Local Real-time Positioning Algorithms And Methods For Improving Location Precision

The essence of both pervasive computing and Internet of Things (IoT) is the mutual fusion between physical world and information systems to meet the various demand of people better. Location-aware technology is one of key technologies among pervasive computing and Internet of Things. In recent years, it have been rated as one of new IT technologies among rising rapidly industry applications by the Gartner corporation, an international well-known IT consulting company. Real time positioning systems (RTLS) based on location-aware technology have wide application prospects in healthcare, logistics, mining, security and many other fields. Location acquisition is the key of implementing ubiquitous computing, IoT and LBS.Although there are a lot of positioning techniques for sensing the location of the object so far, different positioning techniques have different location precision, coverage, cost of setup and maintenance and no one can achieve good performance among all kinds of environment. To address the problem, we have proposed and developed a location-aware middleware integrating various position techniques and filtering algorithms after investigating and analyzing existing local positioning techniques and methods for improving location precision. This middleware uses a plug-in architecture, which means that it can be extended conveniently and dynamically. The main contributions of this dissertation as follows:(1) A WLAN fingerprinting algorithm based on the similarity of AP set is proposed. To solve this problem that the fingerprints collected in real environment are usually from different AP set, we propose a WLAN fingerprint positioning algorithm based on the similarity of AP set. At first, a selecting method of training fingerprints based on region similarity is used to narrow the search region and to improve the accuracy of fingerprint matching. Then, fusing the AP sets similarity as well as the received signal strength (R-SS) information to infer distance calculation formula, and compute the coordinate of a mobile target by K-weighted nearest neighbors method.(2) We propose a particle filter-based WiFi/INS positioning algorithm. Using WLAN positioning technique to compute the initial location of a target and DR based on particle filter to calculate its real-time location. To eliminate the cumulative error of DR, we set certain landmarks that are generally key points with specific characteristics such as doors, stairs or the location of APs deployed. When a target moves to a landmark, its location will be updated automatically.(3) This thesis proposes a switch mechanism of indoor and outdoor positioning. We assign different priorities to GPS, WiFi fingerprinting and WiFi trilateration considering their location precision, coverage and complexity, and choose one with high priority when positioning. We define the priority of WiFi fingerprinting outweigh that of GPS, with the lowest priority to WiFi trilateration, to achieve intelligent switch when positioning from indoor to outdoor and vice versa.