LAUSR

In Wireless Sensor Networks (WSNs), a dense deployment of sensor nodes produce data that contain intra-temporal and inter-spatial correlation. To reduce the intensity of correlation, we propose in-node data aggregation technique that eliminates redundancy in the sensed data in an energy-efficient manner. A novel data-driven approach is adopted to perform in-node data aggregation using an underlying cluster-based hierarchical network. Our proposed approach partially processes the data at each member node and forwards a fraction of the actual data, i.e., fused data, towards the cluster head. At each member node, the raw captured data is categorized into various classes, i.e., stratum. Each member node continuously senses the environment for temperature readings and compares them with the mean values of various strata. If the temperature reading is less than or greater than the mean value, it is compared with the existing Min/Max of that particular stratum. If in case, the new reading is less than/greater than the Min/Max of a particular stratum, it replaces these values, accordingly. Our proposed approach is computationally lightweight, energy-efficient and reduces the degree of correlation among the resource-constrained sensor nodes. As as a result, communication cost, packet collision and network congestion are reduced and the network lifetime is enhanced. The analytical results prove the validity and effectiveness of our proposed approach.