基于WSN的分布式传感器系统关键技术研究

发布时间：2018-05-06 09:10

本文选题：无线传感器网络 + 卡尔曼滤波　；参考：《浙江大学》2016年博士论文

【摘要】：无线传感器网络是21世纪以来最重要的技术之一,其广泛应用极大提高了人类认知和改造世界的能力。无线传感器网络有着分布式的特性,传感器节点是网络的基本单元,其计算、存储和通信能力极其有限。如何合理调度节点资源,实现优化配置,构成功能强大的网络,是传感器管理的核心。本文将结合远程网络监测系统,研究传感器的协同状态估计和传输时序规划,讨论如何利用传感器的局部通信,优化测量信息的融合,实现本地状态估计;以及如何规划数据传输时序,确保远程状态估计满足给定精度指标。远程监测系统中存在诸多监测进程,集中式的处理方式涉及海量数据的传输,影响了系统的整体效率。本文研究相对感测网络的分布式估计问题,以去除冗余信息、减少数据传输。网络中所有节点都能测量其自身与邻居所对应的进程之间的相对状态,锚定节点能额外测量进程的绝对状态。本文利用双向图对网络拓扑进行建模,通过构造最优估计器以论述连通图是全局状态估计的基础条件,设计了两种类似卡尔曼滤波器的次优估计器和一种平均一致的次优估计器,实现了进程全局状态的传感器本地估计。无线传感器网络的通信能力有限,为了避免通信冲突并降低系统能耗,本文研究了并行卡尔曼滤波器的时序规划问题。在单信道通信限制条件下,规划传感器的传输序列,使所有进程在远程中心的状态估计时刻满足给定精度。系统中,进程、传感器和远程估计器是一一对应的。本文定义并计算了实时传输截止期限,将基于协方差的时序规划问题变为任务限期安排问题;在此基础上,设计了基于传输周期的离线规划方法和基于滑动窗口的在线调整方法,以适应传输失败和进程估计阈值的变化。系统中存在大量进程时,单信道通信不能保证数据及时传输。针对这一问题,本文研究了自适应的多信道传输规划方法,在保障估计精度的同时减少信道占用数量。本文论证了一种准确计算进程传输周期的夹挤算法;为适应多信道传输,改进了一种基于缓存的离线规划算法,并设计了相应的在线调整算法;多信道算法简化了调整过程中的计算,提高了空闲信道利用效率,缩短了系统变动后的调整时间,更适应分布式网络。对于提出的各种算法设计,本文在相应章节中给出了具体的理论分析,并通过仿真实验进行了验证。
[Abstract]:Wireless sensor network (WSN) is one of the most important technologies in the 21st century. Its wide application has greatly improved the ability of human beings to recognize and transform the world. Wireless sensor networks have distributed characteristics. Sensor nodes are the basic unit of the network, and their computing, storage and communication capabilities are extremely limited. It is the core of sensor management that how to reasonably schedule node resources, realize optimal configuration and form a powerful network. In this paper, based on the remote network monitoring system, the cooperative state estimation and transmission timing planning of sensors are studied, and how to use the local communication of sensors to optimize the fusion of measurement information to realize local state estimation is discussed. And how to plan the time series of data transmission to ensure that the remote state estimation meets the given precision index. There are many monitoring processes in the remote monitoring system. The centralized processing involves the transmission of massive data, which affects the overall efficiency of the system. In this paper, the distributed estimation problem of relative sensing network is studied to remove redundant information and reduce data transmission. All nodes in the network can measure the relative states between themselves and their neighbors, and anchoring nodes can measure the absolute states of the processes. In this paper, a bidirectional graph is used to model the network topology, and an optimal estimator is constructed to show that the connected graph is the basic condition of global state estimation. Two suboptimal estimators similar to Kalman filter and a suboptimal estimator with uniform average are designed to realize the local sensor estimation of the global state of the process. Wireless sensor networks have limited communication capability. In order to avoid communication conflicts and reduce system energy consumption, the parallel Kalman filter timing planning problem is studied in this paper. Under the condition of single channel communication limitation, the transmission sequence of sensors is planned so that all processes meet the given precision at the state estimation of remote center. In the system, processes, sensors and remote estimators are one-to-one correspondence. In this paper, the deadline for real-time transmission is defined and calculated, which turns the covariance-based time series programming problem into a task deadline scheduling problem. The off-line planning method based on transmission period and the on-line adjustment method based on sliding window are designed to adapt to the change of transmission failure and process estimation threshold. When there are a large number of processes in the system, single channel communication can not guarantee the timely transmission of data. To solve this problem, an adaptive multi-channel transmission planning method is proposed in this paper, which not only guarantees the estimation accuracy but also reduces the amount of channel occupancy. In this paper, an algorithm for accurately calculating the process transmission cycle is demonstrated, and an off-line planning algorithm based on cache is improved to adapt to multi-channel transmission, and the corresponding on-line adjustment algorithm is designed. Multi-channel algorithm simplifies the calculation in the adjustment process, improves the efficiency of idle channel utilization, shortens the adjusting time after system change, and is more suitable for distributed networks. For the various algorithms proposed in this paper, the specific theoretical analysis is given in the corresponding chapters, and verified by simulation experiments.
【学位授予单位】：浙江大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TP212.9;TN929.5

【参考文献】