Multicast Path Switching Without Packet Loss
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current multicast routing protocols face issues with packet loss and duplication during path switching in multicast networks, particularly when transitioning between shared trees and shortest path trees, as they often rely on delaying the discontinuation of the old path and using acknowledgment messages, which can result in traffic being sent on both paths and received twice by downstream routers.
Innovation Solution
Implementing a 'make-before-brake' approach that uses modified Join and Prune messages to prepare intermediary routers for traffic switching, ensuring that the old path is discontinued only after the new path is established, and managing the switch from an upstream router common to both paths, without relying on acknowledgment messages or packet numbering.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If the old path is discontinued before the new path is established, then the network complexity is reduced, but packet loss occurs during path switching
Solution Approach 1:
The patent applies preliminary action by establishing the new path before discontinuing the old path. The upstream router receives a Join message for the new path, marks packets accordingly, and only then stops sending packets on the old path. This ensures seamless path switching without packet loss while managing network complexity.
2Reliability
If the old path is delayed in discontinuation, then packet loss is avoided, but traffic is sent on both paths and received twice by downstream routers
Solution Approach 1:
The patent uses packet marking (analogous to color changes) to distinguish packets from different paths. The upstream router marks packets with a flag indicating they are from the new path. Downstream routers use this marking to identify and discard duplicate packets, preventing harmful duplication while maintaining reliability during the transition period.
Solution Approach 2:
The patent implements feedback through the marking mechanism. When the upstream router marks packets as belonging to the new path, this information flows downstream, enabling downstream routers to make informed decisions about packet handling. This feedback loop prevents duplicate packet processing while ensuring smooth path transition.
3Reliability
If acknowledgment messages are used to manage path switching, then packet loss is prevented, but the device complexity and signaling overhead increase
Solution Approach 1:
The patent extracts the essential function of acknowledgment from complex signaling sequences. Instead of using multi-message acknowledgment protocols, the invention uses a simple packet marking mechanism where the upstream router directly marks packets to indicate path transition. This eliminates the need for complex acknowledgment message exchanges while maintaining reliability.
4Adaptability or versatility
If traffic switching is managed without coordination from upstream router, then routing flexibility is improved, but packet loss and duplication occur at path transition points
Solution Approach 1:
The upstream router acts as an intermediary that coordinates path switching for all downstream routers. By receiving the Join message and marking packets centrally, the upstream router ensures synchronized path transition across the entire multicast tree. This intermediary role maintains routing flexibility while preventing packet loss and duplication at transition points.
Data Source
Figure 1~2
Figure 3~4
Figure 5
AI summary
Multicast signals are transmitted from a source (S) to user terminals (RA, RB, RC) via multicast paths through a set of nodes (R1 to R11) in a network. A first node, such as a 'last-hop' router (R7), may send towards the source (S) corresponding join and prune messages, the join message to establish a new multicast path (R1, R2, R4, R6, R7) from the source (S) to a user terminal (RB) and the prune message to discontinue the old multicast path (R1, R3, R8, R9, R10, R7) from the source (S) to the user terminal (RB). The router (R7) sends the join message and the prune message over the new multicast path (R1, R2, R4, R6, R7) and the old multicast path (R1, R3, R8, R9, R10, R7). A second node (R1) located at the bifurcation of the new multicast path (R1, R2, R4, R6, R7) and the old multicast path (R1, R3, R8, R9, R10, R7) detects the reception of both the join message and the prune message and steers, either directly or via delegated downstream nodes, the switch from the old multicast path (R1, R3, R8, R9, R10, R7) over to the new multicast path (R1, R2, R4, R6, R7) without loss or undue duplication of packets sent form the source (S) to the user terminal (RB).