| Nowadays, since the location based service(LBS) becomes increasingly popular, more and more researchers focus on how to position the indoor users. The algorithms that are based on RSSI of Wi Fi are hot, because a large number of indoor Wi Fi facilities are available. In addition, it is easy to get RSSI of Wi Fi by mobile phones. In order to provide location information to users more effectively, this paper has finished the following work. Firstly, several indoor positioning algorithms based on RSSI of Wi Fi are analyzed. To compare KNN algorithm with Clustering KNN(CKNN) algorithm, they are tested in the RSSI dataset collected by an Andorid demo. The test shows that CKNN could position faster than KNN while the precision is not affected. To improve the precision of these algorithms, this paper proposes a HMM-based Wi Fi positioning algorithm, called HMMA. HMMA makes use of users’ general walking speed, the distribution of RSSI of Wi Fi and the distance between two positions to get parameters of HMM model. Since the user’s historical RSSI is taken into account, HMMA could position more precisely than CKNN algorithm. Compared with CKNN, the HMMA decreases the average positioning error by 30%. Secondly, based on the above research on algorithms, this paper accomplishes an indoor positioning system which could switch the positioning algorithms between CKNN and HMMA in Android smart phones. When the Internet is not available, the client of this system can run CKNN to position users in time. Though the precision is limited, this system could still provide users corrdinates in this case. When the Internet is available, the server of this system runs HMMA to position users precisely. The test of this system in real classrooms shows that when running in client, CKNN is faster than KNN by 0.278 s and faster than HMMA by 8.33 s. In addition, the average positioning error of HMMA when the Internet is available is smaller than CKNN when the Internet is not available by 22%. |
PDF Full Text Request