Cluster Oriented Dynamic Routing for Network Reliability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing routing protocols in large-scale packet networks face challenges in efficiently determining optimal paths due to non-deterministic latency, limited representational structure of identifiers, and the complexity of dynamic network operations, leading to potential routing loops and blackholes.
Innovation Solution
The Cluster Oriented Dynamic Routing (CODR) algorithm, which uses group vectors to represent network topology, allows for distributed route calculation and identifies network identifiers independent of destination identifiers, enabling route determination based on network structure and dynamic status information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional routing protocols (OSPF, IS-IS, BGP) are used to determine loop-free reachability in large-scale networks, then routing reliability is improved, but computational complexity and routing overhead increase significantly
Solution Approach 1:
The patent segments the network into hierarchical levels (Level 0 to Level N) based on cluster relationships. Each node maintains routing information only for its local cluster and uses cluster-level identifiers (CLIDs) for external communication. This segmentation reduces the routing table size from O(N²) to O(N) where N is the number of clusters, directly reducing computational complexity while maintaining loop-free routing through hierarchical level information.
Solution Approach 2:
The patent introduces cluster-level identifiers (CLIDs) as intermediary routing information. Instead of exchanging full node identifiers across the entire network, nodes use CLIDs that represent their cluster membership. This intermediary layer abstracts the complex network topology into manageable cluster groups, reducing routing overhead and computational burden while preserving end-to-end connectivity.
2Adaptability or versatility
If distributed route calculation is implemented in large networks, then routing adaptability is improved, but computational effort and convergence time increase
Solution Approach 1:
The patent divides the distributed routing calculation into hierarchical segments. Level 0 nodes perform local cluster routing calculations independently, while higher-level nodes handle inter-cluster routing. This segmentation allows parallel processing of routing updates across different hierarchical levels, reducing overall convergence time while maintaining distributed adaptability to local network changes.
Solution Approach 2:
The patent pre-establishes hierarchical cluster relationships and assigns level identifiers to nodes before routing operations begin. This preliminary structuring of the network topology enables faster route calculation during dynamic events, as nodes already have the hierarchical framework in place to quickly determine routing paths without performing complete network-wide calculations.
3Ease of operation
If network identifiers are based solely on destination addresses, then routing simplicity is improved, but ability to represent network topology and optimize paths is reduced
Solution Approach 1:
The patent merges destination address information with cluster-level identifier (CLID) information to create enhanced routing entries. Each routing entry contains both the traditional destination address for simple lookup and the CLID for topology-aware routing decisions. This combination preserves routing simplicity through straightforward address matching while adding topological optimization capabilities through hierarchical level information.
Solution Approach 2:
The patent adds a hierarchical dimension to traditional flat routing by incorporating level identifiers into the routing structure. Instead of a single-dimensional destination-based routing table, the system creates a multi-dimensional routing space that includes both destination address and cluster hierarchy level. This additional dimension enables topology-aware path selection while maintaining simple address-based lookup operations.
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
A method for routing traffic between a source node and a destination node in a network comprises: identifying a first group level at which a source group vector and a destination group vector differ; obtaining one or more first route segments between a first source group and a first destination group at the first group level; computing an end-to-end route between the source node and the destination node based on at least the one or more first route segments; and routing traffic between the source node and the destination node on the end-to-end route.