Coverage problem in camera-based sensor networks using the CUDA platform

Abstract

Closed-circuit televisions serve as prevention against crime, and many studies for closed-circuit television deployment have been conducted. The closed-circuit television deployment in downtown is similar to solving coverage problem in wireless camera-based sensor networks. The difference between the two problems is various environmental factors such as buildings, roads, camera capability, and movements of pedestrians. We use a genetic algorithm to increase the efficiency of closed-circuit television deployment in two-dimensional topography. In addition, a parallel experiment using general-purpose computing on graphics processing units is added to improve computing speed, which is a disadvantage in genetic algorithms. The target region is 500 m × 500 m and consists of 50 × 50 grids. The fitness of the evaluation, which refers to a detection rate, is calculated from the corresponding cell when a pedestrian moves to each cell depending on whether the pedestrian is detected. The proposed experiment was superior to the random deployment experiment by approximately 37.5%. There was no significant difference in the detection rate between the CPU experiment and a NVIDIA GeForce GTX 970 experiment in the 95% confidence interval. The efficiency of a CUDA kernel function using the NVIDIA GeForce GTX 970 graphic card was analyzed.