Sensor network for object location determination using machine learning methods

Abstract

The use of closed-circuit television (CCTV) has been commonplace thus far in protecting personnel and assets of key installations, commercial buildings and residential houses. Situations today can make use of sensor networks to locate objects, such as people or furniture within a room. However, CCTV can be overly intrusive for less sensitive applications such as surveillance of elderly people who are left alone in their homes. Sensor networks can then be deployed to detect the location of individuals within the house. Some additional functions can be programmed to report the position (lying down, seated, etc) they are in so that if anything untoward happens, a timely response can be initiated. Thus, a sensor network surveillance system can be used for functions where a regular CCTV will be deemed too invasive of privacy. This project seeks to simulate a sensor network through the use of radial basis function neural networks in MATLAB. A number of scenarios will be simulated and a comparison of the errors incurred under different configurations will be made.