随机系统中能容忍连续丢包和测量时延的卡尔曼滤波

通过转换原线性系统到能容忍连续丢包和测量时延的随机参数系统,推导了一个最优线性滤波器.给出一个仿真例子,比较已存在的结果,仿真结果表明被提出的线性滤波器有优越的性能.然而,该滤波器不能应用于非线性系统.从应用角度,为非线性系统提出了一个增强型的滤波器.而且,该增强型的滤波器能成功地应用于不可靠的无线传感器网络场景来跟踪移动目标.这些滤波器只依靠测量值的达到概率,而不需要知道某一时刻测量是否接收.仿真说明了被提出的增强型滤波器不仅能改善实时目标跟踪的鲁棒性,而且比标准的扩展卡尔曼滤波器能够提供更精确的估计.

Kalman filtering for stochastic systems with consecutive packet losses and measurement time delays

An optimal linear filter is derived through transferring the original linear systems to stochastic parameter systems with consecutive packet losses and time delays. A numerical simulation example is performed with results showing that this linear filter has superior performance to other existing approaches. However, the proposed filter cannot be applied to nonlinear systems. From the practical perspective, an enhanced filter is proposed and is extended to nonlinear systems. This enhanced filter has been applied successfully to an unreliable wireless sensor network (WSNs) scenario to track a moving target. The proposed filters depend only on the measurement arrival probability at all time but do not require knowing whether a measurement is received at a specific time instant. Simulations show that the proposed enhanced filter not only improves the robustness for real-time target tracking in WSNs, but also provides more accurate estimations than the standard extended Kalman filter.