Non-Identical Diffusion Models in MIMO-OFDM Channel Generation

2025-09-01

AI Summary

In MIMO-OFDM systems, pilot-based channel estimation yields highly heterogeneous reliability across channel matrix entries, rendering scalar time-indexed diffusion models inadequate for capturing localized noise evolution. To address this, we propose a non-uniform diffusion model that introduces element-wise time indicators and dimension-wise time embeddings, enabling fine-grained, dynamic control of noise levels per subcarrier and antenna path. We further design a matrix-form time embedding scheme and a non-uniform noise scheduling mechanism, thereby relaxing the conventional uniform noise assumption. Theoretical analysis establishes the consistency of the generative process. Extensive experiments under diverse pilot configurations demonstrate that our model achieves an average 28.6% reduction in normalized mean square error (NMSE) for channel reconstruction compared to standard diffusion models, with particularly pronounced gains in sparse-pilot regimes.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Mixture of Experts (MoE)
• Planning, Routing, and Scheduling: Planning with Language Models

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs

Abstract

We propose a novel diffusion model, termed the non-identical diffusion model, and investigate its application to wireless orthogonal frequency division multiplexing (OFDM) channel generation. Unlike the standard diffusion model that uses a scalar-valued time index to represent the global noise level, we extend this notion to an element-wise time indicator to capture local error variations more accurately. Non-identical diffusion enables us to characterize the reliability of each element (e.g., subcarriers in OFDM) within the noisy input, leading to improved generation results when the initialization is biased. Specifically, we focus on the recovery of wireless multi-input multi-output (MIMO) OFDM channel matrices, where the initial channel estimates exhibit highly uneven reliability across elements due to the pilot scheme. Conventional time embeddings, which assume uniform noise progression, fail to capture such variability across pilot schemes and noise levels. We introduce a matrix that matches the input size to control element-wise noise progression. Following a similar diffusion procedure to existing methods, we show the correctness and effectiveness of the proposed non-identical diffusion scheme both theoretically and numerically. For MIMO-OFDM channel generation, we propose a dimension-wise time embedding strategy. We also develop and evaluate multiple training and generation methods and compare them through numerical experiments.

Problem

• Proposes non-identical diffusion model for MIMO-OFDM channel generation
• Addresses uneven reliability in initial channel estimates across elements
• Replaces scalar time index with element-wise noise progression control

Innovation

• Non-identical diffusion model with element-wise time indicators
• Matrix-based control for element-wise noise progression
• Dimension-wise time embedding for MIMO-OFDM channels