSNs have received the immediate and significant attention of the research community and industry. WMSNs consist of sensor nodes equipped with imaging sensors, cameras, microphones producing multimedia content. These multimedia sensor nodes with the abilities of computing, storage and communication self-organize into distributed sensing networks that are able to collaboratively capture, retrieve, process and transmit the multimedia information such as audio and video streams, still images, and scalar sensor data from the environment covered by WMSNs. Most deployed traditional Wireless Sensor Networks (WSNs) mesure scalar physical phenomena like temperature, humidity, light intensity, or location of objects. However, WMSNs focus on the capture and processing of large multimedia information like audio and video stream, still images, which is the main difference from traditional WSNs. Therefore, novel technologies and methods are needed to solve crucial issues in each functional layer.Similar to traditional wireless sensor networks, Energy consumption is a fundamental concern in wireless multimedia sensor networks, even more than that in traditional wireless sensor networks. In fact, sensors are battery-constrained devices, while multimedia applications, which require higher transmission rates and more extensive processing capacity, produce large quantities of data. With the help of analysising the causes of energy consumption and taking into account of multimedia information sensing, processing and transmission, a type of layer collaboration energy saving system for wireless multimedia sensor networks is proposed in this thesis. On this basis, we design an energy saving localization method, a type of coverage and control algorithm for multimedia nodes, a kind of sleeping scheduling mechanism as well as an information processing and data transmission technology. Finally, a multi-network integration and network access system is built up to further reduce the energy consumption of WMSNs. The main works and contributions of this thesis are as follows:(1) An energy-saving localization algorithm based on difference of link quality is proposed in this thesis. Path loss factor could be obtained with the help of neighbor beacon node’s interaction and the high-accuracy localization information could be acquired on the basis of maximum likelihood estimation as well as the difference values of the link quality between beacon nodes and unknown nodes. Meanwhile, a type of spiral-like beacon motion model is created to reduce the cost of energy consumption. With the help of the mobile nodes, sensing direction of multimedia node could also be obtained.(2) Coverage and optimization schemes about WMSNs are also studied on the basis of their directional sensing model. A greedy algorithm as well as a type of voronoi diagram based method on sensing direction adjustment is proposed. Moreover, to effectively reduce energy consumption of wireless multimedia sensor networks, an energy-saving coverage scheme is proposed based on multi-objective genetic algorithms. Genomes are established according to the different opening sequences of nodes and the outstanding individuals are preserved due to the double cross operation as well as the best genome mutations which could open nodes as little as possible when meeting certain network coverage requirement.(3) We put forward a kind of prediction model on energy consumption and propose an energy-saving data transmission scheme based on life period division. Multi-layer cluster was constructed on the basis of energy consuming prediction and the life periods of sensor nodes were divided to ensure the balance of energy consumption. By changing node’s working mode, network lifetime was further extended. In addition, an information fusion method based on relative entropy for multimedia nodes is proposed. Moreover, areas of the overlapping sensing region between adjacent nodes could be calculated and a sleeping scheduling scheme in WMSNs is carried out.(4) An energy-efficient data sampling and transmission method in wireless multimedia sensor networks is proposed in this thesis as well. Energy consumption on sampling is reduced on the basis of periodical sleeping as well as different sampling granularities. Moreover, the transmission overhead is further brought down by means of color block matching. A compression cost estimation method is also described to minimize the node’s power cost on transmission and the network lifetime is optimized according to the regularity of multimedia event stream. Finally, a multi-path transmission method for multimedia sensor networks is proposed se well. By considering the transmission time, residual energy as well as transmission rate of the wireless links, three classes of paths are established with the help of metadata packets. Network congestion rate is reduced as well by adjusting time intervals of sampling.(5) A type of multi-network integration system based on WMSNs is created at the end of this thesis. The wireless image, audio and video sensor nodes with low power consumption as well as the gateway for WMSNs are developed as well. We also build up a wireless multimedia sensor network access system for types of infrastructure networks to realize energy management on WMSNs.更多还原