Nonmyopic Sensor Scheduling and its Efficient Implementation for Target Tracking Applications

Abstract

We propose two nonmyopic sensor scheduling algorithms for target tracking applications. We consider a scenario where a bearingonly sensor is constrained to move in a finite number of directions to track a target in a two-dimensional plane. Both algorithms provide the best sensor sequence by minimizing a predicted expected scheduler cost over a finite time-horizon. The first algorithm approximately computes the scheduler costs based on the predicted covariance matrix of the tracker error. The second algorithm uses the unscented transform in conjunction with a particle filter to approximate covariance-based costs or information-theoretic costs. We also propose the use of two branch-and-bound-based optimal pruning algorithms for efficient implementation of the scheduling algorithms. We design the first pruning algorithm by combining branch-and-bound with a breadth-first search and a greedy-search; the second pruning algorithm combines branch-and-bound with a uniform-cost search. Simulation results demonstrate the advantage of nonmyopic scheduling over myopic scheduling and the significant savings in computational and memory resources when using the pruning algorithms.

Sensor Resource Allocation for Tracking Using Outer Approximation

“…All time indexes correspond to unless specified otherwise. It is shown in [10] that is a convex function of , where is the relaxation of over the unit interval [0,1]. An outline of the proof in [10] is as follows: 1) we first use the matrix convex function property in [11] to prove that , , and , where the inequality is in a positive semidefinite sense; 2) we obtain since the aforementioned matrix inequality is preserved in the diagonal elements of 's and is a linear function of these diagonal elements.…”

“…We predict the tracking error one-step ahead using an information formulation of the extended Kalman filter (EKF) in which the sensor contribution is expressed as an information matrix [8], [9]. This matrix is obtained by linearizing the measurement model about the one-step predicted target state.…”

Adaptive sampling with sensor selection for target tracking in wireless sensor networks

“…Note that our approach is also different from the nonmyopic sensor scheduling problem, where in the non-myopic sensor scheduling, the problem is to select/schedule sensors over the future certain time steps of tracking [8], [9]. Here, we basically consider a myopic sensor selection scheme where at a given time, t, we first determine when to sample the target t + Δ and then decide which sensors to select at t + Δ. Simulation results demonstrate that adaptive sampling provides similar estimation performance as compared to the fixed and frequent sampling case.…”

On the Convergence of the Modified Riccati Equation

“…It plays an important role in target tracking [1][2][3][4][5][6][7][8][9][10]. Theoretical properties of the modified Riccati equation have been derived in [2,3].…”