基于隐马尔可夫模型的无线传感器网络非视距定位研究

发布时间：2018-01-16 05:12

本文关键词：基于隐马尔可夫模型的无线传感器网络非视距定位研究　出处：《东北大学》2014年硕士论文　论文类型：学位论文

【摘要】：近年来,无线传感器网络这一新兴技术受到国际学术界、工业界的广泛关注,在军事、环境、工业等领域显现出良好的发展态势。其中,获取移动节点的位置信息是无线传感器网络的基本功能。无论是在灾难救援还是在智能家居领域,如何精确地测算出移动节点的坐标信息,对整个传感器网络系统都有着举足轻重的作用。本文针对室内陌生环境,研究非视距的误差的消除策略和算法,并结合仿真实验对其性能进行了分析。针对D/TA(Detection/Tracking Algorithm)算法的不足,本文结合节点自身的运动特点,提出了一种结合了运动惯性的(Improved-Detection/Tracking Algorithm, I-D/TA)定位算法。在完善隐马尔可夫模型(H idden Markov Model, HMM)算法的基础上,相继提出了修正隐马尔科夫模型(Modified-Hidden Markov Model, M-HMM)和更新修正隐马尔科夫模型(Renewal Modified-Hidden Markov Model, RM-HMM)两种改进的定位算法。仿真实验表明,上述三种算法在距离估计和坐标计算环节都有很好的效果,算法定位精度逐步提高,并具有很好的稳定性。考虑到移动节点的速度运动特点,本文提出了一种HMM与IMM混合定位算法,将其速度模型划分为高速模型和低速模型,让节点在移动时不断地评估自己处于两种状态的概率,利用交互式多模型(IMM)对HMM定位的结果进行融合,然后以HMM算法的改进形式与IMM进行融合,以达到更好的精度。仿真实验表明,HMM与IMM混合定位算法在距离估计和坐标计算环节都有很好的效果,所获得的定位精度高于卡尔曼滤波、粒子滤波等其它算法,而且定位结果稳定,具有良好的鲁棒性。在不同的定位环境下,HMM初值定位的条件通常各不相同。针对这一客观事实,本文提出了一种基于粒子群(PSO)和模拟退火混合优化算法,对HMM的初始条件进行优化。并提出了一种降维的优化策略在不影响精确度的情况下加快运行效率,针对传统的边界处理方式容易产生局部最优解的情况提出了两种边界改进处理方法。仿真结果表明,本文提出的初值优化测量与其它的策略相比,具有更快的优化速度,更稳定的优化精度。
[Abstract]:In recent years, wireless sensor network (WSN), as a new technology, has received extensive attention from the international academia and industry, and has shown a good trend of development in military, environmental, industrial and other fields. Obtaining location information of mobile nodes is the basic function of wireless sensor networks. Whether in disaster relief or in the field of smart home, how to accurately calculate the coordinate information of mobile nodes. It plays an important role in the whole sensor network system. In this paper, the non-line-of-sight error elimination strategy and algorithm are studied for the unfamiliar indoor environment. Combined with the simulation experiments, the performance of the algorithm is analyzed, aiming at the deficiency of the D/ TA detection / tracking algorithm. In this paper, according to the motion characteristics of nodes themselves, a novel improved improved detection / tracking Algorithm is proposed, which combines the motion inertia. Based on the improvement of hidden Markov model and H idden Markov model (HMMM) algorithm. Modified-Hidden Markov Model has been proposed one after another. M-HMMM and Renewal Modified-Hidden Markov Model. Simulation results show that the above three algorithms have good effect in distance estimation and coordinate calculation, and the accuracy of the algorithm is improved gradually. Considering the characteristics of velocity movement of mobile nodes, this paper proposes a hybrid localization algorithm of HMM and IMM, which can be divided into high speed model and low speed model. Let the nodes continuously evaluate their probability of being in two states while moving, and fuse the results of HMM localization by using interactive multi-model (IMM). Then the improved form of HMM algorithm is fused with IMM to achieve better precision. The hybrid localization algorithm of HMM and IMM has a good effect in distance estimation and coordinate calculation. The positioning accuracy obtained is higher than that of other algorithms such as Kalman filter particle filter and so on and the localization results are stable. It has good robustness. The initial location conditions of hmm are usually different in different location environment. In view of this objective fact. A hybrid optimization algorithm based on particle swarm optimization (PSO) and simulated annealing is proposed in this paper. The initial conditions of HMM are optimized, and a dimensionality reduction optimization strategy is proposed to speed up the operation efficiency without affecting the accuracy. In this paper, two improved boundary processing methods are proposed to solve the problem that the traditional boundary processing methods are easy to produce local optimal solutions. The simulation results show that the proposed initial value optimization measurement is compared with other strategies. With faster optimization speed, more stable optimization accuracy.
【学位授予单位】：东北大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN929.5;TP212.9

【相似文献】