Let Opportunistic Crowdsensors Work Together for Resource-efficient, Quality-aware Observations

Opportunistic crowdsensing empowers citizens carrying hand-held devices to sense physical phenomena of common interest at a large and fine-grained scale without requiring the citizens' active involvement. However, the resulting uncontrolled collection and upload of the massive amount of contributed raw data incur significant resource consumption, from the end device to the server, as well as challenge the quality of the collected observations. This paper tackles both challenges raised by opportunistic crowdsensing, that is, enabling the resource-efficient gathering of relevant observations. To achieve so, we introduce the BeTogether middleware fostering context-aware, collaborative crowdsensing at the edge so that co-located crowd-sensors operating in the same context, group together to share the workload in a cost- and quality-effective way. We evaluate the proposed solution using an implementation-driven evaluation that leverages a dataset embedding nearly one million entries contributed by 550 crowdsensors over a year. Results show that BeTogether increases the quality of the collected data while reducing the overall resource cost compared to the cloud-centric approach.