Reverse CSPF Path Computation for Multi-Domain MPLS Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for computing Traffic Engineering Label Switched Paths (TE LSP) across multiple domains in MPLS networks are inefficient, as they require computing a large number of paths in each domain, reducing computation efficiency and speed, especially when dealing with a high number of boundary nodes.
Innovation Solution
Implementing a 1-to-n reverse CSPF algorithm or a 1-to-n normal CSPF algorithm, using algorithms such as reverse CSPF-Dijkstra and CSPF-A* or CSPF-Dijkstra and reverse CSPF-A*, respectively, to compute paths in each domain based on the number of starting and ending nodes, thereby reducing the number of shortest path computations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional methods compute all shortest paths in each domain for TE LSP across multiple domains, then path computation completeness is ensured, but computation efficiency and speed deteriorate significantly
Solution Approach 1:
The patent divides the path computation problem into segments across multiple domains. Instead of computing all shortest paths in each domain simultaneously, the system computes paths segment by segment, using reverse CSPF algorithm to determine path segments in reverse order from destination to source, thereby reducing the computational burden in each domain while maintaining overall path computation completeness.
Solution Approach 2:
The patent employs reverse CSPF algorithm that computes paths from destination nodes backward to source nodes, rather than the conventional forward computation approach. This inversion allows the system to determine path segments in reverse order, reducing the number of paths that need to be computed in each domain while ensuring complete path coverage.
2Adaptability or versatility
If the number of boundary nodes increases in multi-domain networks, then network scalability is improved, but computation time and processing load increase significantly
Solution Approach 1:
The patent applies partial action by computing only the necessary path segments rather than all possible paths. The reverse CSPF algorithm computes paths from destination to source, allowing the system to process only the essential path segments needed for TE LSP establishment, thereby reducing computation time while maintaining scalability with increasing boundary nodes.
3Measurement precision
If conventional CSPF algorithm computes shortest paths from all source nodes to all destination nodes, then path computation thoroughness is ensured, but processing speed deteriorates
Solution Approach 1:
The patent inverts the conventional CSPF approach by using reverse CSPF algorithm that computes paths from destination nodes backward to source nodes. This inversion enables the system to compute path segments more efficiently by processing destination nodes first, thereby maintaining computation thoroughness while improving processing speed.
Solution Approach 2:
The patent performs preliminary actions by pre-computing reverse paths from destination nodes to potential source nodes before final path determination. This preliminary computation of reverse paths allows the system to quickly assemble complete paths without recomputing all shortest paths from scratch, thereby maintaining thoroughness while enhancing speed.
Data Source
AI summary
An apparatus comprising a path computation element (PCE) associated with a domain in a network and configured to find a segment of a Multiprotocol Label Switching (MPLS) Traffic Engineering (TE) Label Switched Path (LSP) that crosses a plurality of domains in the network using a Constraint Shortest Path First (CSPF) algorithm or a reverse CSPF algorithm that computes a plurality of shortest paths in the domain of which the segment is selected, wherein the CSPF algorithm or the reverse CSPF algorithm is selected to reduce the number of shortest path computations in the domain based on the number of starting nodes and ending nodes that are considered for computing the shortest paths in the domain.


