Method of error recovery in a data communication system

• US 5,410,536 A
• Filed: 09/25/1992
• Issued: 04/25/1995
• Est. Priority Date: 12/04/1990
• Status: Expired due to Term

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Grainger, Bernard J., Judd, Ian D., Cockburn, Gordon J., Shah, Vinay V.
• Primary Examiner(s): Chin, Wellington
• Application Number:
  US07/916,009
• Time in Patent Office: 942 Days
• Field of Search: 370/13, 370/17, 370/80, 370/60.1, 370/94.1, 370/94.2, 371/32, 371/33, 371/35
• US Class Current: 370/216
• CPC Class Codes: H04L 1/16 in which the return channel carries supervisory signals, e.g. repetition request signals
• CPC Class Codes: H04L 1/1806 Go-back-N protocols
• CPC Class Codes: H04L 1/1848 Time-out mechanisms
• CPC Class Codes: H04L 1/1874 Buffer management

Abstract

Described is a method of error recovery in a data communication system of the kind comprising two nodes connected by a serial link and wherein data is transmitted between the nodes in the form of packets of a predefined format. Each node receives data over an inbound line and transmits data over an outbound line. When an error is detected, both nodes enter a link check state, invoke a Link Error Recovery Procedure (ERP) and exchange status by means of Link Resets. Error recovery is performed separately for each line. Each node is responsible for recovering packets that were lost on its outbound line. In normal operation of the link, the transmitter does not reuse a packet buffer until it has received a response from the connected node indicating that the packet was correctly received. Therefore when an error occurs, the affected packets are still available for retransmission.

Claims

1. A data communication system including

two nodes connected by a serial link over which data is transferred between the nodes in packets of a predefined format,

error detection means in each node for detecting errors,

and transmission error recovery means in at least one of said nodes responsive to detection of an error by the error detection means of that node to cause that node to send an error message to the other node including a sequence number indicative of the last packet received by said at least one node, and to receipt of an error message from the other node to cause said at least one node to send its error message to the other node,

each node being arranged to determine from the error message from the other node the number of packets, if any, that were not correctly received by the other node and to retransmit the missing packets.

2. A system as claimed in claim 1 in which each node includes multiple packet buffers for storing packets to be transmitted over the link or received on the link.

3. A data communication system including

two nodes connected by a serial link over which data is transferred between the nodes in packets of a predefined format,

error detection means in each node for detecting errors,

and transmission error recovery means in at least one of said nodes responsive either to detection of an error by the error detection means of that node to cause that node to send an error message to the other node including a sequence number indicative of the last packet received by said at least one node, or to receipt of an error message from the other node to cause said at least one node to send its error message to the other node,

each node being arranged to determine from the error message from the other node the number of packets, if any, that were not correctly received by the other node and to retransmit the missing packets,

each node includes multiple packet buffers for storing packets to be transmitted over the link or received on the link, and

each node including

means for transmitting an acknowledgement message to the other node upon receipt of a packet,

transmit pointer means for indicating from which packet buffer the last packet was transmitted, and

acknowledgement pointer means for indicating from which packet buffer the last packet to have been acknowledged was transmitted.

4. A system as claimed in claim 3 including means for calculating from the transmit pointer means, the number of packets sent which have not been acknowledged,

and means for comparing this number with the number of packets to be retransmitted,

thereby to identify packet buffers containing packets not required to be retransmitted.

5. An error recovery method for use in a data communication system, said data communication system comprising two nodes connected by a serial link, data being transferred between the nodes in packets of a predefined format, said method comprising the steps of:

monitoring the system for errors with each node;

sending a first message with a first node to a second node, in response to detecting an error by said first node, said first message including a sequence number indicative of the last packet received by said first node;

receiving said first message with said second node;

sending a second message with said second node to said first node, in response to said step of receiving said first message, said second message including a sequence number indicative of the last packet received by said second node;

receiving said second message with said first node;

determining with each node, from the respective one of said first and second messages received from the other node, the number of packets, if any, that were not correctly received by the other node; and

retransmitting a packet if said determining step determines that the packet was not correctly received.

6. The method of claim 5, further comprising the steps of:

sending an acknowledgement with each node to the other node upon receipt of a packet;

calculating, with each node, the number of packets it has sent which have not been acknowledged; and

comparing, with each node, the number produced by said calculating step with the number of packets to be retransmitted, and discarding surplus packets when the number of packets to be retransmitted is greater than the number produced by said calculating step.

7. The method of claim 6, wherein each said packet comprises a plurality of predefined fields, each field consisting of one or more multibit data frames, and wherein the flow of data packets is controlled by means of multiple bit control frames distinguishable from the data frames, the control frames being transmissible independently of the data packets.

8. The method of claim 7, wherein the acknowledgment is one of the control frames.

9. The method of claim 5, wherein said step of retransmitting a packet is not attempted upon the detection of certain predetermined errors.

10. The method of claim 5, wherein each said packet comprises a plurality of predefined fields, each field consisting of one or more multibit data frames, and wherein the flow of data packets is controlled by means of multiple bit control frames distinguishable from the data frames, the control frames being transmissible independently of the data packets.