Coded data generation or conversion – Digital code to digital code converters – To or from run length limited codes
Reexamination Certificate
1999-06-01
2001-02-13
Young, Brian (Department: 2819)
Coded data generation or conversion
Digital code to digital code converters
To or from run length limited codes
Reexamination Certificate
active
06188337
ABSTRACT:
FIELD OF THE INVENTION
This invention relates to methods for encoding digital data. More particularly, this invention relates to serial binary data, and methods for encoding such data in preparation for placing such data on a transmission or storage channel. Still more particularly, this invention relates to methods that adjust the frequency power spectrum of the encoded data.
ART BACKGROUND
Block codes are widely used in optical storage and transmission of data. In accordance with the use of such a code, an input stream of binary bits is partitioned into input words of a fixed length. Each input word is mapped to an encoded output word of a second fixed length that is longer than that of the input words. The encoded output word is referred to as a codeword.
Several advantages may be gained from the use of block transmission codes. These advantages relate generally to error detection and to the quality of the recovered signal. One specific advantage is that the dc component of the encoded signal, i.e., the power spectral content at or near zero frequency, can be suppressed by using an appropriate code. This is desirable in, e.g., optical communication systems because fiber optic receivers often include an ac-coupled input stage. The processing of the recovered signal is simplified, and the quality of that signal is improved, if information content is suppressed at the relatively low frequencies where coupling is inefficient. Another advantage of dc suppression is that it simplifies the problem of recovering clock timing from the signal data.
It should be noted that actual two-level signals may be transmitted as sequences of 1's and 0's, or as sequences of +1's and −1's, or in various other equivalent representations. We will refer to all such signals as binary signals, and for convenience only and without limitation, we will take a sequence of 1's and 0's as exemplary of all such signals.
Various block transmission codes are known to those skilled in the art. For example, the well-known Manchester code maps each input bit into two output bits. Other well-known codes map 5-bit input words to 6-bit codewords. Such codes are referred to as 5B/6B codes. Yet other codes are 8B/10B codes. One example of an 8B/10B code is described in U.S. Pat. No. 4,486,739, issued to P. A. Franaszek et al. on Dec. 4, 1984. In the coding scheme of Franaszek et al., the 8B/10B coder is partitioned into a 5B/6B coder plus a 3B/4B coder.
For the purpose of spectrally adjusting the encoded signal to suppress dc power, it is generally advantageous to employ relatively high redundancy, that is, to employ a relatively high length ratio of the codeword to the input word. One reason for this is that spectral adjustment is achieved, at least in part, when the number of 1's in each codeword is exactly or approximately matched to the number of 0's. The excess of 1's over 0's, or of 0's over 1's, in a codeword is referred to as its disparity. Thus, decreasing the disparity tends to improve the power spectrum. However, of all the words of a given length, only a fraction of them will have zero, or very small, disparity. Thus, a requirement of low disparity reduces the number of available codewords and thus reduces the amount of information that can be transmitted per codeword. To compensate, it may be necessary to increase the redundancy, i.e., to increase the length of the codewords.
On the other hand, increasing the redundancy of the codewords decreases the gross rate at which information can be transmitted over the communication channel. Therefore, there is a tradeoff between redundancy and disparity. Both cannot be minimized simultaneously. There remains a need to find encoding schemes that combine moderate redundancy with moderate disparity.
SUMMARY OF THE INVENTION
I have invented such a coding scheme.
My code is a 16B/18B low disparity code. The 16-bit input word is split into two bytes, i.e., into two 8-bit words, and each byte is mapped to a 9-bit word. The image of each input byte under the mapping depends, at least in part, on its disparity.
In some cases, the resulting image is an intermediate image word, which is then inverted to obtain the output codeword. A word is said to be inverted if each of its 1's is converted to a 0, and each of its 0's is converted to a 1. The decision whether to invert an image depends, at least in part, on the running digital sum (RDS) of the output. The RDS is the sum of all 1's in the data stream from its beginning (or from a designated starting point).
All possible input words are grouped, by disparity, into a plurality of distinct categories. The rule for mapping of input words is different for each of the categories.
REFERENCES:
patent: 4486739 (1984-12-01), Franaszek et al.
patent: 5663724 (1997-09-01), Westby
patent: 6054944 (2000-04-01), Yamashita
Finston Martin I.
Lucent Technologies - Inc.
Young Brian
LandOfFree
Low disparity coding method for digital data does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Low disparity coding method for digital data, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Low disparity coding method for digital data will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2596556