Digital Deep Joint Source-Channel Coding with Blind Training for Adaptive Modulation and Power Control

2025-01-04

AI Summary

Existing Deep Joint Source-Channel Coding (DeepJSCC) systems rely heavily on channel-specific training data, limiting adaptability to unknown wireless environments and compromising both reconstruction quality and energy efficiency. This paper proposes a digital DeepJSCC framework enabling blind training and adaptive modulation-power control—without prior channel knowledge. Our key contributions are: (1) an error-adaptive blind training mechanism that models bit-flip probability as a learnable parameter; (2) a training-aware dynamic encoder-decoder pair selection strategy coupled with joint power control; and (3) an end-to-end differentiable learning architecture built upon differentiable bit-flip optimization and binary symmetric channel modeling. Experiments demonstrate superior PSNR performance over state-of-the-art methods, significantly reduced total power consumption, and robust adaptation to diverse channel conditions using only a minimal set of encoder-decoder pairs—greatly simplifying real-world deployment.

Technology Category

• Search and Optimization: Learning to Search
• Machine Learning: Deep Generative Models & Autoencoders
• Cognitive Modeling & Cognitive Systems: Adaptive Behavior

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Energy management for devices in mobile Web and WoT environments
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

This paper proposes a novel digital deep joint source-channel coding (DeepJSCC) framework that achieves robust performance across diverse communication environments without requiring extensive retraining and prior knowledge of communication environments. Traditional digital DeepJSCC techniques often face challenges in adapting to various communication environments, as they require significant training overhead and large amounts of communication data to develop either multiple specialized models or a single generalized model, in pre-defined communication environments. To address this challenge, in our framework, an error-adaptive blind training strategy is devised, which eliminates the need for prior knowledge of communication environments. This is achieved by modeling the relationship between the encoder's output and the decoder's input using binary symmetric channels, and optimizing bit-flip probabilities by treating them as trainable parameters. In our framework, a training-aware communication strategy is also presented, which dynamically selects the optimal encoder-decoder pair and transmission parameters based on current channel conditions. In particular, in this strategy, an adaptive power and modulation control method is developed to minimize the total transmission power, while maintaining high task performance. Simulation results demonstrate that our framework outperforms existing DeepJSCC methods, achieving higher peak signal-to-noise ratio, lower power consumption, and requiring significantly fewer encoder-decoder pairs for adaptation.

Problem

• DeepJSCC
• Adaptability
• Training Data

Innovation

• AdaptiveTraining
• IntelligentCommunicationStrategies
• AutoPowerModulationControl