Network sensor management for tracking and localization

Abstract

Abstract-This paper addresses the problem of sensor management for a large network of agile sensors. Sensor management refers to the process of dynamically retasking agile sensors in response to an evolving environment. Sensors may be agile in a variety of ways, e.g., the ability to reposition, point an antenna, choose sensing mode, or waveform. The goal of sensor management in a large network is to choose actions for individual sensors dynamically so as to maximize overall network utility.Sensor management in the multiplatform setting is a challenging problem for several reasons. First, the state space required to characterize an environment is typically of very high dimension and poorly represented by a parametric form. Second, the network must simultaneously address a number of competing goals. Third, the number of potential taskings grows exponentially with the number of sensors. Finally, in low communication environments, decentralized methods are required.The approach we present addresses these challenges through a novel combination of particle filtering for nonparametric density estimation, information theory for comparing actions, and physicomimetics for computational tractability. The efficacy of the method is illustrated in a realistic surveillance application by simulation, where an unknown number of ground targets are to be detected and tracked by a network of mobile sensors.

Cognitive Radar Framework for Target Detection and Tracking

“…The expressions in (1)-(3), (15), and (21)-(28) provide the Bayes-Markov tracking recursions, the state estimate, and the predicted conditional Bayes risk expressions for a cognitive sensor/processor system whose objective is single target tracking. The formulation in this paper provides a generalization and formalism to the cognitive radar tracking formulations in [1]- [3], [7]- [20].…”

“…Some possibilities include the expected Kullback-Leibler and Renyi divergences [14], [15] and the entropy [18]. We take an alternative approach and use the same criterion that we used in the single target tracking problem, namely to minimize the trace of the PC-CRLB.…”

On the “near-universal proxy” argument for theoretical justification of information-driven sensor management

“…The "sandwich-inequality" (5) was used to justify the use of the Rényi divergence in a series of other works [13,9,14,15]. Since the "near-universal" proxy argument seems very strong given the properties of information-driven measures that we have been able to identify until so far, we will attempt to repeat the same derivations done in [8] and check the validity of these results.…”

Multi-target tracking via multiple cost-reference particle filtering

“…Another method has been applied to network sensor management and it includes the combination of particle filtering for nonparametric density estimation, information theoretic measures for comparing possible action sequences and artificial physics for cooperation between sensors [15]. Finally, some other efforts for resolving multimodality issues in particle filtering are based on non-parametric mixture models [28] and clustering of particles that are independently tracked [20].…”