This paper investigates the problem of real-time estimation for one kind of linear time invariant systems which subject to limited communication capacity. The communication limitations include signal transmission dela...This paper investigates the problem of real-time estimation for one kind of linear time invariant systems which subject to limited communication capacity. The communication limitations include signal transmission delay, the out-of-sequence measurements and data packet dropout, which appear typically in a network environment. The kernel of filter design is equally to formularize the traditional Kalman filter as one linear weighted summation which is composed of the initial state estimate and all sequential sampled measurements. For it can adapt aforementioned information limitations, the linear weighted summation is then decomposed into two stages. One is a predict-estimator composed by all reached measurements, another is one compensator constructed by those time-delayed data. In the network environment, there are obvious differences between the new hybrid filter and those existing delayed Kalman filters. For example, the novel filter can be optimal in the sense of linear minimum mean square error as soon as all measurements available and has the lowest running time than these existing delayed filters. One simulation, including two cases, is utilized to illustrate the design procedures proposed in this paper.展开更多
基金Supported by the National Natural Science Foundation of China (No.60804064,60772006)
文摘This paper investigates the problem of real-time estimation for one kind of linear time invariant systems which subject to limited communication capacity. The communication limitations include signal transmission delay, the out-of-sequence measurements and data packet dropout, which appear typically in a network environment. The kernel of filter design is equally to formularize the traditional Kalman filter as one linear weighted summation which is composed of the initial state estimate and all sequential sampled measurements. For it can adapt aforementioned information limitations, the linear weighted summation is then decomposed into two stages. One is a predict-estimator composed by all reached measurements, another is one compensator constructed by those time-delayed data. In the network environment, there are obvious differences between the new hybrid filter and those existing delayed Kalman filters. For example, the novel filter can be optimal in the sense of linear minimum mean square error as soon as all measurements available and has the lowest running time than these existing delayed filters. One simulation, including two cases, is utilized to illustrate the design procedures proposed in this paper.
