Next-Generation AI-Native Wireless Communications: MCMC-Based Receiver Architectures for Unified Processing

TL;DR

The paper addresses the escalating complexity and scalability of MIMO receivers in ultra-massive arrays by proposing an AI-native, MCMC-based universal Bayesian computing engine for the receiver. It formulates channel estimation, symbol detection, and decoding as posterior-inference tasks, employing gradient-based MCMC (e.g., Langevin) and modular blocks that enable a unified probabilistic receiver, while integrating data-driven priors via diffusion/energy-based models for end-to-end optimization. Key contributions include (1) a gradient-based MCMC framework for each receiver task, (2) a unified architecture that supports joint optimization and hybrid learning, and (3) performance demonstrations under 3GPP 3D channels showing improvements over traditional LMMSE pipelines and per-block MCMC approaches. The approach advances the 6G vision of AI-native networks by delivering interpretable, scalable, and adaptable receivers that can leverage generative priors and parallel hardware, paving the way for robust end-to-end Bayesian processing in future wireless systems.

Abstract

The multiple-input multiple-output (MIMO) receiver processing is a key technology for current and next-generation wireless communications. However, it faces significant challenges related to complexity and scalability as the number of antennas increases. Artificial intelligence (AI), a cornerstone of next-generation wireless networks, offers considerable potential for addressing these challenges. This paper proposes an AI-driven, universal MIMO receiver architecture based on Markov chain Monte Carlo (MCMC) techniques. Unlike existing AI-based methods that treat receiver processing as a black box, our MCMC-based approach functions as a generic Bayesian computing engine applicable to various processing tasks, including channel estimation, symbol detection, and channel decoding. This method enhances the interpretability, scalability, and flexibility of receivers in diverse scenarios. Furthermore, the proposed approach integrates these tasks into a unified probabilistic framework, thereby enabling overall performance optimization. This unified framework can also be seamlessly combined with data-driven learning methods to facilitate the development of fully intelligent communication receivers.

Figures (6)

• Figure 1: Computing architecture using gradient-based MCMC.
• Figure 2: Downlink channel estimation NMSE in a massive MIMO system with 64 BS antennas and 8 user terminals, each equipped with two antennas.
• Figure 3: BER performance in an ${\text{8}\times\text{8}}$ MIMO system with 16-QAM modulation under Rayleigh fading channels.
• Figure 4: Block error rate (BLER) performance under the 3GPP 3D channel model with 64 BS antennas and 8 or 16 single-antenna users, leading to effective ${\text{64}\times\text{8}}$ or ${\text{64}\times\text{16}}$ MIMO setups with 16-QAM modulation.
• Figure 5: Diagram of the MCMC-based unified receiver architecture. The left part illustrates the overall framework, while the right part details the integration of MCMC with data-driven NNs.