Traffic Flow Manager for TCP/IP Packet Delivery
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
TCP performance is degraded in wireless communication channels due to its inability to distinguish between network congestion and transmission errors, leading to inefficient error recovery and reduced throughput, especially in high-capacity wireless links with multiple simultaneous service data flows.
Innovation Solution
A system with a traffic flow manager that resides between the ARQ L2 layer and the IP stack, capable of receiving and reassembling out-of-order IP packets, prioritizing retransmitted TCP data packets, and limiting TCP ACK packets to maintain stable throughput, thereby reducing round trip time and preventing queue congestion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If TCP retransmission scheme is employed in wireless networks, then error recovery is achieved, but throughput is reduced due to queueing delays and inability to distinguish congestion from transmission errors
Solution Approach 1:
The patent introduces an intermediary mechanism at the network layer that sits between the physical layer error recovery and TCP retransmission. This intermediary uses Forward Error Correction (FEC) codes to recover from transmission errors before packets reach the TCP layer, preventing unnecessary TCP retransmissions and maintaining throughput while ensuring reliable delivery.
Solution Approach 2:
The patent segments the error recovery function into two distinct layers: physical layer FEC for transmission errors and TCP retransmission for congestion loss. By separating these functions and applying appropriate recovery mechanisms to each, the system avoids the throughput penalty of treating all packet losses as congestion events.
2Quantity of substance
If multiple service data flows are carried over wireless channel, then channel capacity is utilized, but error recovery becomes less effective as all flows are affected when link is in error recovery condition
Solution Approach 1:
The patent applies segmentation by assigning different FEC code rates to different service data flows based on their quality of service requirements. This allows some flows to be more robust against errors while others maintain higher throughput, ensuring that error recovery in one flow does not necessarily impact others.
Solution Approach 2:
The patent implements local quality by applying differentiated error protection strategies to different packets or flows based on their importance and tolerance to delays. Critical flows receive stronger FEC protection while less critical flows use weaker protection, optimizing overall system performance during error recovery conditions.
3Reliability
If TCP packets are queued for retransmission, then retransmission can occur, but round trip time increases to several seconds
Solution Approach 1:
The patent applies preliminary action by pre-correcting transmission errors using FEC before packets are transmitted over the wireless channel. This prevents packet loss at the source, eliminating the need for retransmission and the associated queueing delays that would otherwise occur.
Solution Approach 2:
The patent enables skipping by allowing the system to bypass the TCP retransmission queue entirely when FEC successfully recovers from errors. Packets that would otherwise be delayed in the retransmission queue are delivered immediately, reducing round trip time significantly.
Data Source
AI summary
Technologies for managing Transmission Control Protocol/Internet Protocol (TCP/IP) packet delivery include a network interface controller (NIC) of a computing device. The NIC is configured to retrieve identifying data from received IP packets and identify a TCP data stream associated with each received IP packet based on the retrieved identifying data. Additionally, the NIC is configured to determine a service data flow queue associated with the TCP data stream, determine whether a plurality of IP packets presently enqueued in the determined service data flow queue are out of order, and enqueue the received IP packet at a tail of the identified service data flow queue if the NIC determines that the plurality of IP packets presently enqueued in the determined service data flow queue are out of order. Other embodiments are described herein.