基于自适应卡尔曼滤波器的WSN定位算法研究

发布时间：2018-06-15 10:32

本文选题：无线传感器网络 + 自适应卡尔曼滤波器　；参考：《南京大学》2017年博士论文

【摘要】：在无线传感器网络的许多应用中,例如环境监测和室内定位等,无线传感器节点需要被部署在一个复杂的环境中。无线传感器网络是一种自组织类型的网络,节点之间通常没有固定的通信链路,它们是以随机多跳的方式将采集的数据传回给汇聚节点的。对于随机部署的无线传感器网络而言,通常不知道采集数据节点的准确位置。因此,定位技术是无线传感器网络研究领域的热点之一。然而,由于无线传感器网络部署环境的复杂性,获得准确的节点位置仍然是一项具有挑战性的工作。本文围绕复杂环境下无线传感器网络定位研究了基于噪声自适应卡尔曼滤波的传感网定位精化、鲁棒的传感网噪声自适应卡尔曼滤波定位和多径环境下传感网噪声自适应卡尔曼滤波指纹定位优化这三个问题。(1)考虑到无线传感器网络是一种能量、计算能力、通信带宽、成本等资源受限的网络,通常,在节点之间最容易获取的接收信号强度被用来确定无线传感器网络中节点之间的距离,进而再根据多个节点之间的相对距离获得节点在网络中的相对位置。然而,接收信号强度的测量噪声会使这种方法的定位精度变差。为了降低测量噪声对定位精度的影响,卡尔曼滤波器被利用来精化定位的结果。由于在实际的部署环境中,噪声的统计特性往往是未知或时变的,所以我们首先提出一种基于现有自适应扩展卡尔曼滤波器的无线传感器网络多维定标定位精化算法。然后,为了进一步提高精化的效果,我们从自适应扩展卡尔曼滤波器推导出了一种新的自适应无迹卡尔曼滤波器,并且基于这种新的滤波器提出了精度更高的无线传感器网络多维定标定位精化算法。前一种算法具有相对更小的计算复杂度,而后一种算法则具有相对更高的定位精度。同时,我们还提出了一种专门用于精化噪声环境下无线传感器网络指纹定位结果的自适应指纹卡尔曼滤波器。广泛的实验结果表明不论噪声是已知还是未知、是否随时间改变,我们提出的算法均能够较好地改善传统卡尔曼滤波器对无线传感器网络定位结果的精化效果。(2)卡尔曼滤波器是通过无线传感器节点的过程模型和测量模型来估计节点的位置、速度等状态信息。通常,由于一些外部和内部的干扰因素,在建立的过程模型和测量模型里分别会包含表示过程噪声和测量噪声的随机变量。目前,如果这两种噪声的统计特性同时随时间改变时,现有自适应卡尔曼滤波器估计的结果会出现较大的偏差,甚至出现滤波器无法工作失去鲁棒性的情况。为了提高基于自适应卡尔曼滤波器的无线传感器网络定位算法的精度和鲁棒性,我们首先提出了一种鲁棒的自适应扩展卡尔曼滤波器。然后,在这个鲁棒滤波器的基础上,进一步推导出了精度更高的鲁棒自适应无迹卡尔曼滤波器。另外,这两种新提出滤波器的鲁棒性被从理论上进行了严格地证明。在过程噪声和测量噪声都时变的情况下,大量仿真实验的结果表明我们提出的这两种卡尔曼滤波器能够确保无线传感器网络中的节点位置被鲁棒和准确地估计。(3)目前基于自适应卡尔曼滤波器的无线传感器网络指纹定位研究主要是关注如何提高算法的精度和鲁棒性,还很少有关于指纹定位在噪声和多径的环境下如何获得最优节点位置估计的研究。因此,我们首先利用自适应卡尔曼滤波器和多目标演化算法优化了噪声环境下的指纹定位结果。由于自适应卡尔曼滤波器只能过滤掉一部分接收信号强度的测量噪声,所以过滤后的接收信号强度仍然残留一个小的噪声,这会影响指纹定位精化的效果。为了将残留噪声对定位结果的影响降低到最小,我们又使用多目标演化算法进一步优化了噪声环境下的指纹定位结果。由于在多径环境下现有RSSI(received signal strength indication)测距模型会导致优化过程中用于计算节点位置估计的指纹权重与位置权重不匹配,所以为了获得更好的指纹定位优化结果,我们根据多径环境下的信号强度与距离关系表达式推导出了多信道加权RSSI的测距模型。广泛的实验结果表明新建立的多目标演化模型和多信道加权RSSI测距模型能够使自适应卡尔曼滤波指纹定位在噪声和多径环境下获得更优的节点位置估计结果。
[Abstract]:In many applications of wireless sensor networks, such as environmental monitoring and indoor positioning, wireless sensor nodes need to be deployed in a complex environment. Wireless sensor networks are self-organized networks with no fixed communication links between nodes, and they transmit the collected data in a random multi hop manner. Back to converging nodes. For random deployed wireless sensor networks, it is usually not known to acquire the exact location of data nodes. Therefore, location technology is one of the hotspots in the research field of wireless sensor networks. However, due to the complexity of the deployment environment of wireless sensor networks, the acquisition of accurate node location is still a one. Challenging work. This paper focuses on the localization refinement of sensor networks based on noise adaptive Calman filtering, robust sensing network noise adaptive Calman filtering positioning and adaptive Calman filter fingerprint localization optimization of sensing network noise under multi path environment around the complex environment of wireless sensor network location. (1) consideration of the three problems. Wireless sensor networks (WSN) is a network with limited resources, such as energy, computing power, communication bandwidth, cost, and so on. Usually, the most easily acquired signal intensity between nodes is used to determine the distance between nodes in the wireless sensor network, and then the relative distance between multiple nodes is obtained by the relative distance of multiple nodes in the network. However, the measurement noise that receives the signal intensity will make the positioning accuracy of this method worse. In order to reduce the effect of measurement noise on positioning accuracy, Calman filter is used to refinement the result of positioning. Because the statistical characteristics of noise are often unknown or time-varying in the actual deployment environment, we first propose a one. Based on the existing adaptive extended Calman filter, a multi-dimensional scaling localization algorithm for wireless sensor networks is proposed. Then, in order to further improve the effect of the refinement, we derive a new adaptive Untraced Calman filter from the adaptive extended Calman filter, and the precision is more accurate based on this new filter. The previous algorithm has a relatively smaller computational complexity and the latter has a relatively higher positioning accuracy. At the same time, we also propose an adaptive fingerprint Calman filter for the fingerprint localization results of wireless sensor networks in the noise environment. A wide range of experimental results show that whether the noise is known or unknown or not, our proposed algorithm can better improve the refinement effect of the traditional Calman filter on wireless sensor network location results. (2) the Calman filter is estimated by the process model and the measurement model of the wireless sensor nodes. The position, speed, and other state information of a node. Usually, due to some external and internal interference factors, the process model and the measurement model will contain random variables that represent process noise and measurement noise respectively. At present, if the statistical properties of these two kinds of noise are changed with time, the existing adaptive Calman filter can be used. In order to improve the accuracy and robustness of the wireless sensor network location algorithm based on adaptive Calman filter, we first propose a robust adaptive spread spreading Calman filter. Then, the robust filtering is used to improve the robustness and robustness of the wireless sensor network location algorithm. On the basis of the device, a more accurate robust adaptive Untraced Calman filter is derived. In addition, the robustness of the two new proposed filters is strictly proved in theory. The results of a large number of simulation experiments show that the two kinds of Calman filters have been proposed in the case of varying process noise and measurement noise. The wave device can ensure that the node location in the wireless sensor network is robust and accurate. (3) the current research on fingerprint location of wireless sensor networks based on adaptive Calman filter is mainly concerned with how to improve the accuracy and robustness of the algorithm, and few about how to get the best fingerprint location in the environment of noise and multipath. Therefore, we use adaptive Calman filter and multi-objective evolutionary algorithm to optimize the result of fingerprint localization in noisy environment. Since adaptive Calman filter can only filter a part of the measured noise of the received signal intensity, the intensity of the received signal is still remaining after the filter. In order to minimize the impact of residual noise on the positioning results, we use multi target evolution algorithm to further optimize the result of fingerprint localization in noisy environment. Because of the existing RSSI (received signal strength indication) range model in the multipath environment, it will lead to the result. In order to obtain better fingerprint location optimization results, we derive a range model for multi channel weighted RSSI based on the expression of signal intensity and distance in multipath environment. The evolutionary model and the multi channel weighted RSSI range finding model can make adaptive Calman filter fingerprint localization in the noise and multipath environment to obtain better node position estimation results.
【学位授予单位】：南京大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：TP212.9;TN929.5

【相似文献】