Abstract:

When providing location-based dating and other location-based services, social networks will confuse the displayed user distance text to protect the user’s location privacy.In order to verify whether the current location confusion mechanism adopted by social networks can effectively protect user’s accurate location, a social network user geolocating method based on weighted least squares was proposed.The method constructed real-world tests to collect a large number of confused distance texts in location dating service, and identified the real distance boundary of reported distance combined with real distance data.Then, based on the discrimination of the quadrant of the coordinate system where the target user was located, the position of probes was optimally deployed, and multiple preliminary position estimations of the target user were obtained by using the trilateration model.The weights of the longest and the shortest distance of the target user were determined by the estimated position, and the related objective function was constructed.The optimal solution of the objective function was calculated based on the weighted least squares, which was the final geolocating result of the target user.The proposed method inferred the location of social network user based on the distance boundary constraint, which avoided the frequent query of location services and ensured the geolocating efficiency.The actual geolocating results based on 500 WeChat users showed that, the proposed method can accurately geolocate the “eople nearby” users of WeChat.Compared with the existing typical geolocating methods based on space partition and heuristic number theory, the proposed method have better performance in geolocating accuracy and efficiency.The average geolocating error is reduced by more than 10%, and the number of location service quarries in the geolocating process is reduced by more than 50%.

Key words: social network, location privacy, user positioning, weighted least squares