Hybrid CNN-Transformer Based Sparse Channel Prediction for High-Mobility OTFS Systems

2025-10-18

AI Summary

In high-mobility scenarios, OFDM systems suffer from severe channel prediction degradation due to Doppler shifts and channel aging. To address this, this paper proposes a high-accuracy sparse channel prediction framework tailored for Orthogonal Time Frequency Space (OTFS) modulation. Leveraging OTFS’s quasi-static nature in the delay-Doppler (DD) domain, we design a CNN-Transformer hybrid architecture: a convolutional neural network (CNN) captures spatial sparsity patterns in the DD-domain channel response, while a causal-masked Transformer models temporal dynamic dependencies. Evaluated under extreme mobility (500 km/h), the method achieves a 12.2% reduction in root-mean-square error (RMSE) and a 9.4% decrease in mean absolute error (MAE) compared to baseline approaches. These improvements significantly enhance prediction accuracy and robustness—critical requirements for ultra-reliable low-latency communication (URLLC). The framework establishes a deployable, OTFS-native paradigm for channel prediction in high-speed wireless systems.

Technology Category

• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Machine Learning: Mixture of Experts (MoE)
• Application Domains: Mobility, Driving & Flight

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Social Networks and Social Media: Influence propagation, information diffusion, and the prediction on networks
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

High-mobility scenarios in next-generation wireless networks, such as those involving vehicular communications, require ultra-reliable and low-latency communications (URLLC). However, rapidly time-varying channels pose significant challenges to traditional OFDM-based systems due to the Doppler effect and channel aging. Orthogonal time frequency space (OTFS) modulation offers resilience by representing channels in the quasi-static delay-Doppler (DD) domain. This letter proposes a novel channel prediction framework for OTFS systems using a hybrid convolutional neural network and transformer (CNN-Transformer) architecture. The CNN extracts compact features that exploit the DD-domain sparsity of the channel matrices, while the transformer models temporal dependencies with causal masking for consistency. Simulation experiments under extreme $500$ si{km/h} mobility conditions demonstrate that the proposed method outperforms state-of-the-art baselines, reducing the root mean square error and mean absolute error by $12.2%$ and $9.4%$, respectively. These results demonstrate the effectiveness of DD-domain representations and the proposed model in accurately predicting channels in high-mobility scenarios, thereby supporting the stringent URLLC requirements in future wireless systems.

Problem

• Predicting time-varying channels in high-mobility wireless systems
• Addressing Doppler effects and channel aging in OTFS modulation
• Enhancing accuracy for ultra-reliable low-latency vehicular communications

Innovation

• Hybrid CNN-Transformer architecture for channel prediction
• Exploits delay-Doppler domain sparsity with CNN features
• Models temporal dependencies using causal masking transformers