Predictive Acknowledgments for Network Data Flow Management

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current network data flow management systems are inefficient due to the slow-start algorithm in TCP, which leads to prolonged load times for web pages and other data transfers, especially in high-latency networks, and are vulnerable to misbehaving receivers that can cause congestion and denial-of-service attacks.

Innovation Solution

A network data flow management system that employs predictive acknowledgments, using timers and client properties to optimize data flow by sending ACK packets ahead of actual receipt, thereby bypassing the slow-start phase and improving data transfer speed, and includes mechanisms to handle network congestion and packet loss.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If conventional TCP slow-start algorithm is used, then network congestion is avoided through gradual rate increase, but data transfer speed is reduced and load time is prolonged

Engineering Contradiction:
Improvenetwork congestion avoidanceVSAvoiddata transfer speed
Core Design Contradiction:
ReliabilityVSSpeed

Solution Approach 1:

The system performs preliminary actions by sending predictive acknowledgments before the receiver actually sends them, and by pre-increasing the TCP window size beyond what the slow-start algorithm would normally allow. This anticipatory approach bypasses the gradual ramp-up phase, immediately establishing higher data transfer rates while maintaining congestion avoidance through predictive mechanisms.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent introduces an intermediary component that sits between the sender and receiver, managing TCP connections and window sizes. This intermediary can send predictive acknowledgments and adjust window sizes independently of the standard TCP slow-start mechanism, thereby decoupling congestion control from data transfer rate optimization.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Speed

If TCP window size is increased to improve data transfer rate, then data transfer speed improves, but network congestion control becomes less effective

Engineering Contradiction:
Improvedata transfer rateVSAvoidcongestion control effectiveness
Core Design Contradiction:
SpeedVSReliability

Solution Approach 1:

The system implements feedback mechanisms where predictive acknowledgments are sent based on expected data arrival times and network conditions. The TCP window size is dynamically adjusted based on feedback from network performance monitoring, allowing the system to maintain high transfer rates while responding to actual congestion conditions rather than relying solely on slow-start's conservative approach.

Inventive Principle:
Principle #23Feedback

3Productivity

If multiple small TCP connections are used for web page loading, then all content elements can be loaded in parallel, but cumulative latency and jitter dominate the load time calculation

Engineering Contradiction:
Improveparallel content loadingVSAvoidcumulative latency and jitter
Core Design Contradiction:
ProductivityVSLoss of time

Solution Approach 1:

The patent merges multiple small TCP connections into fewer, optimized connections by using predictive acknowledgments and adjusted window sizes that allow larger data segments to be transmitted more efficiently. This reduces the total number of connection handshakes and latency events while maintaining the ability to load multiple content elements in parallel.

Inventive Principle:
Principle #5Merging (Combining)

Data Source

PatentEP2424177B1Method and system for network data flow management
Publication Date: 2016.01.13 SANDVINE CORP
  • EP2424177B1 patent drawingFigure 1
  • EP2424177B1 patent drawingFigure 2
  • EP2424177B1 patent drawingFigure 3

AI summary

A network data flow management device including: a working packet module configured to store data related to a data flow; at least one network interface connected to the working packet module for receiving and sending data related to data flows; and a processor configured to: receive data from the working packet module and determine if the data flow matches a predetermined protocol; and if the data flow matches the predetermined protocol, send predictive acknowledgments based on the data flow via the network interface. A method for network data flow management including: monitoring a data flow for a predetermined protocol; if the data flow matches the predetermined protocol, saving data relating to the data flow; and creating and sending a predictive acknowledgement packet related to the data flow.