| LBS (Location-Based Service), for a long time, is a research hot issue. In the recent years, with the rapid development of mobile Internet, intelligent mobile terminal, which appears more and more in people’s work and life, plays an important role. As one of the most basic function of Location-Based Service, positioning also causes the attention of the users. Though the satellite positioning system, which is a mature application of localization technology, is able to provide users with accurate position service information. The demand of people is more focused on indoor area rather than outside, however, the satellite signal, which can’t penetrate buildings, has been unable to provide positioning service indoors.Smart mobile terminal is an essential part of the people’s communication and entertainment tool in today’s society, especially all kinds of mobile phones, tablets and wearable equipment emerge in endlessly, it has provided great convenience for people to satisfy people’s material and cultural life. Generally, these devices have Wi-Fi (Wireless Fidelity) communication module; in today’s society, These places, such as supermarket, companies, schools and so on, has the Wi-Fi network high coverage rate and Wi-Fi hot spots is able to be seen everywhere, so Wi-Fi indoor localization based on mobile terminal has the good research foundation and broad application prospect.The Wi-Fi indoor positioning technology and Algorithm based on mobile terminal has been deeply researched and discussed in this paper, and the main work of present dissertation is as follows:The background and significance, the research status, and some typical methods of positioning technology of indoor and outdoor is presented and discussed. Android is compared with iOS from the platform system architecture, the programming language and the Wi-Fi class library on mobile terminal; This thesis analysis the reason why Android terminal is selected, and the scheme that indoors positioning is based on mobile terminal is determined, And the nearest neighbor classification algorithm chosen in this article is studied.In the design of Wi-Fi indoor localization algorithm model, in view of the signal is affected by various factors in ranging phase of the localization algorithm based on transmission model, the inaccurate range causes a variety of problems that whether an unknown point is easy to locate or not, and to solve these problems two solutions that the trilateral localization algorithm and the estimation of liner regression model are discussed. It focuses on relevant details of the research work about location fingerprint algorithm from two aspects of offline gathering stage and online positioning stage:on offline acquisition phase, these problems, such as the problem of regional division on positioning, the establishment process of location measure, how to obtain the Wi-Fi signal strength information, how to generate the position fingerprint data and how to construct the fingerprint database, are discussed. On online positioning stage, this paper discusses the dissimilarity metrics, the difference on how to establish fingerprinting between matching algorithm based on determination and matching based on probability and the process of Case-based KNN algorithm. Then the two algorithm models are compared, and this paper chooses the localization fingerprint algorithm model based on mobile terminal, which is easy to implement as the scheme of the thesis.In view of position fingerprint algorithm modification, it puts forward three different methods, which contains the trimmed mean, the median and mean after screening, to settle RSSI original data. Based on the processed data, which makes the position fingerprint data more effective and reliable, so it improves the positioning accuracy. Three different improved strategy are presented on the nearest neighbor classification algorithm:based on weighted distance, based on the cosine similarity measure and the balance of the compromise between the two measurements method. Then in experimental phase, the thesisanalyzesconcretely the result and error of positioning of the traditional KNN algorithm and three kinds of improved strategyalgorithm, respectively in the experimental simulation. Finally, the positioning errors of the above four algorithm under different k value are compared, which verifies the effectiveness of the three kinds of improved strategy algorithm and the smoothness and stability ofthe joint measure algorithm... |
PDF Full Text Request