【作者】 刘洪涛；

【导师】 程良伦；

【作者基本信息】 广东工业大学 ， 控制科学与工程， 2012， 博士

【摘要】 高性能传感器网络(High-Performance Sensor Networks, HPSN)作为信息物理融合系统(Cyber-Physical Systems, CPS)和物联网(Internet of Things, IoT)的重要网络支持层,它的研究涉及了体系结构与评价标准、多元数据流的区分服务、实时可靠数据传输、服务质量(Quality of Service, QoS)保证等,可以解决传统传感器网络无法处理高质量的视频监控、多元化信息采集和处理、复杂任务的调度和对物理对象的控制等诸多问题,同时为业务提供QoS保证,可在军事国防、工业制造业、智能电网、智能交通、抢险救灾等领域获得广泛的应用,获得国内外研究者的广泛关注。在通信和控制方面的应用复杂性不断扩展,需要高性能传感器网络为不同业务提供足够的业务可用性保障,然而高性能传感器网络节点的共享、多跳、分布式、信道非对称、资源受限和拓扑结构动态变化等问题都为其带来巨大挑战。人们要求能在高性能传感器网络中传输有QoS保证的多元化业务,而获取高性能传感器网络节点和路径的可用带宽信息是一个重要前提。只有得到节点和路径的可用带宽信息才有可能使带宽敏感的信道接入协议、QoS路由协议和实时可靠的数据传输协议得到高效运行,达到最大化地利用网络资源的目的。针对如何获取可用带宽信息问题,本文重点研究高性能传感器网络体系结构与评价标准、多元业务的区分服务、节点可用带宽估计、路径可用带宽估计等内容,主要包括：(1)采用跨层优化设计方法,设计了一种高性能传感器网络体系结构。并对高性能传感器网络体系结构的设计思想、主要特点、各层功能模块进行了详细的阐述,提出了系统性能评价指标及评价方法。该结构不仅具备高性能和能量高效等特征,还能为面向高性能的传感器网络各类协议的研究提供一个统一的问题描述和分析的理论框架。(2)针对传统传感器网络不能够为具有不同QoS要求的通信流提供区分服务问题,提出了一种区分服务与速率控制策略。该策略将网络中的通信流分为了六类,每一类通信流设置为不同的优先级,父节点采用基于优先级的速率调整算法为每个子节点分配不同的最大发送速率,从而使高优先级的数据流获得高吞吐量和低延迟,达到区分服务的目的,并能够保证网络吞吐量的稳定。(3)针对节点竞争无线信道导致节点带宽测量不准确的问题,提出了基于网络分配向量的节点可用带宽估计策略。在该策略中,节点首先根据平均竞争窗口大小预测碰撞概率,进而估计出数据帧重传次数；然后通过收集邻居节点周期性发送的Hello分组,获得邻居节点的网络分配向量值；最后采用统计平均的方法估计出节点可用带宽。仿真结果表明该策略能够准确地估计高性能传感器网络节点可用带宽,并且不用发送主动探测分组,不影响网络业务的正常运行,具有较为广泛的适用性。(4)针对源节点无法获知网络状态信息而发送大量数据包,导致高性能传感器网络发生拥塞的问题,提出了一种新颖的路径可用带宽估计策略。该策略在MAC层中引入一个跨层模块,估计出MAC层的带宽和延迟信息,然后利用TCP协议的选项字段进行逐跳累积传输,最后反馈到源节点,用于调整源节点TCP协议的拥塞窗口和慢启动阈值,进而调整源节点数据包发送速率,达到缓解网络拥塞的目的。该策略可以充分利用多变的网络可用容量传输尽可能多的数据包,保证网络资源的利用率,使得高性能传感器网络的可用性能明显提高。(5)针对RFID网络与传感器网络分离建设,不能够充分利用网络资源的问题,设计了一种高性能智能节点,用于融合RFID网络与传感器网络,达到充分利用网络资源的目的。重点设计了高性能智能节点的硬件和软件结构,确保节点具有性能稳定、可靠,功耗最小的特点。同时,针对汇聚节点固定容易造成“热点”问题,设计了一种基于移动汇聚节点的农业立体污染监测系统。该系统利用汇聚节点在监测区域定时移动策略,达到平衡传感器节点能量消耗的目的,有效解决传感器网络环境监测系统的“热点”问题,因而延长了网络的生命周期。整个系统能够远程在线实时监测,具有节能、高效、实时、便捷的优点,在环境监测中具有广泛的应用前景。更多还原

【Abstract】 The High-Performance Sensor Networks(HPSN) is an important network layer in the Cyber-Physical Systems(CPS) and the Internet of Things(IoT). It involves the study of the architecture and the evaluation criteria, differentiated services for multiple data streams, real-time and reliable transport protocols, the quality of service(QoS) guarantee, and so on. HPSN can solve the high-quality video surveillance, and diversified information acquisition and processing, scheduling of complex tasks and control of physical objects, and many other issues, which traditional sensor network can not handle. At the same time, it can provide QoS guarantees for business. HPSN obtained a wide range of applications in the field of military and defense, industrial manufacturing, smart grid, intelligent transportation, rescue and relief, and get the extensive attention of domestic and foreign researchers.The complexity of applications continues to expand in communication and control. It makes HPSN that must provide adequate service availability protection for different business. However, nodes of HPSN have characteristics of shared, multi-hop, distributed, channel asymmetric, resource-constrained and dynamically changing topology, and so on, which are its tremendous challenges. It is hoped that the diversified business with QoS guarantee can be transmitted in HPSN. However, the acquisition of available bandwidth information is an important premise for HPSN to support QoS. By utilizing available bandwidth information, it is possible to develop the bandwidth-sensitive channel access protocol, QoS routing protocol, real-time reliable data transmission protocol, and so on. So it is possible to maximize the use of network resources.To address how to obtain the available bandwidth, the dissertation focuses on the High-Performance Sensor Networks architecture and evaluation criteria, the differentiated service of diversified business, the available bandwidth estimation of nodes, the available bandwidth estimation of paths, and so on. The main contents include:(1)A HPSN architecture is designed using cross-layer optimization design method. First, the HPSN design ideas, the main features and functions of each module are described in detail. Finally, the system performance evaluation indexes and evaluation method are presented. The architecture not only has the characteristics of high performance, energy efficiency and so on, but also provides a theoretical framework of unified description and analysis for the study of various types of protocols of HPSN.(2) Traditional wireless sensor networks can not provide differentiated services for traffics with different QoS requirements, a differentiated service rate control strategy is proposed. The strategy divides network traffics into six categories, and set each type of traffic with a different priority. The parent node assigns a different maximum transmission rate for each sub-node utilizing a priority-based rate adjustment algorithm. Thereby, the high-priority data stream obtains high throughput and low latency. The purpose of differentiated services for different streams is achieved, and the stability of network throughput is ensured.(3) To address the issue of inaccurate measurement of the node bandwidth, an available bandwidth evaluation strategy based on the network allocation vector is proposed. First, the node predicts the collision probability according to the size of the average contention window, and thus estimates the number of retransmission of the data frame. Then the node gets the network allocation vector by collecting Hello packets periodically sent by neighbors. Finally, the node estimates the available bandwidth using the statistical average method. The simulation results show that the strategy can accurately estimate the available bandwidth of HPSN. The strategy does not send the probe packets, and will not affect the normal operation of the network business. So it has the characteristic of wider applicability.(4) The source node sends a large number of data packets into the network without network status information, resulting in HPSN congestion, a novel path available bandwidth estimation strategy is proposed. A Cross-Layer Module(CLM) is introduced in MAC layer. The CLM estimates the bandwidth and delay information of the MAC layer, and then transmits the information hop by hop using the TCP protocol options field. Finally, the information is fed back to the source node. The source node adjusts the congestion window and slow start threshold of the TCP protocol using the information. Thus the packet sending rate is adjusted, and the network congestion is mitigated. The strategy can take full advantage of the changing network available capacity to transmit packets as much as possible, so as to ensure the utilization of network resources, and significantly improved the performance of HPSN.(5) RFID networks and wireless sensor networks separately construct, so network resources can not be able to make full use. A high-performance smart node is designed for the integration of RFID networks and wireless sensor networks. The dissertation focuses on the design of hardware and software structure of the high-performance smart node, which has characteristics of stable performance, reliability, minimum power consumption, and so on. At the same time, due to the "hot spot" issue resulted from the fixed sink node, an agricultural pollution monitoring system based on the mobile sink node is designed. The system uses a strategy of moving the sink node in the monitoring region, balancing sensor nodes energy consumption, effectively addressing the "hot spots" issue, extending the life cycle of the network. The system is capable of remote on-line real-time monitoring, which has characteristics of energy conservation, high efficiency, real-time and convenience. And so, has a wide range of applications in environmental monitoring.更多还原