ENHANCED LARGE DATA TRANSMISSIONS AND CATASTROPHIC CONGESTION AVOIDANCE OVER IPv6 TCP/IP NETWORKS

• US 20160269294A1
• Filed: 05/23/2016
• Published: 09/15/2016
• Est. Priority Date: 09/19/2013
• Status: Active Grant

• Current Assignee: Connectivity Systems Inc.
• Original Assignee: Connectivity Systems Inc.
• Inventors: Rankin, John
• Granted Patent: US 10,432,529 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: H04L 47/12 Avoiding congestion; Recovering from congestion
• CPC Class Codes: H04L 47/193 at the transport layer, e.g. TCP related

Abstract

A system and method of reducing network congestion whereby significant improvement can be gained by the elimination of catastrophic network congestion failure within a TCP/IP network. Exemplary embodiments are implemented by an improved process for dealing with data loss related to IPv6 fragmentation, in which a ICMPv6 error message is transmitted by a destination node containing a bitmap of the unreceived datagram fragments. The sending node receives the error message and prepares only those unreceived fragments for transmission.

Claims

1. A method of congestion avoidance in a packet switched network comprising the steps of:

receiving, at a destination host, one or more received fragments of a fragmented datagram, wherein the fragmented datagram has been fragmented into a total number of fragments comprising the one or more received fragments and one or more lost fragments;

constructing a fragmentation data loss error message comprising;

an identification value identifying the fragmented datagram; and

a bitmap identifying the one or more lost fragments; and

transmitting the fragmentation data loss error message to a source host.

2. A method of congestion avoidance in a packet switched network comprising the steps of:

initializing, at a destination host, a transmission timer;

receiving, at the destination host, a first received fragment of a fragmented datagram;

starting the transmission timer upon receipt of the first received fragment;

receiving, at the destination host, at least one received fragment of a fragmented datagram, wherein the at least one received fragment includes the first received fragment, and wherein the fragmented datagram has been fragmented into a total number of fragments comprising the at least one received fragment and at least one lost fragment;

incrementing the transmission timer to a transmission time that is less than a transmission timeout time

constructing, at a retransmission time that is less than a transmission timeout time, a fragmentation data loss error message comprising;

an identification value identifying the fragmented datagram; and

a bitmap identifying the at least one lost fragment;

transmitting the fragmentation data loss error message to a source host.

3. A method of congestion avoidance in a packet switched network comprising the steps of:

assembling, in an internet layer daemon of a first remote host, a datagram to be transmitted over the network to a second remote host, the datagram comprising;

an address header; and

a payload;

fragmenting the datagram into a plurality of fragments, wherein each fragment in the plurality of fragments comprises;

the address header;

a fragmentation header comprising an identification value identifying the datagram and a fragment offset value; and

a payload fragment,

wherein the payload fragments for each fragment in the plurality of fragments together comprise the payload;

transmitting each fragment in the plurality of fragments to a second remote host;

receiving, a fragmentation data loss error message comprising;

the identification value identifying the datagram; and

a bitmap identifying one or more lost fragments, wherein each lost fragment in the one or more lost fragments corresponds to a fragment in the plurality of fragments; and

retransmitting the one or more lost fragments to the second remote host.

4. The method of claim 3, wherein the step of fragmenting the datagram into a plurality of fragments further comprises the steps of:

dividing the payload into n payload fragments, wherein the first n−

1 payload fragments are equal-length multiples of 8 octets in length; and

appending each of the payload fragments to the address header and the fragmentation header.

5. The method of claim 4, wherein the fragmentation header further comprises:

a more-fragments value set to 1 for the first n−

1 fragments in the plurality of fragments; and

a more-fragments value set to 0 for the last fragment in the plurality of fragments.