Adaptive Bitrate Manager for TCP Flow Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Network congestion control in communication networks, particularly in mobile telecommunications systems, is complicated by the termination of different protocols at various locations, leading to inefficient congestion management and challenges in providing Quality of Service (QoS) bit-rate differentiation among data connections.
Innovation Solution
Implementing a method in network nodes to detect congestion by monitoring sequence numbers and selecting data connections for bit-rate reduction based on time-integrated differences between experienced and targeted bit-rates, with the option to use relative bit-rate weights for Quality-of-Service differentiation, and initiating bit-rate reduction by discarding data packets to manage congestion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If HSPA Flow Control is used to manage congestion in the transport network, then bandwidth sharing among traffic flows is improved, but the ability to support QoS bit-rate differentiation among data connections deteriorates
Solution Approach 1:
The invention segments the congestion control mechanism into two independent parts: HSPA Flow Control handles bandwidth sharing at the radio interface level, while a new TCP-level congestion control mechanism handles QoS bit-rate differentiation at the transport network level. This segmentation allows each mechanism to operate independently without interfering with the other, resolving the contradiction between bandwidth sharing and QoS differentiation.
Solution Approach 2:
The invention introduces an intermediary mechanism - the TCP congestion control at the network node - that mediates between the HSPA Flow Control and the underlying network congestion. This intermediary translates TCP congestion signals into appropriate bit-rate adjustments for different data connections while preserving QoS weights, enabling both bandwidth sharing and QoS differentiation to coexist.
2Reliability
If TCP is used for congestion control in the radio access network, then congestion avoidance is improved, but effectiveness deteriorates due to lower layer retransmissions hiding congestion situations
Solution Approach 1:
The invention places TCP congestion control at the network node (Serving Radio Network Controller) as an intermediary layer between the radio interface and the core network. This position allows the TCP mechanism to directly observe transport network congestion without being obscured by radio layer retransmissions, as the TCP controller operates at a higher protocol level where radio retransmissions are transparent.
Solution Approach 2:
The invention moves TCP congestion control from the traditional end-host dimension to the network-node dimension, creating a new control plane. This dimensional shift allows congestion detection to occur at the network node where transport network congestion is directly observable, separating the detection function from the radio access layer where retransmissions occur.
3Productivity
If a radio base station handles multiple data connections with varying TCP sessions, then network capacity is improved, but congestion control complexity increases
Solution Approach 1:
The invention implements a universal TCP congestion control mechanism at the network node that handles all data connections and TCP sessions uniformly. Instead of implementing separate congestion control logic for each data connection or TCP session, the network node applies a single TCP congestion control algorithm that aggregates traffic from multiple connections and sessions, simplifying the control complexity while maintaining high network capacity.
Solution Approach 2:
The invention merges the congestion control functions for multiple data connections and TCP sessions into a single unified TCP congestion control mechanism at the network node. By combining the control of all connections through one TCP-level controller, the system manages high network capacity without proportionally increasing control complexity, as the merged controller handles aggregate traffic rather than individual connections separately.
Data Source
AI summary
A method is provided for congestion control in a network node (114; 510; 900) of a communication network. The network node handles a plurality of data connections for conveying data packets travelling between a core side and a terminal side of the communication network. The method is performed in the network node and comprises detecting (210) a condition indicative of a congestion for any of the data connections, selecting (220), among the data connections, a data connection (RABdrop) for which a bit-rate reduction is to be caused, and determining (230) a number (Xi) of data packets to be discarded for the selected data connection. The determined number depends on a time-integrated difference between experienced and targeted bit-rates for the selected data connection. Then, the reduction of the bit-rate is initiated (240) for the selected data connection by causing discarding of the determined number of data packets.