Learned codes for broadcast channels with feedback

2024-11-06

arXiv.org

AI Summary

This work addresses the symmetric Gaussian broadcast channel (GBC) with feedback, tackling the underexplored setting of finite blocklength multi-user communication under imperfect feedback. Method: We extend the end-to-end learned error-correcting coding paradigm to this scenario via a lightweight, fully differentiable neural encoder-decoder architecture. The design jointly optimizes encoding and decoding while explicitly modeling non-ideal channel feedback; robust training is achieved through parameterized Gaussian channel simulation. Results: At short blocklengths (n ≤ 128), our scheme achieves frame error rates over one order of magnitude lower than conventional linear codes and state-of-the-art learning-based methods, significantly enlarges the achievable rate region, and exhibits strong robustness to feedback noise. Key contributions include: (i) the first end-to-end learning framework for feedback-aided GBC; (ii) a compact architecture balancing performance and complexity; and (iii) a differentiable system-level modeling approach supporting imperfect feedback.

Technology Category

• Search and Optimization: Learning to Search
• Machine Learning: Neuro-Symbolic Learning
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization

Abstract

We focus on designing error-correcting codes for the symmetric Gaussian broadcast channel with feedback. Feedback not only expands the capacity region of the broadcast channel but also enhances transmission reliability. In this work, we study the construction of learned finite blocklength codes for broadcast channels with feedback. Learned error-correcting codes, in which both the encoder and decoder are jointly trained, have shown impressive performance in point-to-point channels, particularly with noisy feedback. However, few learned schemes exist for multi-user channels. Here, we develop a lightweight code for the broadcast channel with feedback that performs well and operates effectively at short blocklengths.

Problem

• Design error-correcting codes for Gaussian broadcast channels with feedback.
• Enhance transmission reliability using feedback in broadcast channels.
• Develop lightweight codes for short blocklengths in multi-user channels.

Innovation

• Learned error-correcting codes for broadcast channels
• Jointly trained encoder and decoder for feedback
• Lightweight code effective at short blocklengths