Exploiting spatial correlation towards an energy efficient Clustered AGgregation technique (CAG)

In Wireless Sensor Networks (WSN), monitoring applications use in-network aggregation to minimize energy overhead by reducing the number of transmissions between the nodes. We note that nearby sensor nodes monitoring an environmental feature (e.g., temperature or brightness) typically register similar values. In this paper, we propose Clustered AGgregation (CAG), which is a mechanism that reduces the number of transmissions and provides approximate results to aggregate queries by utilizing the spatial correlation of sensor data. The result is guaranteed to he within a user-provided error-tolerance threshold. While a query is disseminated to the network, CAG forms clusters of nodes sensing similar values. Subsequently, only one value per cluster is transmitted up the aggregation tree. We use mathematical models and simulations with synthetic and empirical data to evaluate the efficiency-correctness tradeoff of CAG. Our simulation shows that with highly correlated sensor reading and 10% error threshold, CAG can save the communication overhead by as much as 70.9% over TAG while incurring a modest 1.7% error in result.