Research On Localization Techniques And Location Verification Systems In Wireless Networks

As location-based applications and services become ubiquitous in emerging wireless networks, the stability, reliability and security of a localization system are of growing importance. However, the traditional localization error used to evaluate a localization system cannot be utilized to monitor the system in practice, since it is dependent on the knowledge of the user’s true position. In this thesis, the estimated region, which is the area where a target node resides with utmost probability specified by the corresponding localization algorithm, is proposed to resolve the above problems. Firstly, the relationship between the estimated region and the localization error is derived in theory and then verified through Monte Carlo simulations. The localization error is proportional to the square root of the estimated region’s area. Then, the thesis presented a localization algorithm based on the estimated region and analyzes its performance by comparing it with the original localization algorithm.The traditional approach to address reliability and security is to apply cryptographic authentication. However, the cryptographic authentication is not able to guarantee the security due to the openness of the localization system, and cannot prevent inside attacks. On the other hand, the location verification system (LVS) utilizes the physical layer information, making it much harder to forge, and can be used to differentiate users. Since the output is binary, the threshold of a location verification system determines its performance. This thesis proposes a threshold based on the squared Mahalanobis distance. After presenting a threat model, this thesis derives the theoretical false positive and detection rates, and optimizes the threshold by minimizing the sum of false positive and false negative rates. Finally, the detailed numerical simulations verify the legitimacy of our analytical framework proposed in this thesis.