Research On Wireless Localization Algorithm Based On Crawdad

The rapid developments of wireless network changes people’s life and provides us with a broader create space. With the same time, the process of lighter mobile devices and higher data transfer rate is constantly moving forward, so the wireless network has gotten more and more attention. With the development of IEEE802.11protocol standards, the wireless local area network based on these standards has become the focus of everyone’s attention. Because of its characteristics of high data transmission rate, WLAN has forced many attractive technologies and services, and wireless localization is one of them. Nowadays, in many large public places, people can feel free to enjoy their time that is brought by localization services through mobile cellular or wireless hotspot, and sharing their locations to their friends, navigation, searching surrounding shops and other functions has gradually become an integral part of people’s life.In this paper, the research is about the Wi-Fi fingerprint localization algorithm which is based on RSSI. The basic principle of such an algorithm is to create a fingerprint database to store the fingerprint data that consists of RSSI and the position coordinate. Then, when an terminal transmits a positioning request, the algorithm matches the RSSI that is collected by the terminal with the fingerprint data which is in the fingerprint database. And then, estimate the position of the terminal. This paper mainly studies the traditional classification algorithms for problems and chooses Naive Bayes (NB) which has a better performance in the positioning estimates as the basis for improvement. And we propose an improved Bayesian algorithm. And for the lack of traditional simulation, we use hotspots signal strength data stored at CRAWDAD platform and collected in Seattle to analyze and get the corresponding parameters to optimize the traditional simulation, so as to make "simulation" closer to "true". Finally, we complete an improved algorithm implementation and testing through the experiment platform. This experimental platform is based on MySQL database and Android mobile client, and gets a big enough fingerprint information database through the server and client information collection. Then, the algorithm is validated by the data obtained from experiments. After analyzing the data, we find that the improved algorithm can avoid the zero probability event caused by some accidental circumstances during the probability calculations, and the algorithm’s performance on the positioning accuracy and latency has a distinct advantage.