Optimizing Data Rates via Path Characterization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current congestion control methods in communication networks, such as TCP and MPTCP, often lead to inefficient data transfer due to a reactive approach in setting the congestion window (CWND), resulting in sawtooth-like bandwidth fluctuations and potential network overload or underutilization.
Innovation Solution
A method to calculate and set an optimal data rate for communication paths based on transport characteristics, such as bandwidth, latency, and stability, using a data assignment unit and multipath scheduler, which avoids the iterative increase and decrease of CWND, allowing for efficient data transfer without network overload.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the congestion window (CWND) is increased to transfer more data packets, then the data transfer rate is improved, but network overload and packet loss occur
Solution Approach 1:
The patent applies preliminary action by proactively determining path characteristics (bandwidth, RTT, stability) before initiating data transmission. The system calculates an optimal initial CWND based on these pre-acquired characteristics, avoiding the traditional reactive approach where CWND is only adjusted after congestion occurs. This allows the sender to start with an optimized window size that prevents both underutilization and overload from the beginning.
Solution Approach 2:
The patent implements feedback mechanisms by continuously monitoring path characteristics and using this information to dynamically adjust the CWND. The system receives feedback about network conditions (bandwidth availability, RTT variations, packet loss) and uses this feedback to optimize the data rate, creating a closed-loop control system that adapts to changing network conditions while maintaining stability.
2Reliability
If the congestion window (CWND) is decreased to avoid network overload, then network stability is maintained, but data transfer efficiency is reduced
Solution Approach 1:
The system performs preliminary characterization of the communication path before transmission, determining the maximum stable data rate based on bandwidth, RTT, and stability metrics. This allows the system to set an initial CWND that is optimized for the specific path conditions, avoiding the conservative approach of using small CWND values that prevent overload but also limit throughput potential.
Solution Approach 2:
The patent changes the parameters used for CWND determination from reactive congestion indicators (packet loss, retransmissions) to proactive path characteristics (bandwidth, RTT, stability). By using these different parameters that better represent the true capacity of the network path, the system can set higher CWND values that maintain stability while improving throughput.
3Reliability
If reactive congestion control is used to adjust CWND based on packet loss, then network overload is prevented, but sawtooth-like bandwidth fluctuations occur
Solution Approach 1:
The patent applies preliminary action by determining path characteristics before transmission and using these characteristics to set an optimized initial CWND. This eliminates the need for the traditional slow-start phase where CWND grows exponentially and causes sawtooth fluctuations. The system starts at or near the optimal data rate from the beginning, providing stable bandwidth utilization without the characteristic oscillations of reactive congestion control.
Solution Approach 2:
The patent substitutes the mechanical reactive adjustment mechanism (increasing CWND until packet loss occurs, then reducing it) with a calculated approach based on path characteristics. Instead of relying on the trial-and-error mechanical process of increasing and decreasing CWND, the system uses a formulaic approach that directly calculates the optimal CWND based on measured path properties, eliminating the sawtooth pattern.
4Productivity
If path measurements are used to determine transport capabilities, then data transfer optimization is improved, but system complexity increases
Solution Approach 1:
The patent applies universality by using the same path measurement infrastructure that already exists for other purposes (congestion control, routing decisions) to also determine the optimal data rate. The system leverages existing feedback loops and measurement mechanisms in multipath protocols to gather path characteristics, making the measurement process multi-functional rather than requiring separate dedicated measurement systems.
Solution Approach 2:
The system implements self-service by having the sender autonomously perform path characterization and calculate the optimal data rate without requiring external control or complex centralized coordination. The sender uses its own measurements and calculations to determine the optimal CWND, eliminating the need for additional control infrastructure or complex interaction with network elements.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Techniques to set a data rate for the transmission of data packets over a communication path of a communications system, wherein the method comprises the following steps: • Providing at least a first communication path P1 between a sender and a receiver; • Providing data packet transport characteristics C1 of the at least first communication path P1; • Providing a data assignment unit, wherein a data assignment algorithm is implemented on the data assignment unit that calculates a data rate R1 that shall be send by the sender to the receiver via the at least a first communication path P1 based on the data packets transport characteristics C1 of the at least first communication path P1.