SocialSense: Mobile crowd sensing-based physical distance monitoring model leveraging federated learning for pandemic

Abstract

As recommended by the World Health Organization, testing, isolation, and physical distancing are the keys to combat the pandemic due to COVID-19. However, physical distance monitoring and management is not straightforward, specifically in regions with high population density. Crowdsensing is one of the most feasible solutions where a large group of volunteers with mobile devices collectively share data and analysis on such data is carried out for extracting insights of common interest. This article proposes a novel mobile crowdsensing-based geospatial physical distance monitoring model capable of efficient pandemic monitoring and management. The work consists of two major contributions: analysis of human mobility information to find probable hot-spot regions and monitoring of the physical distance mandate. Another objective of this paper is to devise mobile crowdsourcing analytics model to find out the quality of the crowdsensing information and infer any implicit knowledge without affecting the quality of the output. Furthermore, we have also designed an Android application to implement the mobile crowdsensing system, named SocialSense, and provide effective pandemic management. The proposed model is supported by a theoretical analysis of latency calculation. We observe from the experimental results that the accuracy in hot-spot identification and physical distance monitoring are better in the case of the proposed model than the existing approaches. The trustworthiness of the crowdsourcing data is also improved in terms of accuracy than the existing approaches.