Resource allocation for signal detection with active sensors

— We consider the problem of determining the existence of known constant signals over a set of sites, given noisy measurements obtained by a team of active sensors that can switch between different sites. Since the quality of detection depends on the time that the sensors allocate at every site, maximizing the total detection probability relies on selecting the sites and possibly the order in which these should be monitored. When the switching time between sites is negligible, as in steerable camera networks, we show that optimizing the global detection performance for a team of sensors with uncorrelated measurement noise is a convex problem. On the other hand, for significant switching times, which can be due to path planning for mobile robots in surveillance missions, the detection problem can be approximated by an integer program, known as the orienteering problem. Due to its hardness, even small instances of this problem are difficult to solve. Focusing on the single sensor pro...

• Type: Conference
• Year: 2009
• Where: CDC
• Authors: Jerome Le Ny, Michael M. Zavlanos, George J. Pappas
• Keywords: CDC 2009, Control Systems, Detection, Sensor, Switching Times