With the advances in the hardware technology and ever-decreasing cost of imaging circuitry, cameras have become ubiquitous in our lives. The large amounts of video data generated by multiple cameras necessitate automatic analysis of activities from video. With the introduction of smart cameras, it has now become viable to install many spatially-distributed cameras interconnected by wireless links. A smart camera is a stand-alone unit that not only captures images, but also includes a processor, memory and communication interface. This project takes a holistic view of the problems that need to be solved to build scalable, battery-powered wireless smart- camera networks (Wi-SCaNs). It focuses on providing a unifying solution to perform tasks distributively, and communicate in a P2P fashion over wireless links while making e±cient use of limited resources, such as energy and bandwidth. This research explicitly considers the challenges of wireless channels, and addresses the issues of scarce resources, P2P protocol design and channel usage priorities simultaneously. The expected outcomes of this project include light-weight vision algorithms that are portable to embedded smart cameras, characterization of the energy-bandwidth-delay tradeo®s in Wi-SCaNs, e±cient resource allocation strategies, and P2P wireless communication protocols that leverage the knowledge of the wireless channel conditions. The outcomes will have an important positive impact, because they will fundamentally advance the automatic activity analysis solutions by providing installation of any number of cameras anywhere, and will ¯nd wide-ranging applications including surveillance in military, commercial or public scenarios, elderly care and wildlife monitoring. 1
