RSS Localization Using Multistep Linearization in the Presence of Unknown Path Loss Exponent
Xiaojun Mei, Yanzhen Chen, Xiaofeng Xu, Huafeng Wu
Abstract
Location awareness is essential to numerous wireless sensor network applications. However, it is challenging to have an accurate location estimation, especially for the received-signal-strength-based scheme in the presence of an unknown path loss exponent (PLE). To this end, this letter first develops the maximum likelihood estimator (MLE) with an unknown PLE. Owing to the significant computational complexity and highly nonconvex feature in the MLE, this letter further converts the original problem into a generalized trust regional subproblem by employing the first-order Taylor expansion to different parameters. A multistep linearization (MSL) method is then studied for the scheme, wherein the target location and the PLE are considered variables without prior knowledge. Subsequently, a bisection method integrated with a refined step is presented to figure out the estimate. Moreover, this letter conducts the Cramér–Rao lower bound with unknown PLE to evaluate the proposed method. Simulation results show that the proposed MSL achieves better estimate accuracy than other state-of-the-art methods in different scenarios.