Secure and Scalable Home Monitoring A Federated Learning Architecture

DOI: https://doie.org/10.0307/Jbse.2025203228

Avijit Bose, Dipannita Ghosh Sneha, Ahana Mukherjee, Mrinal Mondal, Ipsita Dutta, Satyajit Chakrabarti

Keywords:

Federated Learning, Home Surveillance, Deep Learning, Pose Estimation.

Abstract:

Video Surveillance is a well-established monitoring technology which has applications in a myriad of sensitive and complex domains. In this following context, we are presenting a novel home surveillance system on pose estimation, fog computing, federated learning and image processing for real time surveillance of small children and old people keeping the sensitivity of personal information in mind. The said system would help in preventing child abuse and monitoring of old age people when no-one is present in the house. It is implemented using areal time analysis algorithm which uses a unique kind of data- preprocessing. We also propose a unique two-level encryption method which would prevent personal data from being divulged and helps in maintaining the privacy of the concerned individual, and as such our solution has an impact on the present society which demands both real-time surveillance and privacy-security. We also demonstrate how our solution can be adapted to the next generation wireless architecture i.e. 5G and 6G when fully deployed.
 

PDF