Piggyback CrowdSensing (PCS): Energy efficient crowdsourcing of mobile sensor data by exploiting smartphone app opportunities

Fueled by the widespread adoption of sensor-enabled smart-phones, mobile crowdsourcing is an area of rapid innova-tion. Many crowd-powered sensor systems are now part of our daily life { for example, providing highway congestion information. However, participation in these systems can easily expose users to a significant drain on already limited mobile battery resources. For instance, the energy burden of sampling certain sensors (such as WiFi or GPS) can quickly accumulate to levels users are unwilling to bear. Crowd system designers must minimize the negative energy side-effects of participation if they are to acquire and maintain large-scale user populations. To address this challenge, we propose Piggyback Crowd-Sensing (PCS), a system for collecting mobile sensor data from smartphones that lowers the energy overhead of user participation. Our approach is to collect sensor data by exploiting Smartphone App Opportunities { that is, those times when smartphone users place phone calls or use ap-plications. In these situations, the energy needed to sense is lowered because the phone need no longer be woken from an idle sleep state just to collect data. Similar savings are also possible when the phone either performs local sensor computation or uploads the data to the cloud. To eficiently use these sporadic opportunities, PCS builds a light weight, user specific prediction model of smartphone app usage. PCS uses this model to drive a decision engine that lets the smartphone locally decide which app opportunities to exploit based on expected energy/quality trade-offs. We evaluate PCS by analyzing a large-scale dataset (con- taining 1,320 smartphone users) and building an end-to-end crowdsourcing application that constructs an indoor WiFi localization database. Our findings show that PCS can ef- fectively collect large-scale mobile sensor datasets (e.g., accelerometer, GPS, audio, image) from users while using less energy (up to 90% depending on the scenario) compared to a representative collection of existing appoaches.