Improving data quality with an accumulated reputation model in participatory sensing systems

The ubiquity of mobile devices brings forth a sensing paradigm, participatory sensing, to collect and interpret sensory information from the environment. Participants join in multifarious sensing tasks and share their data. The sensing result can be obtained in light of shared data. It is not uncommon that some corrupted data is provided by participants, which makes sensing result unreliable accordingly. To address this nontrivial issue, we proposed the accumulated reputation model (ARM) to improve the accuracy of the sensing result. In ARM, participants' reputation will be computed and accumulated based on their sensing data. The sensing data from reputable participants make higher contributions to the sensing result. ARM performs well on calculating accurate sensing results, even in extreme scenarios, where there are many inexperienced or malicious participants.