The effectiveness of centralised management for reducing wasted effort in participatory sensing
Mattias Linnap, Andrew C. Rice · 2014
Participatory sensing projects typically rely on time-based polling or volunteer-initiated data collection. We have investigated the potential benefits of managed sensing, in which requests for information are targeted at particular volunteers as they visit locations of interest. We performed a study comprising of 24 volunteers who were asked to report sightings of bees whilst visiting gardens over a 400m by 400m area. We found that compared to a baseline periodic polling strategy, the centralised management approach achieved 8.6% larger sampling coverage of the visited area with 4 times less wasted participant effort. We investigated the parameter sensitivity of the results in simulation. We show that the results are statistically significant, and that participant response time to questions is the most influential parameter. Centralised management is more effective than simpler approaches if participants respond within 3 minutes, and equivalent for longer response times. Both the experimental and simulated results are informative for parameter and management choices for future participatory sensing studies.