Deep Reinforcement Learning-Based Physical Layer Security Framework for Internet of Medical Things

The Internet of Medical Things (IoMT) is transforming modern healthcare information systems by connecting a diverse array of medical devices and sensors. However, significant security and privacy challenges arise when handling confidential medical data during transmission. This paper addresses these challenges by proposing a Physical Layer Security (PLS) framework integrated with Cell-Free Massive Multiple Input Multiple Output (CF-mMIMO) to enhance security in IoMT environments. The framework introduces a safe zone (SZ), a protected area surrounding legitimate healthcare devices to prevent access by eavesdroppers. This spatial segmentation enables precise beamforming within the SZ while amplifying artificial noise (AN) outside it, significantly boosting the secrecy rate. Additionally, the framework dynamically selects communication devices based on channel quality and orthogonality, optimizing network resources, reducing inter-user interference, and ensuring high-quality communication in densely deployed healthcare settings. Simulation results confirm that our approach adapts to and leverages the spatial dynamics of eavesdroppers, maintaining high secrecy rates even in scenarios with increased eavesdropper presence, thus keeping sensitive medical data secure and unreadable to unauthorized entities.