GB2094107A - Digital code for line transmission - Google Patents

Abstract

A method of transmitting information encoded into a binary digit stream in which successive groups of N binary digits are translated into corresponding groups of M binary digits, where M > N, the translation being accomplished according to at least the following rules:- a) only M-bit groups having a disparity not exceeding +/- (M/2 - 2) are allowable; b) no M-bit group having more than n consecutive ones or zeros at the beginning or end is allowable, where n =M/2; c) no zero disparity M-bit group having more than n-2 consecutive ones (zeros) at the end and n-1 consecutive ones (zeros) at the beginning is allowable. The resulting M-bit code will have the unique features of 1) peak digital sum in every group occurs at the same digit position; 2) maximum run of like digits between successive groups only occurs with the last digit of the run always in the same digit position. f

Description

SPECIFICATION Digital code for line transmission This invention relates to a digital code for line transmission in telecommunications systems.

Binary encoding of analogue signals is an established technique for modern telecommunications systems.

However, with the advent of the latest technology, such as the widespread use of coaxial cable and optical fiber line transmission systems higher and higher digital transmission speeds are being attained. Currently systems are being designed, for example, in which information is encoded for transmission at 140Mbit/sec (a "bit" being defined as one binary encoded digit, i.e.

In designing a digital line system it is important to provide for continued monitoring of performance by detection of errors produced by the system while in service. Detection of errors at the system terminal can use relatively elaborate techniques with no need to economise on power consumption.

The errors may be caused at any location along the route, e.g. by a faulty dependent repeater. The location of the source of errors for maintenance purposes can be accomplished by using at each dependent repeater a special low power consumption error detector, information from which is carried over a telemetry channel to a terminal. The location is performed most rapidly if that detector also operates whilst the system is in service; this avoids the necessity to take the system out of service when the error rate is not high enough to justify this.

One well known method of achieving in-service error detection is to use a line code which incorporates as a feature of the code itself error detection (and in some cases error correction) capabilities. A simple example of this is the use of parity control, i.e. the number of bits of a given significance must always be odd (or even) for a given digit word or block.

A recent development is the use of specially designed line codes which are different from the original information code. A typical example of this is to be found in British patent No. 1,156,279 (D.B. Waters-l), which discloses the concept of translating binary encoded information into a ternary encoded signals for the purposes of line transmission. Specifically, successive groups of N binary digits are translated into groups of M ternary digits, where M < N, such that the cumulative disparity of the ternary encoded signals does not exceed a predetermined value of either polarity. The ternary bit stream has other identifiable characteristics.

For example, the longest possible block of positive or negative bits without a transition is fixed, as is the longest possible block of consecutive zeroes. This means that D.C. balance is obtained together with adequate timing content for regeneration. At the receiver the ternary groups are translated to binary independently. Because of the nature of the original translation from binary to ternary, if the accumulated disparity count in the transmitter causes ternary codes of the wrong disparity to be transmitted no digital errors are introduced in the receiver translation. Digital errors on the line only affect the actual characters mutilated, since there is no disparity count or inversion in the receiver to be upset.

A typical example of such codes is that known as 4B 3T, in which successive groups of 4 binary digits are translated into groups of 3 ternary bits. A later version of this code is known as 6B 4T.

A more recent development is the class of codes which can be generally referred to as NEMB, where N < M. Examples are 5B6B, as disclosed in British patent No. 1,250,908 and 7B8B as disclosed in British patent No. 1,540,617. The translation from NB to MB can provide signals having lower disparity that the original NB signals. The code disclosed in patent 1,540,617 is one in which successive groups of N binary digits are recoded into groups of M binary digits, where M is greater than N and both are positive integers, the recoding being arranged so that some but not all of the groups of M digits have minimum disparity and successive groups of M having non-zero disparity have disparities of opposite signs. The specific code exemplified is a 7B8B code in which the 8B groups beginning or ending with more than three like digits are excluded from use.

7B-8B coding is a good compromise between complexity and efficiency with regard to regenerability.

Additionally, there is adequate redundancy to satisfy the system requirements of error detection at repeaters, error detection at terminals, short word alignment time and the provision of auxiliary channels for such uses as remote alarms and end to end Engineering Order Wire.

In determining disparity in balanced binary codes a binary one is weighted +1/2 and a binary zero is weighted - 1/2 in line with CCITT vocabulary in Rec. G702. In the following text the number of consecutive ones or zeros at the beginning or end of a code word is designated "n". If it is necessary to distinguish between the beginning and the end of a code word these are designated "nb" or "ne" respectively.

In the interest of keeping the digital sum variation (DSV) small and limiting the length of consecutive ones and zeros preference is given to code words of lower disparity and smaller value of n. In the use of 7B-8B encoding it is necessary to use zero disparity, + 1 disparity and + 2 disparity codes words to make up the required 128 combinations. In all there are 70 possible words with zero disparity, 56 words with +1 disparity and 28 words with + disparity.

Figure 1 shows a state-transition diagram for a 7B-8B code using words of zero, +1 and +2 disparity.

Because the zero disparity words have an even number of marks they do not change the end of word disparity and hence the end of word parity can be associated with the end of word digital sum (DS) on the state diagram.

From Figure 1 it can be seen that the max and min end of word digital sums are 11/2 and -11/2 respectively.

With n (zero disparity) 63 the max and min intra-word digital sums occur with zero disparity words as follows: DS=-3 DS = +3 00010111 11101000 00011011 11100100 00011101 11100010 00011110 11100001 00100111 11011000 01000111 10111000 10000111 01111000 The peak digital sum can occur on either word digit 3 or 5.

If the 8 zero disparity words with 3 or more consecutive ones or zeroes in the beginning of the word are rejected it will be seen that the peak DSV of 3 can only occur at word digit 5. In addition, the maximum run of like digits, namely 7, can only occur with the last digit of the run in the word digit 4 pOSitiOn.

Therefore, according to the present invention there is provided a method of transmitting information encoded into a binary digit stream in which successive groups of N binary digits in the stream are translated at a transmitter into groups of M binary digits, where M > N, said translation being accomplished according to at least the following rules: a) only M-bit groups having a disparity of either polarity not exceeding M/2 - 2 are allowable.

b) no M-bit groups having more than n consecutive digits of like significance at the beginning or end of the group is allowable, c) no M-bit group with zero disparity having more than n-2 consecutive digits of like significance at the beginning of the grop is allowable, d) no M-bit group with zero disparity having more than n-1 consecutive digits of like significance at the end of the group is allowable.

In a preferred embodiment of the invention N = 7, M = 8 and n = 4. For this code there are a total of 136 8-bit groups available. The DSV is 6 and minimum number of transitions in each group is 2. The longest run of like digits in any two consecutive groups in 7. In a system required to convey 128 levels of speech, for example, there are 8 spare groups. This code has both the unique features referred to above, namely 1. The peak digital sum in every group occurs at the digit 5 position.

2. The maximum run of 7 like digits can only occur with the last digit of the run in the digit 4 position of a group.

With this code there are several options when the use of the spare groups is considered. Firstly, all the spare groups can be used as alternative coding to provide 8 auxiliary channels at 155 k bit/sec. This results in a parity count d.c. content of 1/232. A second option is where all the spare groups are used to maximise the parity count d.c. content, in which case no auxiliary channel is provided. In this case the parity count d.c.

content is 1/40. The third option is to provide only one 155 k bit/sec auxiliary channel, the parity cound d.c.

content then being 1/44.

A common method of obtaining group alignmentwith block codes is to check for coding rule violations (e.g. by monitoring the end-of-group digital sum). If in the initial phasing a high violation rate is measured then the relative phase of input serial data and group rate clock is slipped. This process is repeated until an alignment is obtained which gives a zero or an acceptably low violation rate. In the correct group alignment violations are only due to line errors. This method suffers from the disadvantage that, although a decoder may be in the correct alignment, a high error rate burst, due say to external interference, can cause the decoder to methodically slip through all phases before getting back to the original alignment phase which it need not have left at all.Increasing the integration time over which the error/violation rate is measured helps in this respect but gives longer re-alignment times in the event of genuine loss of alignment.

There are two methods of overcoming the above shortcoming. The first is to look for unused code groups in all phases simultaneously and choose the phase which give zero or the lowest rate of occurence of these groups. This greatly improves the resilience of group alignment to burst interference, but is somewhat cumbersome when there are 8 phases to monitor followed by the selection logic.

The second method involves organising the coding so that a particular event can be uniquely associated with group alignment. Then by inspecting the serial data stream, in the absence of line errors, group alignment can be instantly recognised with6iit having to establish an error/violation rate. The group strategy to combat line errors can be very similar to normal multiplex frame alignment strategy. The two unique features outlines above, viz. peak digital sum and maximum run of like digits are both particularly well suited for this purpose. The mean rates of occurrence of these unique features are: 7 ones 34 fas 7zeros DS peaks of + 3 9.5 ys for a 140 Mbit/secsystem.

The digital sum method can advantageously be combined with error detection. One alignment strategy is as follows: i) In the search mode So, align digit 5 of the group rate clock with the first DS of 13 received.

ii) Check that the next DS of +3 also aligns with digits. If it does so then go into the aligned mode Ao. If it does not then realign on to this new time position and continue checking Sa until two consecutive DS peaks occur on the same digit.

iii) In the aligned mode line errors are detected by checking for DC = +3 occuring in digits other than digit 5. Confirmation of alignment is obtained by checking for DS = 13 occuring in digits. These checks are used to implement the state diagram shown in Figure 2. It should be noted that: i) In the absence of errors the DS peaks of +3 can only occur in digits, and cannot be simulated by traffic.

ii) In the presence of errors each error indication is in general accompanied by a confirmatory alignment indication. This property gives the alignment very high immunity against errors.

A typical 7B8B code translation table is illustrated in Figure 3.

Whilst no specific hardware implementation for the invention has been given it will be apparent to those skilled in the art that conventional digital data handling technology can be employed, the necessary adaptions for the particular codes to be used being quite straightforward.

Claims (6)

1. A method of transmitting information encoded into a binary digit stream in which successive groups of N binary digits in the stream are translated at a transmitter into groups of M binary digits, where M > N, said translation being accomplished according to at least the following rules: a) only M-bit groups having a disparity of either polarity not exceeding M/2 - 2 are allowable, b) no M-bit groups having more than n consecutive digits of like significance at the beginning or end of the group is allowable, c) no M-bit group with zero disparity having more than n-l consecutive digits of like significance at the end of the group is allowable, d) no M-bit group with zero disparity having more than n-1 consecutive digits of like significance at the end of the group is allowable.

2. A method according to claim 1 wherein N = 7, M = 8 and n = 4.

3. A method of transmitting digital data substantially as hereinbefore described with reference to the accompanying drawings.

4. A digital data transmission system wherein data is transmitted by the method claimed in any preceding claim.

5. A transmitter for transmitting information by the method claimed in any one of claims 1 to 3.

6. A receiver for receiving information by the method claimed in any one of claims 1 to 3.