Deep-OFDM: Neural Modulation for High Mobility

2025-06-20

AI Summary

In high-mobility scenarios, conventional OFDM suffers from severe inter-carrier interference (ICI) due to Doppler shifts and low spectral efficiency caused by excessive pilot overhead. Method: This paper proposes an end-to-end learnable neural modulation framework that jointly neuralizes the transmitter and receiver—eliminating explicit channel estimation and cyclic prefixes. A differentiable convolutional modulator learns low-rank implicit pilot structures directly from data, enabling pilot-free operation. The framework supports MIMO and incorporates optimized nonlinear activations to enhance robustness against Doppler-induced distortions. Contribution/Results: Experiments demonstrate a 40% reduction in block error rate (BLER) and a 2.3× improvement in goodput under high-Doppler conditions. Crucially, reliable communication is maintained without pilots, significantly outperforming conventional OFDM and state-of-the-art neural baseline methods.

Technology Category

• Machine Learning: Multimodal Learning
• Search and Optimization: Learning to Search
• Natural Language Processing: Learning & Optimization for NLP

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
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Orthogonal Frequency Division Multiplexing (OFDM) is the foundational waveform in current 5G deployments due to its robustness in quasi-static channels and efficient spectrum use. However, in high-mobility scenarios, OFDM suffers from inter-carrier interference (ICI), and its reliance on dense pilot patterns and cyclic prefixes significantly reduces spectral efficiency. In this work, we propose Deep-OFDM: a learnable modulation framework that augments traditional OFDM by incorporating neural parameterization. Instead of mapping each symbol to a fixed resource element, Deep-OFDM spreads information across the OFDM grid using a convolutional neural network modulator. This modulator is jointly optimized with a neural receiver through end-to-end training, enabling the system to adapt to time-varying channels without relying on explicit channel estimation. Deep-OFDM outperforms conventional OFDM when paired with neural receiver baselines, particularly in pilot-sparse and pilotless regimes, achieving substantial gains in BLER and goodput, especially at high Doppler frequencies. In the pilotless setting, the neural modulator learns a low-rank structure that resembles a superimposed pilot, effectively enabling reliable communication without explicit overhead. Deep-OFDM demonstrates significant improvements in BLER and goodput at high Doppler frequencies across various scenarios, including MIMO systems. Comprehensive ablation studies quantify the role of nonlinear activations and characterize performance-complexity trade-offs. These results highlight the potential of transmitter-receiver co-design for robust, resource-efficient communication, paving the way for AI-native physical layer designs in next-generation wireless systems.

Problem

• Reduces inter-carrier interference in high-mobility OFDM
• Improves spectral efficiency without dense pilot patterns
• Enables reliable communication in high Doppler scenarios

Innovation

• Neural modulation framework for OFDM
• End-to-end trained CNN modulator and receiver
• Low-rank structure enables pilotless communication