Optimal Sensor Placement for Large Scale Systems Using Boosted Clustering

Applications such as smart cities, smart weather monitoring etc. involve installing a large number of sensors. Installing these sensors and maintaining them is a cumbersome exercise and quite often involves huge cost. As a solution, one can install lesser number of sensors and monitor the entire area by interpolating the missing values (locations which are not measured). The approximation error obtained depends on two things (i) number of sensors installed (ii) placement of these limited number of sensors. The proposed work focuses on the second aspect i.e. optimal placing of sensors assuming the number of sensors available to be placed are fixed. Traditional methods like [1, 2, 3] estimate the optimal locations by posing them as an optimization problem solved using mathematical or heuristic approach. However, for large-scale systems, which deal with thousands of sensors, solution strategies are inefficient owing to their computational complexity.