Adaptive Coding and Modulation Using Neural Network Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional communication systems independently design source coding, channel coding, and modulation, which limits their performance and does not effectively handle various channel models, especially non-linear channels.
Innovation Solution
An adaptive coding and modulation method using neural networks to jointly perform source coding, channel coding, and modulation by generating mapping and de-mapping functions based on communication channels, allowing data to be transmitted in a multi-dimensional space.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If channel coding and modulation are performed independently based on channel quality, then the system is easier to implement and design, but the communication performance is limited
Solution Approach 1:
The patent combines channel coding and modulation into a unified joint coding and modulation (JCM) framework. Instead of independently designing codebooks for channel coding and constellation maps for modulation, the system generates a single joint codebook that simultaneously performs both functions. This merging allows the system to achieve better communication performance by exploiting the synergistic effects between coding and modulation while maintaining manageable implementation complexity through unified codebook generation algorithms.
Solution Approach 2:
The joint codebook serves multiple functions simultaneously: it provides channel coding for error protection, modulation mapping for signal generation, and adaptive capability for varying channel conditions. The codebook generation process universally handles different modulation orders (QPSK, 16QAM, 64QAM, etc.) and channel qualities through a single framework, eliminating the need for separate design procedures for each coding and modulation scheme.
2Reliability
If symbols are transmitted in a two-dimensional space with independent coding rate and modulation index selection, then the system design is simpler, but the pair-wise Euclidean distances between symbols cannot be maximized
Solution Approach 1:
The patent extends the traditional two-dimensional symbol space into a multi-dimensional joint codebook structure. By organizing codebooks across multiple dimensions (different modulation orders, coding rates, and channel conditions), the system can maximize pair-wise Euclidean distances between symbols while maintaining manageable complexity through structured codebook generation and selection algorithms.
3Adaptability or versatility
If conventional modulation schemes are designed for linear channel models, then the design process is straightforward, but the schemes perform poorly on non-linear channels
Solution Approach 1:
The patent implements dynamic codebook generation that adapts to different channel models including non-linear channels. The codebook generation process dynamically adjusts parameters such as modulation order, coding rate, and constellation mapping based on the detected channel characteristics. This dynamic adaptation allows the system to maintain optimal performance across varying channel conditions without requiring completely separate design procedures for each channel type.
Data Source
AI summary
A method for adaptive coding and modulation. The method includes generating a set of mapping functions and transmitting a tth set of transmit symbols where 1≤t≤T and T is a maximum number of symbol transmissions. Transmitting the tth set of transmit symbols includes transmitting each transmit symbol in the tth set of transmit symbols. Each transmit symbol is transmitted by a respective transmitter. Transmitting each transmit symbol includes generating a tth set of mapped symbols, generating each transmit symbol from the tth set of mapped symbols, and transmitting each transmit symbol. Generating the tth set of mapped symbols includes applying a mapping functions subset of the set of mapping functions on a respective data vector. Each mapping function in the mapping functions subset depends on a respective mapped symbol in an rth set of mapped symbols where 0≤r≤T.