Bayes-based Indoor Localization With A Wireless Sensor Network

Recent years have witness more and more modernization in China. A number of public constructions like skyscrapers, stadiums as well as numerous government facilities have been appearing, along with the need of indoor navigation for customers or movement tracking of janitors in a whole day for security reasons.Thus, there is an increasing demand for indoor localization techniques nowadays. Aimed at developing such a system with low cost but high accuracy, this thesis conducts research about how to make use of Naive Bayes classification method and the distinction of Received Signal Strength Index (RSSI) value at different positions. Wireless Sensor Network (WSN), a network consisting of hundreds of small devices able to sense the physical environment, compute and process simple tasks as well as communicate with each other, is the main infrastructure of the whole system to generate and deliver the RSSI dataset. Apparently, besides monitoring the wild field or working space in factories, indoor localization is another crucial application of WSN. Taking into account the market need and the cost of establishment, we first introduce the use of the RSSI dataset gathered from different positions to train a Naive Bayes Classifier and then use it to analyze the RSSI data from a certain position in order to estimate its location.which fulfills the function of indoor localization system of low cost and scalability.Especially we come up with a practical application of the theory mentioned above to develop a prototype system. This thesis conducted preliminary test-bed experiment and the results show that the proposed system can achieve a relatively high precision with low cost in establishment and energy consumption.