并行卡尔曼滤波器的多信道序列规划算法

摘要

针对基于卡尔曼滤波器的并行序列规划问题,本文提出了一种基于多信道的改进算法,在规划各进程所对应传感器的测量信息传输序列,以使相应滤波器的状态估计误差协方差满足给定指标的同时,尽可能减少通讯信道占用.文中论证了基于夹挤方法的周期算法,以提高计算进程传输周期的效率;改进了基于缓存的离线序列规划算法,以适应多信道传输的情景;设计了在线序列调整算法,利用空闲信道进行传输失败后的数据复发,快速调整序列以适应指标变化的需求.算法的运行效果将通过仿真实验进行验证与分析.%In this paper, an improved multi-channel algorithm is designed to schedule parallel processes of Kalman filters. The goal is to schedule the transmission of sensor measurements for all processes to meet individual estimation error requirements with the least communication channels occupied. Our algorithm includes the following methods:A fast cycle estimation method based on the squeeze method, which accelerates the search for transmission cycles significantly;An off-line schedule method based on the Buffer Scheme, which is suitable for scheduling transmissions with multiple communication channels;An on-line reschedule method, which utilizes idle channels against transmission losses and ad-justs the transmission sequence quickly in case of process requirement changes. The effect of the algorithm is verified by simulations.