An increasing emphasis of the US Army lies in supporting efficient operation in contested environments, where access to air (e.g., aerial surveillance and pictures) and access to the spectrum (e.g., wireless communication) are highly reduced or denied. A data fusion system, called Athena, was recently developed to optimize information collection for decision making in this context . In this paper, we discuss the use of Athena to track realistic targets on city blocks using deployed security cameras. The work has two objectives: (i) investigate the efficacy of monitoring with sensors available in a city environment (e.g., security cameras, as opposed to, say, aerial imagery), and (ii) minimize communication needs among sensors to better handle bandwidth limitations. 284 vehicle trajectories were collected by inspecting USC campus security cameras to emulate realistic targets. Predictors were trained to anticipate future locations of targets based on past observations. We evaluate the efficacy of such predictors regarding tracking accuracy versus resource savings (e.g., size of footage collected for tracking).