TCP Packet Loss Detection via Selective Acknowledgment and Probing

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

TCP performance is inefficient on long latency and high loss network links, particularly in wireless data networks, leading to unacceptable latency and throughput issues for transactional and real-time applications, due to its inability to accurately detect packet loss and adapt bandwidth usage.

Innovation Solution

A computer-implemented method that assigns sequential order numbers to TCP packets, uses ACK and SACK segments to determine received packet order, and retransmits unacknowledged packets with lower order numbers, along with a probing mechanism to expedite packet loss detection and retransmission, compatible with existing TCP stacks.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If standard TCP is used on long latency and high loss network links, then compatibility with existing networks is maintained, but packet loss detection accuracy deteriorates and throughput decreases

Engineering Contradiction:

ImprovecompatibilityVSAvoidthroughput

Core Design Contradiction:

Adaptability or versatilityVSProductivity

Solution Approach 1:

The patent segments the acknowledgment process by introducing SACK (Selective Acknowledgment) segments that separately acknowledge received packets. This allows the receiver to inform the sender which specific packets were received out of order, enabling more precise packet loss detection without changing the core TCP protocol, thus maintaining compatibility while improving throughput.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent introduces an intermediary mechanism (probe packets) that acts as a mediator between sender and receiver to detect packet loss. These probe packets are sent to trigger acknowledgments from the receiver, allowing the sender to determine whether data packets were lost without requiring modifications to the underlying TCP stack.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Device complexity

If RTO-based retransmission is used, then implementation simplicity is maintained, but retransmission latency increases

Engineering Contradiction:

Improveimplementation complexityVSAvoidretransmission latency

Core Design Contradiction:

Device complexityVSLoss of time

Solution Approach 1:

The patent performs preliminary actions by sending probe packets before the RTO timeout expires. These probe packets are designed to trigger acknowledgments from the receiver, allowing the sender to detect packet loss earlier than the standard RTO mechanism would allow, thereby reducing retransmission latency without adding complex detection algorithms.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent enhances feedback mechanisms by utilizing SACK segments and probe packet responses to provide real-time information about packet reception status. This feedback allows the sender to make informed retransmission decisions based on actual network conditions rather than relying solely on timeout-based mechanisms.

Inventive Principle:

Principle #23Feedback

3Measurement precision

If modified TCP stacks are deployed to improve packet loss detection, then detection accuracy improves, but deployment difficulty and cost increase

Engineering Contradiction:

Improvepacket loss detection accuracyVSAvoiddeployment ease

Core Design Contradiction:

Measurement precisionVSEase of manufacture

Solution Approach 1:

The patent achieves universal applicability by designing a solution that works with existing TCP stacks. The method uses standard TCP components (ACK packets, SACK segments, probe packets) that are already supported by most TCP implementations, allowing the solution to be deployed without requiring modified TCP stacks on either end of the connection.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Solution Approach 2:

The patent enables self-service by allowing the system to detect and respond to packet loss using existing TCP mechanisms. The sender can autonomously determine packet loss by analyzing SACK segments and probe packet responses without requiring external network management systems or modified protocol stacks, making deployment straightforward.

Inventive Principle:

Principle #25Self-service

4Productivity

If TCP window is scaled up for long fat pipe networks, then bandwidth utilization improves, but packet loss detection capability deteriorates

Engineering Contradiction:

Improvebandwidth utilizationVSAvoidpacket loss detection capability

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The patent replaces the mechanical packet-by-packet acknowledgment system with a more efficient information-gathering mechanism using SACK segments and probe packets. This substitution allows the system to maintain large TCP windows for high bandwidth utilization while simultaneously achieving accurate packet loss detection through selective acknowledgment information and targeted probing.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS9246636B2Method for detecting TCP packet losses and expediting packet retransmission

Publication Date: 2016.01.26 APPEX NETWORKS CORP

AI summary

A computer-implemented method for detecting TCP packet losses and expediting packet retransmission is disclosed. The method includes assigning to each packet transmitted or retransmitted a sequential order number, determining from an ACK packet or SACK segment thereof the sequential order number of a received packet, recording the determined sequential order number as a highest received order number in case the determined sequential order number is greater than a current highest received order number, and retransmitting all unacknowledged packets having sequential order numbers less than the highest received order number.