【作者】 刘广怡；

【导师】 于宏毅；

【作者基本信息】 解放军信息工程大学 ， 通信与信息系统， 2009， 硕士

【摘要】 无线传感器网络被认为是21世纪最重要的技术之一,在军事、环境、工业等领域具有相当广阔的应用。其中定位技术因为与很多实际应用直接相关而尤为受到关注。本文针对WSN低成本、高灵活性的特点,对使用接收信号强度指示(RSSI,Received Signal Strength Indicator)进行定位的技术进行了研究,并在已有的软、硬件平台上完成了用于煤矿巷道中的定位系统。本文首先研究了基于信号衰减模型的定位法。对基于RSSI的定位技术研究主要分为两个方面,基于信号衰减的定位法是其中之一。在充分研究国内外研究现状的基础上,本文通过大量的实地测量实验,分析了RSSI的概率特征,进而给出了一种加权最小二乘确定权值的方法。针对信号衰减模型参数未知或估计不准确的情况,提出了两种利用RSSI差值计算位置的方法。遮挡问题是影响RSSI定位精度的一个主要问题,本文对信号传播路径上存在人体随机遮挡的情况进行了一系列实验,通过数据统计分析,最终给出了一种加权概率模型的最大似然估计法,实验证实该方法在遮挡情况下,较ML估计能明显提高定位精度。其次,本文对基于RSSI的定位技术的研究另外一个主要方面——经验模型定位法展开了研究。现有的经验模型定位方法(包括神经网络在内),都存在无法很好适应应用环境变化的问题。针对这一不足,本文提出一种基于动态数据库和位置信标的定位方法,并提出了保守性准则和开放性准则以控制数据库的动态调整过程,通过选取不同的准则可以达到更好的定位效果。利用多次定位结果之间的空间相关性特征,引入贝叶斯滤波器,使得定位精度进一步增加。作为算法研究一个分支,本文利用一个独立的章节对位置估计的克拉美-罗下界(CRLB)进行了推导,并根据CRLB的性质给出了一种衡量参考节点布放对定位误差影响的因子——区域平均定位方差下界。通过这一衡量因子建立了求解参考节点最佳拓扑形式的数学模型,通过求解得到了在区域平均定位方差下界最小意义下的参考节点布放方式。其结论对于节点布设等问题具有理论上的指导意义。在实现部分,结合对于RSSI定位技术的研究,完成了煤矿井下位置监测与管理系统中定位子系统的设计与实现工作。在嵌入式操作系统TinyOS和windows操作系统中分别利用NesC语言和Microsoft Visual C++编译环境完成了定位子系统中信息获取、有线传输和定位计算各个模块。长期测试结果表明,该系统运行稳定,能够为用户提供较高定位精度的服务。更多还原

【Abstract】 Wireless sensor network is regarded as one of the most important technologies in 21st century. It is broadly applied in the fields of military, circumstance and industry. Positioning technique is of great importance because of its direct relative to practical application. Aimed at the low-cost and high-flexibility characters of WSN nodes, this thesis studies the positioning technology with RSSI (Received Signal Strength Indicator); it also develops a mine-laneway positioning system grounded on the existing software and hardware platform.Firstly, positioning technology based on signal attenuation model is investigated, which is one part of the technology with RSSI. After deep research on the status quo at home and abroad, this thesis gives out the probability characters of RSSI by mounts of field measure test, and proposes a method based on weighted least-squares approach. Two positioning methods are brought forward aimed at the unknown or incorrect of signal attenuation parameters. For obstruct is one of key problems affecting RSSI based position accuracy, the situations that there are barriers on the transmitting path are tested repetitiously. From the statistical analysis, ML estimation algorithm based on a model of weighted probability is achieved, and experiments testifies this algorithm can improve position accuracy in case of obstruct environment.Secondly, the other part of RSSI based position, namely empirical model positioning method, is discussed. The problem that algorithms cannot adapt to circumstance variety properly is exiting in all empirical models positioning method including neural network. So, this article proposes positioning algorithm on the basis of dynamic database and location beaconing as well as using two rules to control the dynamic adjustment of database. The introduction of different rules can help to gain different position effect. To get further improvement upon position accuracy, the spatial correlation of multi-position results and bayes filter are introduced.Thirdly, CRLB of position estimation is deduced, and a factor that measure the effect of reference nodes’location on position errors called the lower bound of minimum area average position variance is also introduced. By the measuring factor, mathematical model of calculating optimum reference nodes’topology is built, and the result is gained at the lower bound of minimum area average position variance. The conclusion it is of directive significance for nodes’placement.Finally, associated with RSSI-based positioning technology, position subsystem in miner monitoring and management system is designed and implemented. The modules such as information extraction, wire transportation and position calculation are finished using NesC and Microsoft Visual C++ based on embedded operation system TinyOS and windows system. Long-term tests show that the system runs stable and can provide high position accuracy.更多还原