Data encoding and demodulation system.

0001The invention relates generally to the field of signal demodulation and more particularly to the encoding of data to facilitate demodulation of signals recorded on mass storage media, for example, signals used for storing data in a digital data processing system, to identify the digital data represented thereby.

0002Binary digital data is typically recorded on mass storage media as a pattern of transitions in a storage medium. For example, data on magnetic storage media is represented by changes of magnetic polarity, and data on optical storage media is represented by changes in reflectivity or transmissivity. The transition patterns correspond to digital data which has been encoded to facilitate recording. When a magnetic memory is read using a head which is inductively read, or an optical memory is read using an optical scanner, an analog signal is generated with relative positive and negative peaks or large and small signals corresponding to the transition pattern. The analog signal, which may be distorted by system noise and other influences, is then demodulated to extract the original transition pattern as faithfully as possible and interpret it as a series of binary encoded bits. The binary encoded bits must then be decoded to reproduce the original digital data.

0003Signal demodulation may become increasingly difficult as the density of data recorded on the disks is increased. With higher recording densities, the medium, or disk, space allotted for the recording of a transition signal, that is, a transition cell, is, in effect, reduced. The signals read from the transition cells tend to be smaller relative to the background noise as a result and they can be more readily misinterpreted.

0004The signals may be misinterpreted because of system noise, which can distort the signals read from the "small" transition cells, or because of interference from surrounding transition cells, which can cause signal peaks to shift either within a transition cell or even to adjacent transition cells. The misinterpretation of the transition signals results in errors in the binary encoded bits which, after decoding, results in errors in the digital data.

0005The need to encode data specifically to facilitate demodulation and minimize the effects of peak shifts, etc., is well known. One such encoding scheme is shown in United States Patent 4,146,909 to Beckenhaur et al., assigned to International Business Machines, Inc. (IBM). Data which may already be encoded for error correction is further encoded using a demodulation code, for example, the IBM code, before it is recorded on the storage medium. Thus the data is recorded in the form of a series of transitions corresponding to the demodulation code words.

0006When a user requests data from storage the demodulation code words corresponding to the data are retrieved; that is, an analog signal is produced from the stored transition pattern. The analog signal corresponding to the code words must then be demodulated. Demodulation of the code word signal requires first determining the binary values corresponding to the signal transitions, and thus, the bit values, and next, finding the beginnings of the code words and decoding them. The resulting data may then be further decoded, for example, for error correction, to reproduce the actual data.

0007In order to facilitate demodulation the data is encoded such that the effects of signal transitions on nearby transition cells, for example, peak shifts, in the data signal are minimized. The data is typically encoded using a code which separates consecutive binary ONES, which correspond to transition cells containing signal transitions or peaks, by a minimum of one or more binary ZEROS, which correspond to transition cells without signal transitions.

0008To further facilitate the determination of the bit values associated with the recorded signal, the data is encoded such that the data signal, which typically includes clock information which enables the demodulator to find and synchronize to the signal transition cells, has a limited number of consecutive binary ZEROS. A ZERO is represented by a transitionless cell, and thus, too many consecutive ZEROS may result in the demodulator losing clocking information. Codes having these limitations for both ONES and ZEROS are commonly referred to as "run-length-limited" codes.

0009The IBM code discussed above is a run-length-limited code which separates consecutive ONES in a code word by at least two ZEROS. The parameter d is commonly associated with this minimum number of ZEROS, and thus, d=2. The IBM code allows at most seven consecutive ZEROS. The parameter k is commonly associated with the maximum number of ZEROS, and thus, k=7. Codes for which d and k are both small numbers and relatively close to each other are best suited for demodulation encoding.

0010The IBM code using these d and k parameters encodes groups of two, three and four data bits into any of seven code words, but does not offer an advantageous code-rate, i.e., the ratio of total number of data bits, to the total number of encoded bits; nor does it offer a higher information content for each flux reversal. The number of valid code words in a code determines the complexity of the corresponding encoder and demodulator. Thus codes which contain fewer code words may use less complex encoders and demodulators.

0011It is desirable to improve the d=2, k=7 modulation code further. The present invention approaches the problem by using novel decoding and a code rate of 1/2; i.e., every data bit is mapped to two encoded bits.

0012The invention comprises a data storage system including an encoder and a demodulator which encodes and demodulates data, respectively, using an improved d=2, k=7 demodulation code. The encoder encodes one data bit to two code word bits, and thus, the code rate is 1/2.

0013The invention in its broad form resides in a system and a method for encoding data including the step of encoding data by using the following code: <tables id="tabl0001" num="0001"><table frame="all"><tgroup cols="2" colsep="1" rowsep="0"><colspec colnum="1" colname="col1" colwidth="78.75mm" /><colspec colnum="2" colname="col2" colwidth="78.75mm" /><thead valign="top"><row><entry namest="col1" nameend="col1" align="center">DATA BITS</entry><entry namest="col2" nameend="col2" align="center">CODE WORDS</entry></row></thead><tbody valign="top"><row><entry namest="col1" nameend="col1" align="right">00</entry><entry namest="col2" nameend="col2" align="left">000X</entry></row><row><entry namest="col1" nameend="col1" align="right">01</entry><entry namest="col2" nameend="col2" align="left">0100</entry></row><row><entry namest="col1" nameend="col1" align="right">10</entry><entry namest="col2" nameend="col2" align="left">100X</entry></row><row><entry namest="col1" nameend="col1" align="right">011</entry><entry namest="col2" nameend="col2" align="left">10000X</entry></row><row><entry namest="col1" nameend="col1" align="right">111</entry><entry namest="col2" nameend="col2" align="left">100100</entry></row><row rowsep="1"><entry namest="col1" nameend="col1" align="right">0000</entry><entry namest="col2" nameend="col2" align="left">10000001</entry></row></tbody></tgroup></table></tables> where X is a ONE if the last two bits of the preceding code word are both ZEROS and X is a ZERO otherwise, and the right-­most bits of the data and the code words are the first in time.

0014Alternatively, the encoder of the system encodes data to form any of six possible code words according to the following code word chart: <tables id="tabl0002" num="0002"><table frame="all"><title>CHART 1</title><tgroup cols="2" colsep="1" rowsep="0"><colspec colnum="1" colname="col1" colwidth="78.75mm" /><colspec colnum="2" colname="col2" colwidth="78.75mm" /><thead valign="top"><row><entry namest="col1" nameend="col1" align="center">DATA BITS</entry><entry namest="col2" nameend="col2" align="center">CODE WORDS</entry></row></thead><tbody valign="top"><row><entry namest="col1" nameend="col1" align="right">00</entry><entry namest="col2" nameend="col2" align="left">000X</entry></row><row><entry namest="col1" nameend="col1" align="right">01</entry><entry namest="col2" nameend="col2" align="left">0100</entry></row><row><entry namest="col1" nameend="col1" align="right">10</entry><entry namest="col2" nameend="col2" align="left">100X</entry></row><row><entry namest="col1" nameend="col1" align="right">011</entry><entry namest="col2" nameend="col2" align="left">10000X</entry></row><row><entry namest="col1" nameend="col1" align="right">111</entry><entry namest="col2" nameend="col2" align="left">100100</entry></row><row rowsep="1"><entry namest="col1" nameend="col1" align="right">0000</entry><entry namest="col2" nameend="col2" align="left">1000000X</entry></row></tbody></tgroup></table></tables> where X is a ONE if the last two bits of the preceding code word are both ZEROS and X is a ZERO otherwise, and the right-most bits of the data and the code words are the first in time.

0015Other aspects and features of the invention are defined in the accompanying claims to which reference is hereby made.

0016The invention also extends to a method for storing data in a storage medium including applying the data encoded as a run length limited code, according to the encoding methods defined in the claims, to the storage medium.

0017In this regard, the invention extends to a method of retrieving data stored as a run length limited code in a storage medium including demodulating the data in accordance with the demodulating methods defined in the claims.

0018The invention further extends to a system for storing encoded data in a storage medium including means for applying the data encoded by the encoding systems, defined in the claims, to the storage medium.

0019This aspect of the invention also extends to a system for retrieving encoded data from a storage medium including means for reading the encoded data and means for demodulating the encoded data in accordance with the demodulating systems defined in the claims.

0020First, an encoder examines four incoming data bits. If the four bits are all ZEROS the encoder next examines the last two bits of the previous code word to determine if the bits are 01. If they are, the encoder encodes the first two of the four ZERO bits to form one of the four-bit code words in the Chart 1 above. The encoder selects the four bit code word which preserves the limited run length, and thus, it selects 0000. Otherwise, the encoder encodes the four ZERO bits to form one of the two eight-bit code words in the code word Chart 1 above. The encoder selects the eight-bit code word which preserves the limited run length of the code. For example, if the two previous bits are 00 the encoder selects 10000001.

0021If the four data bits are not all ZEROS, the encoder next examines just the first two data bits. If these bits are both ONES the encoder encodes the two data bits along with the third data bit to form a six-bit code word according to the chart. Otherwise, the encoder encodes the first two data bits to form a four-bit code word according to the chart.

0022After these data bits are encoded, the encoder applies the code word to storage, for example, a disk drive, where it is stored for future use. The encoder also retains a copy of the last two bits of the code word. The encoder then continues to encode any incoming data according to the chart and applying it to the disk.

0023When data is requested from storage, the code words corresponding to the data, that is, the recorded signals, are retrieved from the disk. The data signal is then applied to a demodulator. The demodulator synchronizes to the signal, assigns bit values to it, and then attempts to decode code words. Specifically, it first attempts to decode the first eight bits of the encoded data. It sets the first two encoded bits to ZEROS and then compares the eight bits, two of which are set to ZERO, to the valid code word 10000000. If the encoded data bits match a code word the code word decodes to four data ZEROS. Otherwise, the demodulator attempts to decode the first six bits to three data bits and if this fails it decodes the first four bits to two data bits. If the code word bits can not be decoded according to the chart an error is declared for the first two data bits.

0024The invention may be better understood by referring to the following description of a preferred embodiment, given by way of example, and to be studied in conjunction with the accompanying drawings, in which: <ul id="ul0001" list-style="none"><li>Figure 1 is a block diagram of an encoding and demodulation system;</li><li>Figure 2 is a flow chart of the operations of an encoder encoding data in accordance with a preferred embodiment the invention; and</li><li>Figure 3 is a flow chart of the operations of a demodulator decoding the encoded data.</li></ul>

0025A data storage and retrieval system 10 includes an encoder 12, storage 16, for example, a disk drive, and a demodulator 18. The encoder 12 receives from a data bus 11 data bits which may have already been encoded for error correction purposes. The encoder further encodes the data bits into a code word using a d=2, k=7 code designed to facilitate demodulation, as discussed below with reference to Figure 2. The encoder then applies the code word to the disk drive 16. The encoder also retains in a buffer 12A a copy of the last two bits of the code word.

0026When the encoder 12 encodes the next few data bits it examines the retained bits and selects a code word for the data which preserves the limited run length (d=2, k=7) of the code. Once the encoder 12 has encoded the data the disk drive 16 stores the code words on, for example, a magnetic disk as a series of signal transitions, i.e., a series of flux reversals.

0027When a user requests stored data a data request signal 14 is sent to the disk drive 16. The drive 16 retrieves from the magnetic disk the signal corresponding to the data and applies the signal to the demodulator 18. The demodulator 18 synchronizes to the signal and assigns binary bit values to it. The demodulator 18 next decodes the data, if possible, as discussed below with reference to Figure 3. The demodulated data may be applied to an error correction decoder (not shown) for further decoding and correction, as appropriate.

0028To encode the incoming data the encoder 12 uses the code words set forth in Chart 1 below: <tables id="tabl0003" num="0003"><table frame="all"><title>CHART 1</title><tgroup cols="2" colsep="1" rowsep="0"><colspec colnum="1" colname="col1" colwidth="78.75mm" /><colspec colnum="2" colname="col2" colwidth="78.75mm" /><thead valign="top"><row><entry namest="col1" nameend="col1" align="center">DATA BITS</entry><entry namest="col2" nameend="col2" align="center">CODE WORDS</entry></row></thead><tbody valign="top"><row><entry namest="col1" nameend="col1" align="right">00</entry><entry namest="col2" nameend="col2" align="left">000X</entry></row><row><entry namest="col1" nameend="col1" align="right">01</entry><entry namest="col2" nameend="col2" align="left">0100</entry></row><row><entry namest="col1" nameend="col1" align="right">10</entry><entry namest="col2" nameend="col2" align="left">100X</entry></row><row><entry namest="col1" nameend="col1" align="right">011</entry><entry namest="col2" nameend="col2" align="left">10000X</entry></row><row><entry namest="col1" nameend="col1" align="right">111</entry><entry namest="col2" nameend="col2" align="left">100100</entry></row><row rowsep="1"><entry namest="col1" nameend="col1" align="right">0000</entry><entry namest="col2" nameend="col2" align="left">1000000X</entry></row></tbody></tgroup></table></tables> where X is a ONE if the last two bits of the preceding code word are both ZEROS and X is a ZERO otherwise, and the right-most bits of the data and the code words are the first in time.

0029With reference to Figure 2, the encoder 12 examines the next four incoming data bits to determine if all the bits are ZEROS (steps 20-22). If the bits are all ZEROS the encoder 12 examines the last two bits of the previous code word stored in buffer 12A (Figure 1). Unless the two stored bits are 01, the encoder 12 encodes the four ZERO bits into an eight-bit code word which preserves the limited run- length of the code (steps 28-34). Thus, if the stored bits are both ZEROS, the encoder encodes the data bits to form the code word 10000001 to avoid too many consecutive ZEROS (steps 30-32). If the stored bits are 10 the encoder encodes the data bits to form the code word 10000000 to avoid consecutive ONES (steps 30, 34). If the stored bits are 01 the encoder encodes the first two of the four ZERO bits to form a four-bit code word 0000 (step 26).

0030If the four data bits are not all ZEROS the encoder examines the first two data bits to determine if they are both ONES (step 24). If they are, the encoder encodes these two data bits and the next one to form a six-bit code word according to Chart 1 (step 36). For example, if the third data bit is a ONE the encoder 12 encodes the three data bits, 111, to form the code word 100100.

0031If the first two incoming data bits are not both ONES the encoder encodes the two data bits to form a four-bit code word according to Chart 1 (step 26). For example, the two data bits may be 01 and the encoder produces the corresponding code word 0100.

0032After the encoder formulates a code word it applies the code word to the disk drive 16. The encoder also retains a copy of the last two bits of the code word in buffer 12A (steps 38-40).

0033The encoder 12 then examines the next four data bits which have not yet been encoded. For example, if the encoder first examines four bits, b4, b3, b2, b1, and encodes the first two data bits b2 and b1 to form a four-bit code word, the encoder next examines the two bits which have not yet been encoded, bits b4 and b3, along with the next two incoming data bits, bits b5 and b6. If the encoder encodes these four examined data bits b6, b5, b4, and b3 to a code word, it then examines the next four incoming data bits b10, b9, b8, and b7 to determine if these bits are all ZEROS. Thus the encoder encodes the following data bits: 01 00 00 00 00 00 00 01 00 00 10 00 01 where the right-most bit is the first incoming data bit, to form the code words: 0100 0000 <u style="single">1000 0000</u><u style="single">1000 0001</u> 0000 0100 <u style="single">1000 0000</u> 1001 0000 0100 where the underlined code words are the eight-bit code words corresponding to four all ZERO data bits. Referring now to Figure 3, when a data request 14 arrives at the disk drive 16, the disk drive 16 retrieves from the magnetic disk the signal corresponding to the data (step 42). The disk drive then sends the retrieved signal to the demodulator 18. The demodulator 18 assigns bit values to the signal and then attempts to decode the bits as code words. To decode the code words, the demodulator 18 sets the first two retrieved bits to ZEROS and then attempts to decode the first eight, six or four bits to four, three or two data bits, as set forth below (steps 44-46).

0034The decoder sets the first two bits to ZEROS to force to a given format a code word which includes an "X" set to a ONE, for example, 0001 for data bits 00. The value of the "X" term does not add any information to the code word, it simply preserves the desired limited run length. Thus setting the first two bits to ZEROS, and thereby forcing the X terms to ZEROS, simplifies the demodulator 18 by limiting the number of code words which the demodulator 18 must recognize.

0035The demodulator 18 compares the first eight bits, including the two set to ZEROS, with the eight-bit code word 10000000 shown in Chart 1. If the bits match the code word the demodulator decodes the bits as four data ZEROS (step 54).

0036If the eight bits do not match the code word, the demodulator 18 compares the first six bits, including the two bits set to ZEROS, with the two valid six-bit code words shown in Chart 1. If the bits match one of the code words the demodulator 18 decodes the code word to the corresponding three data bits (steps 48, 58). Otherwise, the demodulator 18 attempts to decode the first four bits, again including the two bits set to ZEROS, to two data bits according to Chart 1 (steps 50, 56). If the four retrieved bits do not form a valid code word the demodulator 18 declares a detected error in the first two data bits (step 52).

0037The demodulator 18 easily decodes four-bit code words to two data bits by assigning the data bits the values of the two most significant bits of the four bit code word. Thus the code word 0100 decodes to 01.

0038Once the demodulator has decoded a first code word or declared the first two bits erroneous, it examines the next eight encoded bits, sets the first two to ZEROS and attempts to decode them to four data bits, and so forth, until allthe code words corresponding to the requested data have been decoded.

0039In order to ensure that the bits which are examined for decoding are bits at the start of a code word rather than bits in the middle of a code word, the demodulator 18 first examines the placement of the ZEROS and ONES in the first few retrieved bits. The demodulator 18 finds the beginning of a code word when it finds a ONE in bit position bo, o being an odd number, which is followed by three or more ZEROS. Similarly, it finds the beginning of a code word if it finds, starting at a bit position bo, either of two eight-bit patterns: 00000100 or 01000100. The demodulator 18 may instead find the end of a code word by finding a ONE in bit position be, where e is an even number. The demodulator 18 determines that a particular bit is in an even or odd bit position based on information in the corresponding clock signal, in a manner which is well known to those skilled in the art.

0040If the demodulator 18 finds a repeated pattern of 0001 in the retrieved data with the ONE in a location corresponding to an odd bit position, the demodulator 18 determines that the retrieved sequence is shifted two bit positions, that is, that the sequence is advanced or delayed by two bits. Thus the demodulator advances two encoded bits to ensure that the pattern 0100 is the repeated one and the demodulation is thus synchronized to the data. It is rare that a properly functioning demodulator advances or delays data by an odd number of bits.

0041The encoder 12 and the demodulator 18 may encode and decode data using a second code word chart, as set forth below: <tables id="tabl0004" num="0004"><table frame="all"><title>CHART 2</title><tgroup cols="2" colsep="1" rowsep="0"><colspec colnum="1" colname="col1" colwidth="78.75mm" /><colspec colnum="2" colname="col2" colwidth="78.75mm" /><thead valign="top"><row><entry namest="col1" nameend="col1" align="center">DATA BITS</entry><entry namest="col2" nameend="col2" align="center">CODE WORDS</entry></row></thead><tbody valign="top"><row><entry namest="col1" nameend="col1" align="right">00</entry><entry namest="col2" nameend="col2" align="left">000X</entry></row><row><entry namest="col1" nameend="col1" align="right">01</entry><entry namest="col2" nameend="col2" align="left">0100</entry></row><row><entry namest="col1" nameend="col1" align="right">10</entry><entry namest="col2" nameend="col2" align="left">100X</entry></row><row><entry namest="col1" nameend="col1" align="right">011</entry><entry namest="col2" nameend="col2" align="left">10000X</entry></row><row><entry namest="col1" nameend="col1" align="right">111</entry><entry namest="col2" nameend="col2" align="left">100100</entry></row><row rowsep="1"><entry namest="col1" nameend="col1" align="right">0000</entry><entry namest="col2" nameend="col2" align="left">10000001</entry></row></tbody></tgroup></table></tables> where X is a ONE if the last two bits of the preceding code word are both ZEROS and X is a ZERO otherwise, and the right-most bits of the data and the code words are the first in time.

0042The only difference between this code and the code shown in Chart 1 is that there is only one valid eight-bit code word. The four data ZEROS are encoded to the eight-bit code word only if the last two bits of the previous code word are ZEROS. Otherwise, the first two ZERO bits are encoded to 0000.

0043Using the demodulation codes set forth above, a demodulator easily determines the binary values associated with ananalog data signal. The code is well suited to synchronizing the demodulator to the signal because it encodes a stream of data ZEROS to code words which preserve the desired limited run-length. It also facilitates synchronization of the demodulator to the beginnings or endings of codewords by encoding the data to code words with ONES in particular, that is, even or odd, bit positions. In addition, the encoder and demodulator required for the code may be less complex than those used in prior art systems because the present code has only six valid code words.

0044The foregoing description has been limited to two specific embodiments of this invention. It will be apparent, however, that variations and modifications may be made to the invention, with the attainment of some or all of the advantages of the invention. Therefore, it is the object of the appended claims to cover all such variations and modifications as come within the true scope of the invention.