Stochastic optimization for sensor allocation using AEGIS-FISST Conference Paper uri icon

abstract

  • In Space Situational Awareness (SSA) not only is it desired to maintain the two line elements of objects that have already been detected, but it is also desired to maintain the catalogue by updating it whenever new objects are detected for the first time. Hence, one of the main goals of SSA is to search for and detect new objects. The main challenge to achieving this goal is the fact that the sensor resources available in the SSA system are very limited compared to the very large size of the search space and number of objects (that number is in the hundreds of thousands of objects that are one centimeter and larger in size). This search and detection task needs to be performed using sensors that are often the same ones used for maintaining the tracks of detected objects. Thus, in this paper our goal is to develop a multi-object informationbased objective function for sensor allocation. However, conventional informationbased approaches to the sensor allocation problem (see, for example, Ref. [1]) are mostly dedicated to the problem of sensor allocation for multi-object tracking (without detection). Thus, we develop a Finite Set Statistical (FISST) approach to sensor allocation for joint search, detection and tracking. We demonstrate the procedure on a simple two-object SSA problem. The resulting information objective function will then be used to optimally task the sensor. The resulting optimization problem is computationally intractable. Therefore, we use a stochastic optimization technique that was developed by the authors to solve for the optimal sensor tasking.

author list (cited authors)

  • Hussein, I. I., Sunberg, Z., Chakravorty, S., Jah, M. K., & Erwin, R. S.

publication date

  • January 2014