Vector Matrix Mapping for Variable-Length Semantic Communication
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Semantic communication systems in the related art do not involve classical data encoding and require transmitted sentence lengths to be identical, limiting their integration with classical communication systems and restricting data compression potential.
Innovation Solution
A vector matrix determination method that encodes a message into a codeword matrix, which includes regular and irregular codewords, and transforms it through a group of fully-connected networks to achieve codewords with the same lengths, enabling integration with classical encoding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If semantic communication systems use fixed-length One-Hot vectors for encoding, then the system can process data with consistent dimensions, but the system cannot handle information with variable length and cannot integrate with classical communication systems
Solution Approach 1:
The encoding system is segmented into two independent modules: a classical encoding module that handles variable-length data compression and a semantic encoding module that processes fixed-length semantic vectors. This segmentation allows each module to operate independently with its own optimal structure, resolving the contradiction between handling variable length information and maintaining fixed-length processing requirements.
Solution Approach 2:
A mapping module is introduced as an intermediary between the classical encoding module and the semantic encoding module. This mapping module transforms the variable-length codeword output from classical encoding into a fixed-length format that can be processed by the semantic encoding module, enabling integration of the two systems while preserving their respective advantages.
2Productivity
If semantic communication systems require identical transmitted sentence lengths, then the encoding process is simplified, but the application scenarios are greatly limited and integration with classical communication is not conducive
Solution Approach 1:
The system separates data compression functionality (handled by classical encoding module with variable-length codewords) from semantic processing functionality (handled by semantic encoding module with fixed-length vectors). This allows the system to achieve high compression ratios for diverse application scenarios while maintaining consistent internal processing dimensions.
Solution Approach 2:
The mapping module dynamically adjusts the representation parameters of codewords based on the output from classical encoding. By changing the parameter format from variable-length to fixed-length while preserving the information content, the system achieves both high compression potential and broad application scenario compatibility.
3Adaptability or versatility
If classical encoding and semantic encoding are integrated, then the system can process variable length information, but the system complexity increases due to the need for mapping between different encoding formats
Solution Approach 1:
The mapping module serves as a dedicated intermediary component that handles the format transformation between classical encoding output and semantic encoding input. By isolating the complexity of format conversion into a single specialized module, the overall system structure remains manageable while achieving successful integration of the two encoding systems.
Solution Approach 2:
The mapping module is designed with universal functionality to handle various types of codeword transformations. This multi-functional component can adapt to different classical encoding schemes and output formats, reducing the need for multiple specialized mapping modules and thereby controlling system complexity while maintaining high integration capability.
Data Source
Figure 1~2
Figure 3
Figure 4~6
AI summary
The present disclosure provides a vector matrix determination method and system, a storage medium, and an electronic device. The method comprises: coding a message to be transmitted into a codeword matrix, wherein the codeword matrix comprises a regular codeword matrix and an irregular codeword matrix, and the codeword matrix is a three-dimensional codeword matrix; and converting the codeword matrix by means of a fully-connected network group to obtain a vector matrix having the same codeword length. The technical solution solves the technical problems in the related art that existing semantic communication systems do not relate to classical data coding and the semantic communication systems cannot fuse classical communication due to sentences required to be transmitted having the same length.