Invention Reference Number
Traditional table-based routing protocols in traffic engineering limit the ability to define specific packet paths, leading to suboptimal routing and network congestion, while effective flow allocation management requires complex calculations for optimal path selection. This invention addresses these challenges with a twofold approach that enhances traffic engineering through path-aware routing.
By employing segment routing alongside a polynomial residue number system (RNS), it improves routing flexibility and failure recovery. Additionally, it integrates AI techniques, including supervised learning for traffic flow analysis and deep reinforcement learning for network optimization, allowing the system to predict traffic patterns and adapt dynamically to changes. This approach facilitates real-time bandwidth predictions and the selection of optimal network paths, significantly enhancing overall network performance and reducing congestion.
Description
SDN-based traffic engineering (TE) tools utilize classification techniques or deep reinforcement learning to identify optimal network TE solutions demonstrated through simulation. Routing control conducted via source routing tools. A novel framework leverages AI to practically implement TE on a real network in collaboration with a source routing tool. Utilizing real-time traffic statistics, AI techniques compute optimal paths, which are then communicated to segment routing for flow allocation. Several contributions demonstrate a practical implementation of this framework tested using an emulated ecosystem that mimics a real P4 testbed scenario. This work is valuable for the development of engineered self-driving networks, effectively translating theory into practice.
Benefits
- Juniper solutions
- Cisco Segment routing
- DriveNets
Applications and Industries
- Path-based monitoring solution tool
- AI-based path selection tool
- Flow-based path optimizations
- Traffic engineering tool
Contact:
To learn more about this technology, email partnerships@ornl.gov or call 865-574-1051.
