Spatial Partitioning in Self-Organizing Smart Camera Systems

We propose a decentralized, self-organizing system architecture for wireless networked smart cameras (SCs) with pan, tilt, and zoom abilities. Each SC communicates with its neighbors and independently calculates the optimal position for its field of view. Thereby, SCs autonomously organize themselves and spatially partition the area they observe. This is achieved by a decentralized algorithm that makes way for self-organization in SC systems. The system quickly adapts to new situations caused by joining and failing nodes. Simulations with hundreds of cameras show that scalability and reliability are achieved. We analyze the performance of different camera densities and show that optimal surveillance coverage is achieved in a short time (20 s) while maintaining low communication traffic using a simulated 350-node outdoor SC system.