Distributed Space Situational Awareness (D-SSA) with a Satellite-assisted Collaborative Space Surveillance Network

Abstract

AbstractThis paper addresses the problem of tracking space-borne objects of interest (OoIs) with a satellite-assisted collaborative space surveillance network (SSN). It is formulated as a multiple moving-object tracking problem in a hybrid distributed wireless sensor network (WSN) made of both mobile and static nodes. The static nodes may be considered as ground-based sensors with fixed locations, while mobile sensor nodes can be satellites in space. At any given time, due to relative motion among nodes and OoI's, not all sensors are capable of observing a given target. The number of space OoI's (targets) of interest is several orders of magnitude larger than the number of nodes in the SSN, and each node can observe only a finite number of targets at a time. The challenge is to be able to schedule the nodes so that the best possible node-target assignments are made. A distributed approach is more desired in terms of scalability and robustness. However, for a distributed scheduling scheme to be efficient the distributed scheduling decisions need to be consistent across the network, which in turn requires the distributed local tracking estimates to be consistent across the network. As a result, the problem turns out to be a nonlinear consensus tracking problem on a time-varying graph with incomplete data and noisy communications links. We propose a distributed and collaborative framework in which each node in the network updates its tracking estimate of an object based on its own observations via an extended Kalman filter (EKF), followed by a consensus update stage in which nodes exchange their local estimates in order to arrive at a consensus on the distributed estimates leading to conflict-free node scheduling. Simulations of a two-dimensional hybrid SSN have been carried out in order to evaluate the performance of the proposed distributed tracking with consensus algorithm. The performance results show that the proposed algorithm indeed performs very well under conditions that can be expected in a realistic SSN, paving way for possible distributed scheduling.