Method for digital input signal conversion for optical light guide transmission systems.

The invention relates to a method for converting digital input signals from a first binary code with 32 different code words with five binary digits corresponding to 32 possible amplitude values ​​in output signals in a second binary code with codewords of six binary digits with a positive and a negative fashion, with which Fashion yields according to the running digital sum at the end of the previous codeword with six binary digits occur.

The development of low-loss glass types for fiber optic cables on the one hand and increasing the lifetime of the light emitter on the other hand allows the introduction of fiber optic transmission systems as a cost-effective alternative to energy .and copper cable systems. Due to the nonlinear characteristics of the optical transmission elements are preferably digital signals - and in binary code - transfer, where "light on" is the binary 1 and "light" of the binary 0 or associated vice versa. This results in two problems.

As in intermediate regenerators in general takes place both an amplitude and a timing regeneration of the signal, the transmission signal must contain sufficiently large clock information. This is not guaranteed for any data signal. Secondly, the average value (DC component) of the signal should be constant, so that the receiver only the alternating component must be strengthened and the constant and known DC component must be taken into account until the amplitude decision. Both problems can be safely achieved only in that the signal redundancy is added. Since fiber optic systems in general, a small increase in the transmission bandwidth can be taken into account, has for fiber optic systems with 34 or 140 Mbit / s enforced block coding, to be implemented at the block, five binary digits of the input signal into six binary digits of the output signal , The input signals can dabei.auch of code words with n blocks consist of five binary digits that are implemented accordingly in n blocks, each with six binary digits.

A method for transcoding of messages that are available in a binary code, is known from DE-AS 23 39 868th In this method the n elements are then transmitted each code word of the first code on the first n pixel locations of a code word of the output signal in the original order and on the last n element locations in reverse order.

Also known from DE-AS 27 47 018 a method for converting binary code words is known. In this method, the binary code words are converted into ternary code words.

From the periodical "Frequency" 34 (1980) 2, pages 45 to 52, a method of the initially mentioned type is known. existing input signals In this known method also consists of five binary digits converted into output signals containing six binary digits and which occur in a positive or in a negative fashion. The positive fashion thereby contains only code words in which the number of binary digits 1 to the number of binary digits is 0 or one greater than this, while the code words occurring in the negative mode are configured so that the number of binary digits 0, the number of binary digits 1 the same or a greater than this is. The occurrence of the code words in the positive or negative mode results from the running digital sum of sechsziffrigen codewords of the output signal, which is formed by reacting rated the binary digit 1 with a value of +1 and the binary digit 0 to -1.

In Table 2 of the publication "Frequency", the assignment between input and output code words is shown. For the reaction in addition to a series-parallel and parallel-to-serial converter, a corresponding clock implementation and receive side, besides the actual transcoder means for synchronization is necessary. This results in a relatively high cost, which may increase further by monitoring devices for the listed components.

The .Aufgabe the invention is to provide a process of the aforementioned type, which can be realized with comparatively little effort.

According to the invention the object is achieved that the 32 possible amplitude values ​​of the input signal is divided into two roughly equal parts, each with 16 code words, while the first region, the amplitude values ​​between a minimum amplitude value and a first average amplitude value and the second region, the amplitude values ​​between a second average amplitude value and a maximum amplification comprises tudenwert that the code words for the positive and negative mode the output signal of the first region are chosen so that the largest possible match between input and output results, and that the code words for the second area is characterized showed that for the lowest amplitude value of the second region, the code word for the positive mode the inverted codeword of the negative mode the maximum amplitude value of the first region and the code word for the negative mode corresponds to the inverted code word of the positive mode of the highest amplitude value of the first region that for the second lowest amplitude value of the second region, the code word for the positive mode the inverted codeword of negative fashion the second highest amplitude value of the first region and the code word for the negative mode corresponds to the inverted code word of the positive mode of the second highest amplitude value of the first region that according to the code words be formed for other amplitude values ​​of the second region and that finally for the highest amplitude value of the second region, the code word for the positive mode the inverted codeword of the negative mode the lowest amplitude value of the first region and the code word for the negative mode the inverted codeword of the positive mode the lowest amplitude value of the first region corresponds.

The described division into two areas and the symmetry between the two areas is the advantageous possibility of carrying out the assignment with the aid of gates and controlled inverter stages.

To further reduce the complexity, it is expedient that the respective code words for the positive and negative mode are equal in both areas for the lowest and the highest amplitude value.

A preferred embodiment of the inventive method is described in the patent claim. 3 In carrying out this assignment results in an advantageous reduction of error multiplication of averaged 1.28 over the method according to the prior art, in which an error multiplication factor is achieved averaged 1.41.

Further embodiments of the inventive method are described in the patent claims 4 to 6th

The application is intended to be explained in more detail below with reference to the drawing. In the drawings<ul><li>Fig. 1 is a table for the conversion of five-bit words, in six-bit words and</li><li>FIG. 2 is a supplementary table to table to Fig.1.</li></ul>

In Table 1, the 32 possible different codewords are represented from five binary digits, which may take the input signal, the current numbers 0 to 31, indicating the respective amplitude level. To the five-bit words followed by the six-bit words of the output signal are shown in the positive and in the negative mode, which are released as a result of the reaction. In addition, corresponding to the six-bit words, amplitude levels, and the two modes are illustrated sequence. As already mentioned, the six-bit words - controlled by the running digital sum - appear in a positive or in a negative fashion. The order of the five-bit words and the six-bit words is indicated by AB C ..., wherein each A is first read or comment. It can be seen that the 32 possible amplitude values ​​of the input signal can be divided into two equally large regions, the one region the amplitude values ​​from 0 to 15 and the other region, the amplitude values ​​comprises 16 to 31st

The association between denSechs-bit words and the five-bit words in the first region was chosen so that in each case the largest possible match between input and output signal obtained when the first point of the six-bit word is not considered. This is not only simple, but also fast assignment in the occurring amplitude ranges ensured.

Now With regard to a simple realization bypassing consuming only memory allocation for the second region has been chosen so that the lowest amplitude value of the second region, the code word for the positive mode corresponds to the inverted codeword of the negative mode the maximum amplitude value of the first region. The code word corresponding to the amplitude value 49 provides binary words, the inverted code word for the amplitude value 14. Since the highest amplitude value of the first region, the code words for positive and negative mode are the same, they are the also the lowest amplitude value of the second region.

For the second lowest amplitude value of the second region, the codeword 57 for the positive mode corresponds to the inverted codeword 6 of negative fashion the second highest amplitude value of the first region, also corresponds to the codeword 17 for the negative mode the inverted codeword 46 of the positive mode of the second highest amplitude value of first region. The table 1 is thus assuming a line of symmetry between the amplitude values ​​15 and 16 of the five-bit word, and the corresponding values ​​of the six-bit word is inverted diagonally symmetrical. Accordingly therefore the code word 19 for the fourth-highest amplitude value of the second region is equal to the inverted codeword 44 for negative fashion the fourth highest amplitude value of the first region; because in this case the code words 44 for the positive and negative mode are the same in the first region, they are the also the fourth lowest amplitude value of the second region. Finally, the code word for the positive mode of the highest amplitude value of the second region is equal to the inverted code word of the negative mode the lowest amplitude value of the first region. At the lowest amplitude value of the first region, the code words for the positive and negative mode are the same, so that they are essential for the highest amplitude value of the second region.

At the receiving end the reverse conversion of binary words in sechsziffrigen fünfziffrige binary words are recognized using the same allocation accordingly in the reverse direction.

At the reception side reconverting the problem may occur that during the transmission by a fault occurs a distortion of the originally emitted six-bit word. Assuming that only a single bit has been falsified by the disturbance, various sequences are possible. It may turn out that after decoding is not one bit, but multiple bits are corrupted or not a bit is corrupted, it can also be a codeword not originally used have originated. The error multiplication factor is 1.28 for the bit assignment according to Table 1 and 2 on the average. The error multiplication is thus comparatively very low.

In Table 2, the mapping between the six-bit words and five-bit words is shown for the fault cases. It lists the six-bit words that are at the transmission end is not used in normal operation, the two code words corresponding to the amplitude values ​​of 0 and 63 are only used during fault location operation and can occur only upon the occurrence of multiple errors.

The realization of the assignments according to Tables 1 and 2 by means of an optimized by known methods combinatorial network via gates and controllable inverters can conveniently be carried out in a highly integrated circuit, for example, a so-called mask-programmable logic device of the series SH100.