Pilot Contamination Aware Transformer for Downlink Power Control in Cell-Free Massive MIMO Networks

Atchutaram K. Kocharlakota,Sergiy A. Vorobyov,Robert W. Heath Jr

2024-11-28

Abstract:Learning-based downlink power control in cell-free massive multiple-input multiple-output (CFmMIMO) systems offers a promising alternative to conventional iterative optimization algorithms, which are computationally intensive due to online iterative steps. Existing learning-based methods, however, often fail to exploit the intrinsic structure of channel data and neglect pilot allocation information, leading to suboptimal performance, especially in large-scale networks with many users. This paper introduces the pilot contamination-aware power control (PAPC) transformer neural network, a novel approach that integrates pilot allocation data into the network, effectively handling pilot contamination scenarios. PAPC employs the attention mechanism with a custom masking technique to utilize structural information and pilot data. The architecture includes tailored preprocessing and post-processing stages for efficient feature extraction and adherence to power constraints. Trained in an unsupervised learning framework, PAPC is evaluated against the accelerated proximal gradient (APG) algorithm, showing comparable spectral efficiency fairness performance while significantly improving computational efficiency. Simulations demonstrate PAPC's superior performance over fully connected networks (FCNs) that lack pilot information, its scalability to large-scale CFmMIMO networks, and its computational efficiency improvement over APG. Additionally, by employing padding techniques, PAPC adapts to the dynamically varying number of users without retraining.

Machine Learning,Information Theory

What problem does this paper attempt to address?

### Problems the paper attempts to solve

This paper aims to solve the downlink power control problem in large - scale cell - free massive multiple - input multiple - output (Cell - Free Massive MIMO, CFmMIMO) systems. Specifically, the paper focuses on the following key issues:

1. **Computational complexity**: Traditional iterative optimization algorithms have high computational complexity when dealing with large - scale networks, resulting in difficulties in real - time applications. Learning methods can reduce online iterative steps, but existing learning methods fail to fully utilize the internal structure of channel data and pilot allocation information, leading to poor performance.

2. **Pilot contamination**: In large - scale networks, due to the limited number of available orthogonal pilots, pilot contamination becomes a major problem. Existing learning methods usually ignore the impact of pilot contamination, which limits their application in large - scale networks.

3. **User dynamic changes**: The number of users in the network may change dynamically, and a model that can adapt to different numbers of users without retraining is required.

### Solutions

To solve the above problems, the paper introduces a new pilot contamination - aware power control (Pilot Contamination - Aware Power Control, PAPC) model based on the Transformer neural network. The main features of PAPC are as follows:

1. **Attention mechanism**: Utilize the attention mechanism of the Transformer to preserve the structural integrity of the two - dimensional large - scale fading coefficient matrix within the network range, avoiding the destruction of key inter - user relationships by common flattening operations.

2. **Pilot contamination matrix integration**: By modifying the standard Transformer block, introduce a new masking mechanism to integrate the pilot contamination matrix into the neural network, enabling PAPC to effectively handle pilot contamination scenarios.

3. **Pre - processing and post - processing stages**: Increase the accuracy of the model by adding pre - processing and post - processing stages. Pre - processing increases the input dimension, allowing the model to learn more abundant features from the large fading coefficients of each user; post - processing adjusts the dimension of the output power control coefficients and ensures compliance with power constraints.

4. **Adaptability**: Through customized post - processing modules and padding techniques, PAPC can handle different numbers of users without redesign or retraining.

5. **Reduction of hyper - parameters**: The hyper - parameters of the PAPC Transformer only depend on the number of base stations, not on the number of users as in the fully - connected network model, which improves the scalability and efficiency in large - scale CFmMIMO networks.

### Performance evaluation

The paper verifies the performance of the PAPC model by comparing it with the Accelerated Proximal Gradient (APG) algorithm. The results show that PAPC significantly improves computational efficiency while maintaining spectral efficiency fairness. Simulation results indicate that PAPC outperforms the fully - connected network (FCN) lacking pilot information in handling pilot contamination scenarios and has good scalability in large - scale CFmMIMO networks.

### Conclusion

The PAPC model effectively solves the downlink power control problem in large - scale CFmMIMO systems by combining the attention mechanism and the integration of the pilot contamination matrix, especially performing well in scenarios of pilot contamination and user dynamic changes. This provides strong support for the actual deployment of large - scale CFmMIMO networks.