Enhancing IoT-Based Intelligent Video Surveillance through Multi-Sensor Fusion and Deep Reinforcement Learning

Aymen Hussein¹, S. Ahmed², Shorook K. Abed³, Noor Thamer⁴

¹Department of Medical instruments engineering techniques, Alfarahidi University, Baghdad, Iraq

²Al-Turath University College, Baghdad, 10021, Iraq,

³Department of Computer Techniques Engineering, Mazaya University College, Thi Qar, Iraq

⁴ Accounting Department, Al-Mustaqbal University College , 51001 Hillah, Babylon , Iraq.

Emails: Aymen.hussein@alfarahidiuc.edu.iq; saif.saad@turath.edu.iq; shurooqkamel7@gmail.com; noorthamer2020@mustaqbal-college.edu.iq

Abstract

Currenlty, wireless communication that is successful in the Internet of Things (IoT) must be long-lasting and self-sustaining. The integration of machine learning (ML) techniques, including deep learning (DL), has enabled IoT networks to become highly effective and self-sufficient. DL models, such as enhanced DRL (EDRL), have been developed for intelligent video surveillance (IVS) applications. Combining multiple models and optimizing fusion scores can improve fusion system design and decision-making processes. These intelligent systems for information fusion have a wide range of potential applications, including in robotics and cloud environments. Fuzzy approaches and optimization algorithms can be used to improve data fusion in multimedia applications and e-systems. The camera sensor is developing algorithms for mobile edge computing (MEC) that use action-value techniques to instruct system actions through collaborative decision-making optimization. Combining IoT and deep learning technologies to improve the overall performance of apps is a difficult task. With this strategy, designers can increase security, performance, and accuracy by more than 97.24 %, as per research observations.

Keywords: Machine Learning; Internet of Things; DRL; Intelligent Video Surveillance; Mobile Edge Computing; Fusion System Design.