Analysis On WSN Positioning Error And Precision Control Method

Recent years, localization of wireless mobile devices draw more and more attention.Among them, an important branch is positioning of WSN(Wireless Sensor Networks), in which some location-aware sensors are called anchors and others are unknown sensors.An unknown sensor collects distance or AOA(Angle of Arrival) from itself to anchors in its communication range, and tries to decide its own position. In fact, with radio propagation model, it is possible to deduce T-R(Transmitter and Receiver) distance through RSS(Received Signal Strength). RSS based ranging method is widely adopted and researched,because it doesn’t demand extra hardware, and the algorithm is quite simple. However,there are two drawbacks hindering its application in accuracy crucial scenarios.1) Due to fast and slow fading of wireless channels, accuracy of RSS based ranging method is not satisfying. Slow fading is also referred to as shadow fading, whose influence to ranging accuracy has been widely researched. However, fast fading, caused by multipath e?ect, is rarely discussed. Though it is assumed that fast fading can be mitigated by multi time ranging and averaging, how many times we need to measure RSS values to approach certain precision is still an open question.2) Accuracy of the system is not only decided by that of ranging, but also related to distribution of anchors. This means di?erent sensor has di?erent positioning precision,which makes the system very unstable.Concerning these two problems, contributions of this thesis lie on the following aspects.1) It is very important for RSS based ranging method to estimate parameters of radio propagation model. Theoretically, it is possible to measure RSS values on a small amount of locations, and deploy least square method to compute parameters of wireless channel.This method is proven to be e?ective and accurate in realistic environments.2) Influence of fast fading on ranging accuracy is thoroughly analyzed. Firstly, it is assumed to follow log-normal distribution, and PDF(Probability Density Function) of ranging error is deduced. It is proven that the original ranging method is biased and can be amended by a small correction. To evaluate its performance, CRLB(CramerRao Lower Bound) of RSS based ranging method is deduced, which indicates the lower bound of variance of ranging results. Besides, variance of ranging error is proven to be proportional to T-R distance.3) With ranging results, we will obtain a series of circular equations. In this thesis,we will solve them by linear iteration. Generally, least square method is deployed, which is not optimal in this certain scenario. So, our another contribution to find the optimal iterative matrix to minimize positioning error.4) To improve the integrity and stability of the system, we define GDOP(Geographical Dilution of Precision) and propose APC(Automatic Precision Control) algorithm, which adjusts measure time according to GDOP through iteration. Performance of this algorithm has been proven by both theoretical analysis and simulation results.