Measurement Data Delay Under Incomplete Measurements Filtering Research

Nowadays, more and more target tracking systems begin to adopt multi-sensor network to detect and track the targets. In the network, especially the wireless one, delay or dropouts of the data packets exist widely, which in turn make the whole filtering system face the challenge of partial observations. Under this circumstance, the filter should be modified to guarantee the convergence of the system and inhibit the disturbance of measurements uncertainty. Previous researches mainly discuss one typical phenomenon of the measurement uncertainty-missing measurements and its influence on the filtering system. However, this paper concerns another typical phenomenon of the measurement uncertainty-delayed measurements and differentiates it with the missing ones. Also, the modification of the filter to adapt the delayed measurements is proposed.First, this paper has constructed different mathematic models for missing and delayed measurements in the filtering system under different conditions of measurements uncertainty, such as different number of measurement channels, whether knowing the detection probability or not. This is the precondition for designing the suitable filters.Then, the paper discusses the filtering theory and method when there exist randomly distributed missing measurements. Two different filtering methods are proposed based on whether konwing the detection probability or not. The numerical examples in the end of this part test those methods.Next, the filtering theory and method considering the existence of measurement delay are discussed. Those methods are different due to the fact that the target tracking system has single measurement channel or multiple ones. In the case of single measurement channel, the innovation fix method is proposed to utilize the delayed measurements, while the innovation re-organization method is applied when the system owns multiple measurement channels. Numerical examples and the relating analysis also follow.Finally, considering both the missing and delayed measurements, the paper combines the filtering theories and methods discussed above and solves the filter design problems. The filtering equations are deduced under the system with multiple measurement channels. In the end, the numerical experiments have proved the effectiveness of the method.