Method and arrangement for tcp flow control

Abstract

A communication entity is located at one end of a high latency communication link to support Transmission Control Protocol (TCP) communications between a first transceiver entity and a second transceiver entity. The communication entity comprises proxy logic arranged to inspect a received segment and, in response to identifying that the received segment does not contain data, the proxy logic transparently allows a plurality of synchronising segments to pass between the first transceiver entity and the second transceiver entity through the proxy logic such that the proxy logic generates locally at least one acknowledgement message based on the synchronising segments and the inspected received segment.

Description

technical field [0001] The present invention relates to TCP (Transmission Control Protocol) flow control, in particular (but not exclusively) to TCP flow control in wireless communication systems. Background technique [0002] The Transmission Control Protocol (TCP) is a protocol of the well-known TCP / IP (Internet Protocol) family of communication protocols (see [RFC 793] for a complete description of the TCP protocol). TCP provides a connection-oriented reliable byte stream service, and was originally designed for wired networks with very low error rates. [0003] TCP uses a window-based flow control mechanism that limits the amount of unacknowledged data sent by a sender. This restriction is based on the minimum of the congestion window (cwnd) and the advertisement window (awnd). The congestion window is controlled by the sender of the TCP data and is based on well known slow start and congestion avoidance mechanisms. Slow start is used to slowly increase the amount of ...

AI Technical Summary

Problems solved by technology

Thus, as achievable UL rates get higher and higher, in many cases the ULTCP rate will be limited by the server "awnd" rather than by prevailing air interface conditions

[0009] Another problem associated with simply increasing awnd in a wireless system with high latency is that the DL TCP rate is severely affected due to the loss of any data packets due to congestion in the core network, and due to the associated high latency , so the recovery time is longer

[0017] A potential problem with this situation in known architectures is that an RLC Move Receive Window (MRW) command or reset can occur while the connection is maintained

Furthermore, with completely separated TCP connections at UE side and network side, it is difficult to provide an alternative retransmission scheme to overcome this problem

[0018] Also, in flows without end-to-end TCP functionality, potential problems occur when the UE changes cells due to the fact that different (i.e., out-of-sync) TCP sessions take place on the PC side and on the network side

Images