Qute: quality-of-monitoring aware sensing and routing strategy in wireless sensor networks

Wireless Sensor Networks (WSNs) are widely used to monitor the physical environment. In a highly redundant sensor network, sensor readings from nearby sensors often have high similarity. In this work, we are interested in how to decide an appropriate sensing rate for each sensor node, in order to maximize the overall Quality-of-Monitoring (QoM), while ensuring that all readings can be transmitted to the sink. Note that a feasible sensing rate allocation should satisfy both energy constraint on each sensor node and flow conservation through the network. In order to capture the statistical correlations among sensor readings, we first introduce the concept of correlation graph. The correlation graph is further decomposed into several correlation components, and sensor readings from the same correlation component are highly correlated. For each correlation component, we defined a general utility function to estimate the QoM. The utility function of each correlation component is a non-decreasing submodular function of the total sensing rates allocated to that correlation component. Then we formulate the QoM-aware sensing rate allocation problem as a utility maximization problem under limited power supply on each node. To tackle this problem, we adopted an efficient algorithm, called Qute, by jointly considering both the energy constraint on each node and flow conservation through the network. Under some settings, we analytically show that Qute can find the optimal QoM-aware sensing rate allocation which achieves the maximum total utility. We conducted extensive testbed verifications of our schemes, and experimental results validate our theoretical results.