Fusion-Based Sensor Selection for Optimal State Estimation and Minimum Cost (Intelligent Optimization Approach)

This paper proposes a heuristic method for the sensor selection problem that uses a state vector fusion approach as a data fusion method. We explain the heuristic to estimate a stationary target position. Given a first sensor with specified accuracy and by using genetic algorithm, the heuristic selects second sensor such that the fusion of two sensor measurements would yield an optimal estimation in a target localization scenario. Optimality in our method means that a trade-off between estimation error and cost of sensory system should be created. The heuristic also investigates the importance of proportion between the range and bearing measurement accuracy of selected sensor. Monte Carlo Simulation results for a target position estimation scenario showed that the error in heuristic is less than the estimate error where sensors are used alone for estimation, while considering the trade-off between cost and accuracy. Index Terms — data fusion, sensor selection, multi-objective optimization, genetic algorithm.