Query based nWSN data processing for spatial thermal mapping

This work discusses enhancing the neuroWSN architecture that has been developed for spatial thermal mapping with query system for time synchronization and relevant aggregation at the sensor level. The neural net training process at the cluster head requires synchronization of the incoming data from all the portable nodes. An approach for buffering and query management system is proposed here to fulfil this requirement. An QnDP (Query based nWSN Data Processing) algorithm is developed to validate the scenario. It can also handle the common aggregate functions along with the temperature and location data of each portable sensor node through the use of memory buffers. Every portable and infrastructural sensor node has its own cache/buffer to accommodate data storage for each parameter. The parameters of a query request are dynamically assigned to the response data when it propagates to any sensor node. The Query Based neuro Wireless Sensor Network (QB-nWSN) is designed using IBMs Mote Runner environment. Queries transmit and receive times are recorded for a sample of tests. The results and experimental setup revealed its viability to test further on a real time test bed.