Semantic IoT Transformation: Elevating Wireless Networking Performance through Innovative Communication Paradigms

Abstract

This paper addresses the privacy concerns inherent in semantic communication within the Internet of Things (IoT) and proposes a Secure Semantic Communication Framework (SSCF) to ascertain confidentiality and communication accuracy without compromising semantic integrity. The proposed framework uses the Advanced Encryption Standard (AES) for encryption to address privacy breaches in semantic communication. Additionally, it introduces a novel approach employing Deep Q-Networks (DQN) for adversarial training to maintain semantic communication accuracy in both unencrypted and encrypted modes. SSCF combines universality and confidentiality, ensuring secure and efficient semantic communication. Experimental evaluations showed that SSCF, with its adversarial encryption learning scheme, effectively ensures communication accuracy and privacy. Regardless of encryption status, SSCF significantly hinders attackers from restoring original semantic data from intercepted messages. The integration of heuristic algorithms enhances performance and security. The proposed framework is based on a shared database for training network modules. The originality of the proposed approach lies in the introduction of a DQN-based adversarial training technique to balance confidentiality and semantic communication accuracy, address key privacy concerns, and enhance the security and reliability of IoT communication systems.