A novel semantic-based multi-packet parallel transmission scheme for networked control systems

Abstract

This paper proposes a novel semantic-based multi-packet parallel transmission (MPPT) scheme for networked control systems (NCSs), aimed at addressing the challenge of limited network bandwidth in accommodating the growing data traffic demands in MPPT. Following the principle of ‘analyzing before transmission’, the core of the proposed semantic-based MPPT is ‘semantic extraction $\to$ semantic transmission $\to$ semantic recovery’. Firstly, semantic information of raw data is extracted in the form of natural language descriptions. Considering the ‘cultural clash’ between these descriptions and bitstream-based system components, fuzzy mathematical theories and methods are introduced to facilitate the mutual conversion between natural and computer languages. With the establishment of data semantic feature library, natural language descriptions are further converted into digitized semantic representations for network transmission. Subsequently, the semantic recovery module receives these digitized semantic representations and generates reconstructed data based on shared knowledge, which are then used as inputs for the controller. Furthermore, the concept of semantic errors is introduced to systematically model the inherent discrepancies between raw and reconstructed data. Through the skillful controller design, the asymptotic stability of the studied system is guaranteed in the presence of semantic errors while satisfying the $H_{\infty }$ performance constraint. Additionally, a parameter optimization algorithm is provided to configure semantically-related parameters, thereby optimizing semantic reconstruction effect. Finally, the effectiveness of the proposed method is validated in a six-area power system.