Infrastructure that can weather the storm: AI network management for natural disasters

Network management for natural disasters and climate change

The telecommunications sector faces escalating challenges from climate change. Extreme weather—which is becoming more frequent and severe—disrupts networks and causes significant financial losses. For example, in 2017, T-Mobile faced €250 million in hurricane-related costs, and wildfires disrupted 77% of Australia’s telecom infrastructure. Over the past decade, weather-related losses for telecom operators have surged to $476 billion globally. Regulatory bodies, including Ofcom and the FCC, are mandating resilience measures, such as disaster roaming and infrastructure hardening, to mitigate these risks.

Against this backdrop, the GenAI-powered toolkit for network and service management - Phase II Catalyst, which won the Outstanding Catalyst – Innovative & Futuristic Catalyst Award at Innovate Asia 24, provides a timely and innovative response. Building on its award-winning Phase I, the initiative addresses the dual challenge of ensuring continuous connectivity during crises while aligning with increasingly stringent regulatory requirements for network management in natural disasters.

How AI enhances network management for natural disasters

The project team has used AI to provide a unified, intelligent system for resilient network management. The solution combines fault, performance, and alarm management into an automated framework, streamlining operations and enhancing network reliability during extreme weather events.

The solution uses a variety of TM Forum assets for this purpose. The TMF921 Intent Management API is used to translate high-level service requests—such as disaster roaming or RAN reconfiguration—into actionable network operations. Meanwhile, TMF640 Service Order Fulfilment protocols are used to execute these operations seamlessly, ensuring that services are restored or optimized in real time. The integration of TMF639 for resource inventory management and TMF620 for service catalogues enables precise resource alignment, reducing inefficiencies during critical incidents.

A key innovation is the integration of a knowledge graph which maps relationships between network elements, faults, and inventory data. Additionally, related data from outside the telecommunications domain—such as weather forecasts, unplanned events (e.g. festivals or strikes), and the impacts of other non-telecom networks—enhances the graph’s capabilities. This graph drives predictive fault correlation, enabling the system to anticipate and mitigate potential disruptions before they escalate. For example, during heatwaves, the system identifies overheating risks in base stations and proactively redirects traffic while shutting down vulnerable nodes. This predictive capability helps ensure minimal disruption while protecting infrastructure.

Fault management is further enhanced by machine learning algorithms that analyze network logs and user reports, significantly improving fault diagnosis and response times. Continuous monitoring of KPIs identifies patterns and predicts potential performance issues, while alarm management prioritizes critical alerts, allowing operators to focus on issues with the highest service impact.

The solution’s RAN self-healing and optimization capabilities are rooted in an event-driven architecture where not only negative events, but also positive and neutral events, are taken into account to manage the network. The AI autonomously adjusts network configurations based on real-time telemetry data. During emergencies, it triggers intent-driven workflows to reallocate capacity, reroute traffic, or harden infrastructure—all without manual intervention.

This Catalyst builds on TM Forum’s Autonomous Network framework, integrating closed-loop automation for end-to-end service assurance. By aligning with these standards, the system meets regulatory requirements, such as Ofcom’s resilience mandates and the FCC’s Disaster Response Initiative. Automated compliance reduces the operational burden on CSPs while enhancing service continuity.

Energy efficiency is another key feature of the system. The AI dynamically matches network capacity to demand, shutting down underused stations during low-traffic periods. This reduces energy consumption and operational costs while maintaining service quality, aligning with sustainability objectives and reducing the carbon footprint of telecom networks.

The future of network management for natural disasters

This Catalyst offers a regulatory-compliant solution to the impact of climate change on telecom infrastructure, highlighting its importance in network management for natural disasters. While the investment may seem substantial, its cost-effectiveness is notable: Ofcom’s proposed power backup mandates could cost UK CSPs £1.8 billion, but this Catalyst achieves full coverage at an estimated 40% of that cost. As extreme weather events grow more frequent and intense, the relevance of this solution across global markets is undeniable, especially in high-impact regions such as Asia-Pacific and North America, where annual damages are projected to reach $165 billion and $159 billion, respectively, by 2025.

The solution also supports critical industries reliant on connected infrastructure, such as utilities, transportation, and emergency services. By ensuring continuous connectivity, it mitigates secondary impacts of disasters, including delayed emergency responses and service disruptions. In the US, reducing emergency response times by just one minute has been shown to decrease mortality rates by up to 17%.

AI-driven energy optimization further enhances the system’s value. During non-emergency periods, it matches network capacity to demand, powering down underused stations without compromising service. This reduces operational costs, extends equipment lifespan, and aligns with global sustainability goals, including SDG 13 on climate action.

As extreme weather events continue to test infrastructure resilience worldwide, this Catalyst provides a replicable model for adaptive network management. Its scalability and cost-effectiveness make it particularly attractive for CSPs in emerging markets, where resource constraints often hinder the implementation of traditional resilience measures.

This Catalyst drew on contributions from participants at Clarity, Huawei, Panamax, Tech Mahindra, Technarts, Wavenet and Waylay. The team believes that the success of this project demonstrates the power of teamwork and diversity, and that collaborating across cultures not only enriched their perspectives but also fuelled their innovative solutions, particularly in harnessing genAI to drive efficiency and sustainability. Despite the challenges faced, the participants forged strong relationships, demonstrating that their diverse backgrounds were key both to their resilience and their ability to create meaningful impact.