Localised path optimisation for mobile data collection in industrial sensor networks

Distributed sensor networks are used for asset monitoring within industrial environments. Energy efficient operation of such systems is an important characteristic. Mobile data collectors have been proposed as an approach for reducing energy requirements of data dissemination. Macro level optimisation techniques for selecting the overall best route for data collection exists, but they often oversee dynamic localised conditions which also need to be considered. A novel localised optimisation technique designed to be used in conjunction with existing global path strategies is presented here. The proposed technique considers dynamic radio conditions and optimises localised collection strategy accordingly. This is achieved using a naive Bayesian classifier trained to detect and react to different radio conditions observed through onboard diagnostics. The approach is evaluated using a series of real world experiments. Results indicate localised optimisation to provide measurable improvements, especially under complex and unpredictable radio conditions riddled with path loss, reflection, and noise. Incidentally these are characteristics which can be typical expected within industrial environments.