Wireless multimedia sensor networks are networks of multimedia sensors (e.g., web cameras) that are interconnected and collaboratively work on monitoring an area of interest. They have wide range of applications in areas such as monitoring, surveillance, tracking, immersive technologies, and health care. In this proposal we identify and pursue the problems that arise on the system side of these networks. First, we consider target coverage: how can a set of sensors be most effectively deployed and controlled to monitor and track a set of targets of interest. The basic problem can be expressed using a convex optimization formulation. We propose extensions to this formulation to capture additional considerations such as the presence of obstacles and vulnerable borders. Second, we propose application sensitive multimedia networking protocol that are able to intelligently transfer the desired multimedia streams while maintaining Quality of Service and providing in-network synchronization of correlated streams and events. Finally, we consider the issue of collaborative storage and indexing of the sensor data. Often the analysis of the camera data requires expensive human intervention (the video has to be viewed to determine what happened). Thus, it is critical for the system to identify interesting periods of operation using lightweight algorithms to reduce the amount of human intervention required.