GB2177280A

GB2177280A - Digital data transmission

Info

Publication number: GB2177280A
Application number: GB08611375A
Authority: GB
Inventors: John Philip Chambers
Original assignee: British Broadcasting Corp
Current assignee: British Broadcasting Corp
Priority date: 1985-05-09
Filing date: 1986-05-09
Publication date: 1987-01-14
Also published as: GB8611375D0; GB8511804D0; GB2177280B

Abstract

Data in packets is transmitted with a continuity index to enable detection of a lost packet and with parity information to enable detection of errors within the packet. For transmission the continuity index and parity information are combined, e.g. by modulo-2 addition, so that at the receiver a correctly-received packet gives rise to a parity check output not of zero but indicating instead the continuity index. Departures from the continuity index sequence can be detected to indicate whether a packet has been lost or whether there is a single perturbation of the sequence indicating a parity error. By combining the continuity index and parity information for transmission a reduction in data rate can be obtained. <IMAGE>

Description

SPECIFICATION Digital data transmission This invention relates to the transmission of data in packets. The term transmission is used in this specification for convenience to include not only conveying from one point to another but also to cover processing or storage, and the terms transmitter and receiver should be interpreted accordingly.

It is common practice in telecommunications systems to use a cyclic redundancy check (CRC) or block check character (BCC) to provide an indication that a block of data has been received correctly and completely. Such techniques are described in, for example, the "Handbook of Data Communications" published in 1982 by the National Computer Centre.

It is also common practice in a packet multiplex data system for consecutive blocks of data intended for a common destination to include a "continuity indicator" (Cl), whose content changes in a regular and predictable way (such as a simple increasing binary count) from one block to the next. Use of the Cl information in a receiving terminal equipment allows the detection of the fact that one or more packets are absent from the sequence, or, that an unwanted packet has appeared in the sequence.

According to this invention there is provided a method of transmitting data in which the data is transmitted in packets, each packet having associated with it a continuity indicator identifying that packet and which changes in a predetermined sequence from one packet to the next, and the data being protected by error checking codes derived from the data at the transmitter, and at the receiver the error checking codes are combined with the received data to produce parity codes which have a predetermined value in the absence of errors, and the continuity indicator is monitored to detect the absence of a packet or the insertion of a spurious packet, characterised in that the continuity indicator is used at the transmitter in forming the error checking codes and is not separately transmitted, and at the receiver the parity codes represent the continuity indicators modified by any errors introduced in transmission, the parity codes being monitored for departures from a steady sequence representative of the continuity indicators. In this way the space normally used by the continuity indicator is made available for transmission. The invention also provides a transmitter and a receiver for use in the method.

Examples of the invention will now be described with reference to its implementation in the United Kingdom teletext system, in which spare lines in a broadcast television system are used to carry additional information in the form of digital data.

It is proposed that a system for auxiliary data signalling use packets of data such that each packet occupies a television data line.

The "Magazine and Row Address Group" (MRAG) of that line (see 2.1.1 of September 1976 "Broadcast Teletext Specification") would be chosen so as not to interfere with the existing services and their proposed extension. Fifteen such MRAG's are currently available, corresponding to rows 30 or 31 of any of the magazines 1-8 with the exception of magazine 8, row 30 whose function is already allocated. The remaining 40 eight-bit bytes of each such data packet would be divided between address information, error protection (detection or correction), continuity indication, control information and the "message" data itself. If more than one division of the 40-byte packet is envisaged, the control information would signal which format is in use. It could also signal which method of error protection is in use.It would clearly be advantageous to send the control information in a fixed position at the start of the 40-byte packet, as it controls the interpretation of the remainder.

A block check character (BCC) is usually made as a longitudinal parity check of the data block. This means that bit n of the BCC byte is the result of a defined (odd or even) parity check on the bits n of all the other bytes defined to be part of the check. In accordance with this invention another byte which is not actually sent as part of the packet would be included in the BCC calculation. This "Cl" byte would change from one packet to the next (of the same address) in a predictable way.It could simply be complemented on alternate packets (e.g. 00001111 and 11110000), it could follow a sequence of "counting" where consecutive bytes had a large "Hamming distance" (e.g. the four-byte sequence 0000000,00011111, 11111000, 11100111), or it could be a longer sequence with such suitable properties, or even a simple increasing or decreasing binary count modulo256. In the receiver the BCC checking operation, instead of producing a typical "all correct" message (e.g. 00000000), would reproduce the chosen sequence of CI bytes. It can safely be assumed that most of the time the packets are received without error, so that receiver circuitry could readily "lock on" to the incoming sequence with a "flywheel" generator which could be used to imitate and so remove the effect of the added CI byte, allowing normal interpretation of the BCC.

In the case of a cyclic redundancy check (CRC) a check word, typically of 16 bits, is produced as a function (polynomial division modulo-2) of a defined string of bits to be checked. A receiver would reproduce this process to give, typically, an "all correct" message of 16 zeroes. The CRC could be modified by a Cl sequence in many ways, but they are all equivalent to adding "bit-wise modulo2" a sequence of 16-bit words to the CRC.

Again it can be assumed that most of the time the packets are received without error, so the receiver circuitry could "lock" to the sequence and reproduce it. The options for the sequence are as before, but the (normally) greater length of a CRC allows the possibility of a longer sequence, even a full 16-bit count of 65,536 packets. Such long Cl sequences may be of advantage in applications where conditional access scrambling is in use.

In either event the CI byte or sequence does not have to be sent separately and the space it would have required can be used by other data. Thus the advantage of the technique proposed here is increased efficiency, i.e., more information can be sent per packet. This advantage is greater where a long Cl sequence is required, such as for conditional access. Disadvantages are that buffer storage (which, of course, may already be available) is required in the receiver if the "first ever packet is to be reliably interpreted (this may not be a requirement where there is a "permanent" service channel in operation), and a modest addition to the receiver complexity.

When an error is detected at the receiver, a determination can be made as to whether the sequence of messages before and after the apparent error indicates loss of a packet. In the steady state the data packets of a particular service are received without error, so the parity check results in the predictable sequence. Prima facie a packet whose parity check does not continue the sequence is a corrupted packet. More often than not the next packet (of the particular sequence: there may be more than one) will, when tested, he seen to continue the sequence after 'jumping over' one missing state. This effectively confirms that one and only one packet was corrupted, and also allows the corrected parity check of the corrupted packet to be inferred by subtracting the effect of the missed state.

Provided there is no actual loss or gain in packets this process can be conducted over any pattern of corrupted packets once the 'flywheel' is 'locked' to the incoming continuity index sequence, and providing there is sufficient buffer storage at the receiver. An escape route is provided to allow the process to reacquire lock after a packet loss or gain, which is indicated by a series of checks which is consistent with the sequence but not in the expected phase relation.

In a modification to the CRC system described above, Cl word takes the form of a single 8-bit byte. In order to combine this with the 16-bit CRC word, the Cl word is repeated to make another 16-word. This duplication then makes it possible to distinguish between a 'good' and 'bad' packet even on first reception, i.e. without a flywheel, with high, though not complete, confidence. This is possible because the 'good' packet will show the repeated pattern of the Cl byte whereas only 1 in 256 of 'bad' packets will happen to show this pattern.

A block schematic diagram of a transmitter and a receiver embodying the invention is shown in the sole figure of the accompanying drawing.

A conventional data channel has at the transmitter 10 a data transmission circuit 12, and a parity generator 14 for transmitting parity information. At the receiver 20 the data is received in circuit 22 and parity information in circuit 24. A parity checking circuit 26 compares the data and parity and initiates corrective action in the absence of correspondence between them.

In accordance with this invention, the transmitter includes a parity modifier 30 which modifies the parity information in accordance with information from a pattern generator 32 which takes an input of the continuity sequence generator 34. At the receiver a pattern recogniser 40 looks at the output of the parity generator and applies it to a sequence regenerator 42 which regenerates the Cl sequence.

This is applied to a pattern regenerator 44 which controls a parity check modifier 46 to inhibit an output if message is correctly decoded or if a continuity error alone is detected, but initiates corrective action if a parity error is detected.

The circuit may also signal a continuity index error to other parts of the receiver.

Variations in the detailed manner of operation are possible, and in particular the operation may be implemented by a computer or micro-processor, in which case the block diagram of the figure should be considered as an information flow chart describing the relevant software.

Claims

1. A method of transmitting data in which the data is transmitted in packets, each packet having associated with it a continuity indicator identifying that packet and which changes in a predetermined sequence from one packet to the next, and the data being protected by error checking codes derived from the data at the transmitter, and at the receiver the error checking codes are combined with the received data to produce parity codes which have a predetermined value in the absence of errors, and the continuity indicator is monitored to detect the absence of a packet or the insertion of a spurious packet, characterised in that the continuity indicator is used at the transmitter in forming the error checking codes and is not separately transmitted, and at the receiver the parity codes represent the continuity indicators modified by any errors introduced in transmission, the parity codes being monitored for departures from a steady sequence representa tive of the continuity indicators.

2. A method according to claim 1, in which the error checking codes comprise block check characters.

3. A method according to claim 1, in which the error checking codes comprise a cyclic redundancy check word.

4. A method according to claim 3, in which each error checking code comprises a repeated cyclic redundancy check word.

5. A transmitter for transmitting data packets, comprising means for receiving a sequence of data packets each having associated with it a continuity indicator identifying that packet and which changes in a predetermined sequence from one packet to the next, means for generating error checking codes derived from the data to protect the data, and means for modifying the error checking codes in accordance with the continuity indicator to transmit the continuity indicator and error checking codes simultaneously.

6. A receiver for receiving data packets transmitted by the transmitter of claim 5, comprising means for receiving and detecting the transmitted data and modified error checking codes means for combining the error checking codes with the received data to produce parity codes which have a predetermined value in the absence of errors, the parity codes representing the continuity indicators modified by any errors introduced in transmission, and means for monitoring the parity codes for departures from a steady sequence representative of the continuity indicators.

7. A receiver according to claim 6, including flywheel means for locking to the sequence of continuity indicators to detect departures from the predetermined sequence.

8. A method of transmitting data packets substantially as herein described with reference to the drawing.