Dynamic Graph Routing for Interconnection Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing interconnection networks fail to guarantee fully connected operation when multiple links fail, leading to fatal errors and network shutdowns, as they are limited to handling single link failures using fixed binary-coded routing tables.
Innovation Solution
A dynamic routing algorithm that uses graph theory to explore and update the interconnection fabric, identifying alternative routes by distinguishing between terminating and non-terminating vertices and operative links, and writing routing registers to handle multiple link failures.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If fixed binary-coded routing tables are used to identify routes, then single link failure can be handled, but multiple link failures result in fatal errors and network shutdown
Solution Approach 1:
The patent transitions from static binary-coded routing tables to dynamic graph-based routing that adapts to network conditions. The system continuously updates the graph representation of the interconnection fabric and dynamically computes alternate routes when link failures occur, enabling adaptation to multiple simultaneous failures rather than being restricted to predetermined single-failure scenarios
Solution Approach 2:
The patent changes the fundamental parameter representation from fixed binary codes to flexible graph structures with vertices and edges. This parameter change allows the routing system to represent and process complex network topologies and failure scenarios that cannot be captured by traditional binary routing tables, thereby improving adaptability to multiple link failures
2Ease of operation
If simple table lookup is used for routing, then single bad link can be worked around, but multiple bad links are not guaranteed to have viable alternate paths
Solution Approach 1:
The patent introduces a graph structure as an intermediary between the physical network and the routing decision-making process. This graph intermediary explicitly models network topology, link status, and connectivity relationships, enabling the system to systematically explore and identify viable alternate paths through graph traversal algorithms when multiple links fail, rather than relying on insufficient simple table lookups
3Productivity
If hard coded tables examine links in hardware-supported order, then routing register values can be determined, but the system is restricted to single bad link scenarios
Solution Approach 1:
The patent performs preliminary actions by continuously maintaining an up-to-date graph representation of the network fabric before failures occur. When link failures happen, the system can immediately query the pre-maintained graph structure and rapidly compute alternate routes through graph algorithms, combining the speed advantage of pre-computation with the flexibility to handle multiple simultaneous failures
Data Source
AI summary
Routes between cells in an interconnection network are represented with a graph using graph theory. Failed links are identified, and using the graph, alternative routes between cells in the graph are found.