A GPU-based implementation of PKF for INS/CNS integrated Navigation System

By using navigation system, carrier arrives at a predetermined position follow the route within the stipulated time. INS (Inertial Navigation System) has disadvantage of navigation error accumulate with time. CNS (Celestial Navigation System)´s has a low data update but the navigation error does not accumulate with time. Our goal is to develop a high-precision and high-performance navigation system by using integrated navigation method and Parallel Kalman Filter (PKF). As GPU (graphic processor units) have shown to provide much speedup in matrix computing which can be used on the filter algorithm of integrated navigation system, we will explore in this work for system performance optimization. Experiments results demonstrated that proposed solutions by using GPU accelerated Kalman filter has a scalable speedup as the measurement dimension increases than a pure CPU counterpart.