Spatio-temporal Pattern Discovery in Sensor Data: A Multi-valued Decision System Approach

Discovering novel and interesting spatio-temporal patterns in the collection of sensor data is a significant challenge in many scientific domains. Sensors are being employed in many practical applications that require complex data sampling, analysis and integration. Many practical applications require deployment of a large number of unattended high-density sensor nodes in a sensor field in changing environment. For example, real world applications like monitoring water quality in a watershed or in a treatment plant, hazardous chemical spill, bio-terrorism, and traffic fleet management require analysis of sensor data at upstream /downstream or at different street / pipeline intersections at different intervals of time. Such events are, in principle, preventable if sensor systems are capable of discovering characteristic spatio-temporal patterns or templates long time before exposure. Given that the spatio-temporal patterns represent the physical processes that govern the phenomena being measured, finding spatio-temporal patterns from such dataset can potentially uncover a great deal of information about the underlying processes By identifying generic patterns from data streams without human supervision, it is possible that one can extract the most relevant information with high fidelity and remove the irrelevant patterns and consequently help develop control mechanisms. The proposed framework also has a potential to reduce the data communications. Besides, due to the use of pattern IDs during data transmission, our method is inherently secure and there is no need to distribute encryption or decryption key between the cluster-heads and the sinks. Original paper submitted in May 2014. Revised paper was resubmitted for review in November 2014.