Asynchronous Track-to-Track Fusion by Direct Estimation of Time of Sample in Sensor Networks

Asynchronous data fusion is inevitable in track-to-track fusion for tracking high-speed targets. For low-speed targets, e.g., the movement of clouds, synchronization is insignificant and, depending on the application, may be disregarded. Real-time asynchronous fusion is a demanding task in sensor networks when the sensors are not synchronous in sampling-rate or in sampling-phase. In the method proposed in this paper, an estimator in the fusion center estimates the actual time of the sample with respect to the time-reference of the fusion center upon receiving the data from a sensor. Then, the computer of the fusion center uses predictions to transfer all the received data to the data corresponding to the start of the next fusion period. This process synchronizes the data, which is necessary for real-time uncorrelated track-to-track fusion. Finally, the pseudo-synchronized data of all the sensors are fused with an element-wise linear minimum variance unbiased estimator algorithm before the start of the next fusion period. Simulation and comparison with some benchmark algorithms are demonstrated to verify the effectiveness of the proposed algorithm.