AI-Defined SDN Routing Agent for Network Convergence
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional computer networks face challenges in dynamic routing protocols, such as increased convergence time and computational overhead, especially in large routing domains, and require manual administration, which can be error-prone and resource-intensive.
Innovation Solution
Implementing AI-defined networking using reinforcement learning-trained neural network agents that make routing decisions based on network policies, trained through digital twin simulations, to automate and optimize routing decisions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional dynamic routing protocols are used to determine packet path selection, then routing decisions can be made based on network conditions, but convergence time increases and computational overhead increases substantially in large routing domains
Solution Approach 1:
The patent replaces traditional mechanical routing protocols (BGP, EIGRP, OSPF) with an AI-based neural network system that processes routing decisions. The neural network agent learns optimal routing policies through reinforcement learning and directly determines packet paths without undergoing traditional convergence processes, thereby eliminating the time penalty associated with protocol convergence while maintaining adaptability to network conditions.
Solution Approach 2:
The patent creates a digital twin simulation environment that replicates the physical network topology and conditions. The neural network agent is trained in this virtual copy of the network, allowing it to learn routing strategies without impacting the actual network. This enables rapid iteration and optimization of routing policies without the convergence time penalties of traditional protocols.
2Adaptability or versatility
If traditional dynamic routing protocols are used to determine packet path selection, then routing decisions can be made based on network conditions, but computational overhead increases substantially
Solution Approach 1:
The patent performs routing policy optimization in advance through training the neural network agent in a digital twin simulation environment. The agent learns optimal routing strategies beforehand through reinforcement learning, so that during actual network operation, routing decisions can be made quickly with minimal computational overhead. The heavy computational work is done preliminarily during training rather than in real-time network operation.
Solution Approach 2:
The patent replaces computationally intensive traditional routing protocol calculations with a trained neural network model that makes routing decisions through pattern recognition. Once trained, the neural network can evaluate routing options and select optimal paths with significantly lower computational requirements than protocol-based route calculation and convergence processes.
3Adaptability or versatility
If manual configuration and tuning of routing protocols is performed to enforce business specifications, then routing behavior can be optimized for specific traffic types or links, but human error increases and substantial effort is required
Solution Approach 1:
The patent implements self-service routing optimization where the neural network agent autonomously learns and determines optimal routing policies to satisfy business specifications. Instead of requiring manual configuration and tuning by human administrators, the system automatically adapts routing behavior through reinforcement learning, eliminating human error and the substantial effort required for manual route tuning and traffic engineering.
Solution Approach 2:
The patent replaces manual human configuration and tuning operations with an AI-based neural network system. The neural network agent automatically learns routing strategies that satisfy business specifications without requiring human administrators to manually configure protocols or tune parameters, thereby eliminating human error and reducing operational effort while maintaining full adaptability to business requirements.
4Reliability
If multiple routes are allocated and held in routing tables to provide primary, secondary and tertiary paths, then routing redundancy is improved, but hardware resources and table space are consumed
Solution Approach 1:
The patent implements dynamic route selection where the neural network agent determines the optimal routing path based on current network conditions rather than pre-configuring multiple static routes. The system maintains routing flexibility and redundancy through learned policies that can adapt to changing conditions, eliminating the need to allocate and store multiple candidate routes in hardware tables while preserving reliability through intelligent dynamic path selection.
Data Source
AI summary
A system including one or more processors and one or more non-transitory computer-readable media storing computing instructions that, when executed on the one or more processors, perform certain acts. The acts can include receiving a deployment model selection of a software-defined-network (SDN) control service. The deployment model selection includes one of a centralized model, a decentralized model, a distributed model, or a hybrid model. The acts also can include deploying the SDN control service in the deployment model selection to control a physical computer network. The SDN control service uses a routing agent model trained using a reinforcement-learning model. Other embodiments are described.