Density-aware compressive crowdsensing.

Crowdsensing systems collect large-scale sensor data from mobile devices to provide a wide-area view of phenomena including traffic, noise and air pollution. Because such data often exhibits sparse structure, it is natural to apply compressive sensing (CS) for data sampling and recovery. However in practice, crowd participants are often distributed highly unevenly across the sensing area, and thus the numbers of observations collected over different areas may vary wildly - an issue we call density disparity. Density disparity leads to inaccuracy in low density areas, and potentially undermines the recovery performance if conventional compressive sensing is applied directly, which equally treats data from areas of different density. To address this challenge, we propose a probabilistic accuracy estimator, based on which we devise two recovery algorithms: Threshold Recovery (TR) and Weighted Recovery (WR). As general-purpose recovery algorithms, TR and WR improve the performance of CS in the scenarios with density disparity, and also provide better guarantees in terms of ℓ2-norm accuracy compared with conventional CS recovery algorithms. We also conduct extensive experiments based on synthetic and real-life datasets. Our results show that TR/WR typically reduce ℓ2-norm error by more than 60% compared to state-of-the-art baselines.