OPTIMIZING URBAN MONITORING BETWEEN STATIONARY, OPPORTUNISTIC VEHICULAR, AND HYBRID SENSING

Abstract. Urban monitoring based on wireless sensor networks is a recent paradigm that exploits a large number of low-cost sensors deployed in certain places or/and on mobile devices to collect data ubiquitously at a large scale. In this study, we explore and compare the coverage of stationary and opportunistic vehicular sensing methods with respect to the requirements of a task at hand. We distinguish spatial granularity, temporal granularity, and budget constraints. First we compare the spatio-temporal coverage of stationary sensing and opportunistic vehicular sensing for various tasks, which demonstrates that these two sensing methods are suitable for different tasks. Then we propose a hybrid sensing deployment framework integrating a genetic algorithm to achieve the maximum spatio-temporal coverage for specific tasks. Experiments with a real-world vehicle trajectory dataset demonstrate that the proposed hybrid sensing framework achieves the maximum spatio-temporal coverage in various tasks. Our results provide fundamental guidelines on network planning for urban monitoring applications.