Semi-SemNet: A Evolved Servitization Network Architecture from 5G to Semantic Communication

Recently, a new communication paradigm, semantic communication, has garnered significant attention due to its superior information transmission efficiency compared to conventional communication. However, current research on semantic communication primarily focuses on the design of end-to-end communication systems from the perspective of goal orientation or information reconstruction, without considering integration with the evolving technological landscape of the 5th generation mobile communication (5G) networks. To address this gap, in this paper, we propose an evolved servitization network architecture, called Semi-SemNet, which provides goal-level semantic service capability for the proposed converged goal-oriented semantic communication. To enable dynamic semantic-aware load balancing, the contexts of network functions (NFs) are decoupled from themselves and managed by a common NF, rendering each NF stateless. Additionally, we present an overview of enhanced NFs and system procedures for semantic communications. To meet different target nodes, such as application servers and user equipment (UE), we introduce a semantic translation mechanism to eliminate performance degradation when there are multiple independent wireless links between the source node and the target node. Finally, we use fire video remote monitoring as a case study. Experimental results show that Semi-SemNet achieves an average 26.67% PSNR and 27.21% MS-SSIM performance improvement in information reconstruction, and saves 40.51% bandwidth resources in goal execution. Moreover, with the stateless mechanism and semantic-aware load balancing decisions, it can reduce semantic data transmission delay by an average of 60.59% compared to the benchmark methods. The all code and original data of this paper are publicly available at https://github.com/hwniu/sbi_semantic_network.git.