Architectural framework to control and optimize non-terrestrial networks with Artificial Intelligence

AI-controlled non-terrestrial networks: advancing connectivity with distributed intelligence and low-latency control

Background: With the growing use of non-terrestrial networks such as drones, satellites, and other infrastructure-less nodes, the challenge of reliably controlling and optimizing these networks has become more pronounced. Traditional centralized control systems are inadequate for managing the dynamic nature of non-terrestrial networks, which often operate in challenging environments with limited connectivity to ground-based control centers. The latency associated with centralized control can significantly impact network performance and reliability. The need for intelligent, distributed control systems that can operate autonomously while maintaining optimal performance is critical for the success of non-terrestrial network deployments.

Technical Overview: Northeastern researchers have developed a two-tiered architectural framework that leverages artificial intelligence (AI) for the control and optimization of non-terrestrial networks. The system integrates distributed AI capabilities at both the network edge and centralized control levels to provide low-latency, intelligent network management. The framework employs machine learning algorithms to predict network conditions, optimize resource allocation, and autonomously adapt to changing environmental conditions. The architecture is designed to operate effectively even with intermittent connectivity to ground-based control systems.

Benefits:

• Reduces Latency: The distributed AI framework lowers the delay traditionally associated with centralized control systems
• Enhances Network Reliability: Autonomous operation capabilities ensure continued functionality even with limited ground connectivity
• Optimizes Resource Utilization: AI algorithms efficiently allocate network resources based on real-time conditions
• Scalable Architecture: Framework can adapt to networks of varying sizes and complexity
• Predictive Capabilities: Machine learning enables proactive network optimization

Application:

• Non-Terrestrial Network Optimization: Effective for satellite and drone network management where centralized control is impractical
• Emergency Communication Systems: Providing reliable connectivity in disaster scenarios
• Remote Area Connectivity: Extending network coverage to underserved regions
• Military and Defense Applications: Secure, autonomous network operations

Opportunity:

• Research collaboration
• Licensing