Bus-Based Sensor Deployment for Intelligent Sensing Coverage and k-Hop Calibration

Abstract

Drive-by sensing is a promising concept that employs public transport as a mobile sensing platform to achieve high spatio-temporal coverage for urban sensing tasks. At the same time, the low-cost nature of mobile IoT sensors necessitates their more frequent calibration to ensure data accuracy and reliability. Manual or lab-based calibration of a large number of mobile sensors may no longer be feasible and thus new approaches for automatic calibration are needed. Most prior work on optimal mobile sensor deployment focuses on coverage aspect without considering the sensor calibration. In this study, we present a joint approach for optimising the placement of bus-based sensors for maximising the total unique sensing area and combining the optimal reference sensors geo-placement for maximising k-hop calibrate requirements on the selected routes. A metric-based system developed in our model uses geographical set operations which includes both spatial and temporal joins to quantify the contribution of each bus route and rank them accordingly. We formulate the coverage optimisation problem as a mixed integer linear program (MILP) solve it with a greedy algorithm, and demonstrate this method’s potential using real-world bus-transit data from Toronto, Canada and Manchester, UK. Our approach involves a metric-based system which quantifies each bus route unique coverage contribution for determining an optimal set of bus routes and bus stops for bus-based and reference sensor deployment, to minimise sensor network costs and maximise spatio-temporal coverage. The comparison with a random baseline algorithm indicates that our method outperforms in terms of deployment and coverage efficiency. Our results also include the potential of our weighted method in improving drive-by sensing for air quality monitoring by comparing it with a separate benchmark scheme with different criteria.