Reducing the complexity order of position estimators with combined radiolocation measurements

As a growing attention is currently paid to low-cost energy-efficient radiolocation functionalities in a wide variety of wireless networks, one recurrent concern is to reduce the required processing time and the complexity of practical positioning estimators under deployment constraints (i.e. in terms of infrastructure, embedded computational capability and/or available measurements), while meeting simultaneously the best achievable accuracy. In this paper, we present a general method that reduces the complexity order of such position estimators with no noticeable performance degradation. The main idea consists in minimizing the number of observations that actually take part in the location computation. For this purpose, we propose to linearly combine available measurements into low-dimension observations that lead to the same Fisher Information Matrix (FIM) as that of more classical formulations (i.e. using measurements as direct observations). As an illustration, we show here how to apply this solution to two well-known position estimators, namely the linearized Least Squares (LS) estimator and the Extended Kalman Filter (EKF).