Topics In Demand
Notification
New

No notification found.

Building the Future of 6G Mobile Networks with Causal AI
Building the Future of 6G Mobile Networks with Causal AI

December 6, 2024

AI 5G

36

0

As research efforts in Mobile Wireless Networks (MWNs) transition from 5G to 6G, one of the most promising transformations is the integration of Artificial Intelligence (AI) as a native element of these networks. Unlike in previous generations, where Machine Learning (ML) and AI could be incorporated later for specific tasks, 6G envisions AI as a core component from the outset, driving network planning, operation, and optimization. To fully realize this vision, AI systems must be both robust and explainable, ensuring trust and reliability in their applications.

The Need for Robust and Explainable AI in 6G Networks

The transition to AI-native networks introduces significant challenges, particularly regarding the robustness and explainability of AI systems. On one hand, robustness ensures that AI systems handle a variety of scenarios with consistent performance, even when faced with unexpected changes such as shifts in data distribution. On the other hand, explainability is crucial for human operators to trust and understand the decisions made by or proposed by AI systems. Without these qualities, the deployment of AI in MWNs, especially for critical applications such as network optimization, will remain a major challenge.

The Role of Causality towards 6G Networks

A promising method for enhancing both robustness and explainability in AI systems is the integration of causality, a concept often missing from current AI and ML models. Although modern ML and deep learning techniques excel in predictive capabilities, they are constrained by a lack of process understanding and an inability to generalize beyond the data on which they were trained. To create more robust models, grasping the underlying causal relationships between variables is essential. Causal approaches enable AI systems not only to recognize patterns in data but also to comprehend the fundamental factors that drive these patterns. As Judea Pearl notes in The Book of Why: The New Science of Cause and Effect, “while probabilities encode our beliefs about a static world, causality tells us whether and how probabilities change when the world changes, be it by intervention or by act of imagination.” This insight highlights the transformative potential of harnessing causality. It can enable AI systems to adapt to real-time changes, providing more reliable and actionable insights in dynamic environments such as in mobile networks and 6G systems.

A Novel Methodology for Causal Network Optimization

The Wireless Global Delivery Team of CYIENT, has recently proposed a data-driven methodology that leverages unsupervised learning and causal concepts to optimize network performance in MWNs. This methodology, detailed in the article Enhancing Robustness for Automated Mobile Network Optimization by Uncovering Causal Relationships, moves beyond traditional AI/ML approaches that often lack causal foundations. The proposed methodology is depicted in Figure1.

Proposed methodology for causal network optimization
                                    Figure 1- Proposed methodology for causal network optimization.

This methodology uses both time-based inputs from Performance Management (PM) data and Configuration Management (CM) data collected from multiple Base Stations (BSs). Three processing modules then analyse these data, utilizing AI/ML models to generate target optimization actions supported by causal relationships. The processing modules are described below:

  • Unsupervised Performance Inquiry: This module identifies distinct performance patterns within the network using PM data. By applying a Temporal Convolutional Network (TCN)-based autoencoder, the methodology reduces the complexity of multivariate time series data, enabling more effective clustering and the recognition of performance patterns.
  • Radio Coverage Contextualization: Acknowledging that radio coverage is a critical factor influencing network performance, this module estimates metrics such as Synchronization Signal Reference Signal Received Power (SS-RSRP) and Signal-to-Interference-plus-Noise Ratio (SINR), based on a propagation model known as Ubiquitous Satellite Aided Radio Propagation (USARP)1. These metrics serve as additional inputs for the causal analysis, enriching the process with essential contextual information.
  • Performance Causal Learning & Knowledge Integration: At the core of the methodology, this module uses the identified performance patterns, along with configuration management (CM) data and radio coverage metrics, to uncover causal relationships affecting network performance. By utilizing a Long Short-Term Memory (LSTM) model integrated with SHAP values for interpretability and conditional independence tests, this methodology can derive actionable insights for network optimization based on causal foundations.

Evaluation and Results

The proposed methodology was evaluated using data from a live 5G network. The evaluation demonstrated its ability to identify causal associations between specific network configurations and performance issues, leading to clear optimization actions that can improve network performance. For instance, the evaluation found that certain BSs with lower performance were linked to specific CM parameters, allowing for targeted adjustments to enhance overall network efficiency.

Implications for the Future of 6G Networks

The introduction of causal AI approaches into MWNs can mark a significant advancement toward achieving more autonomous and intelligent networks. By embedding causality into AI models, we not only improve their robustness and explainability but also create scalable solutions adaptable to diverse network architectures. This integration promises to drive the evolution of more efficient and reliable networks, unlocking new avenues for optimization and innovation. Advancing towards 6G, causal AI will likely play a pivotal role in shaping the future of network technology, making it smarter and more capable of meeting the demands of an increasingly connected world.


That the contents of third-party articles/blogs published here on the website, and the interpretation of all information in the article/blogs such as data, maps, numbers, opinions etc. displayed in the article/blogs and views or the opinions expressed within the content are solely of the author's; and do not reflect the opinions and beliefs of NASSCOM or its affiliates in any manner. NASSCOM does not take any liability w.r.t. content in any manner and will not be liable in any manner whatsoever for any kind of liability arising out of any act, error or omission. The contents of third-party article/blogs published, are provided solely as convenience; and the presence of these articles/blogs should not, under any circumstances, be considered as an endorsement of the contents by NASSCOM in any manner; and if you chose to access these articles/blogs , you do so at your own risk.


Cyient (Estd: 1991, NSE: CYIENT) delivers Intelligent Engineering solutions across Product, Plant and Network for digital, autonomous and sustainable future to over 300 customers, including 40% of the top 100 global innovators. As a company, Cyient is committed to designing a culturally inclusive, socially responsible, and environmentally sustainable tomorrow together with our stakeholders.

© Copyright nasscom. All Rights Reserved.