Indoor Positioning Improvement Based On RSSI Fingerprinting

As the proliferation of mobile smart terminals and wireless networks, the application requirements for Location Based Services (LBS) is growing rapidly, and getting more and more attentions in the field of emergency rescue, medical treatment, social network, navigation, monitoring and so on. Because LBS need the support of location information, the development of positioning technology is bind with the development of LBS closely. Reliable and efficient positioning technology is the key to high quality LBS.According to different usage scenarios, positioning technology is mainly divided into outdoor positioning and indoor positioning. Highly mature satellite positioning technology and cellular network positioning technology have become a de facto standard in outdoor environment. Because outdoor positioning is difficult to be applied into complex indoor environment, the indoor positioning gradually shows good prospects for research and application. Among indoor positioning technology, Received Signal Strength Indicator (RSSI) fingerprinting is widely used. Benefited by high positioning accuracy, simple deployment and low cost, RSSI fingerprinting positioning becomes a hot research topic in recent years. This paper makes the following contributions in the improvement of RSSI fingerprinting based indoor positioning technology under different circumstances.Firstly, an improved indoor positioning algorithm based on location continuity is presented, which is to solve the problems, such as the huge positioning error, RSSI fluctuation, and high computational overhead. After the analysis of the internal relations between fingerprint similarity and positioning error, the location continuity is introduced into the algorithm. The algorithm is based on the naive Bayesian method. On the one hand, the fingerprints searching space is reduced, ambiguity locations could be removed, and the probability of major positioning error goes down. On the other hand, the computational overhead of real-time positioning in large-scale scene can be effectively reduced.Secondly, aiming at the problem that the indoor positioning algorithm based on minimum error Bayesian method is not optimal in positioning precision, this paper puts forward an optimal precision fingerprinting positioning algorithm, which is based on the least risk Bayesian method. With the introduction of positioning error loss, the Euclidean or path distance between locations is taken as a reference measurement. Each positioning reaches the minimum expectant error, so the average positioning precision reaches the minimum. The algorithm achieves optimal in average position precision, and avoids the contradiction phenomenon between real positioning effect and the data evaluation results.Lastly, to solve the problem of high accumulated error and resonance error in the continuous positioning algorithm, this paper puts forward a fusion algorithm that localize the continuous positioning process based on Hidden Markov Model (HMM). This algorithm combines with inertial positioning, translates the positioning process into several continuous trajectory segments for decreasing the coupling of the entire process. The accumulated error and resonance error could be decreased efficiently, so this algorithm shows high positioning accuracy and robustness to RSSI fluctuation.