Error detection/correction and fault detection/recovery – Pulse or data error handling – Digital data error correction
Reexamination Certificate
1998-07-30
2001-02-27
De Cady, Albert (Department: 2784)
Error detection/correction and fault detection/recovery
Pulse or data error handling
Digital data error correction
C375S240000, C375S340000
Reexamination Certificate
active
06195778
ABSTRACT:
FIELD OF THE INVENTION
This invention relates to optical-disk systems, and more particularly for methods to demodulation of run-length-limited (RLL) coded data.
BACKGROUND OF THE INVENTION
Optical disks such as compact-disks ROMs (CD-ROMs) and digital-versatile disks (DVDs) store data as pits on the surface for CD or between sandwiched layers for DVD. As a laser scans over these pits in a track, light is scattered, reading a zero, while the absence of pits allows the laser to be reflected back to a photo-sensor, which reads a one.
These pits have a minimum length and spacing. The minimum length for DVD is two bits, while the maximum length is 10 bits. A maximum pit length is needed so that a clock signal can be encoded with the data and extracted during playback.
The pits are sequenced so that analog signals reading the disk do not vary significantly form an average voltage. Minimizing voltage excursions allows more precise small-signal electronics to be used for the read pickup. If the read signal tended to sweep in voltage over time, or had a D.C. buildup, then the small-signal electronics could operate away from its bias point and thus operate with less precision.
Modulation information is added to the user data before recording to achieve a zero D.C. bias. For CD-ROMs, each 8-bit user data byte is expanded during modulation to 14 bits to define the physical pits on the disk. DVDs expand each byte to 16 bits. Rather than simply add 6 or 8 bits to each byte, the byte is completely remapped to a 16-bit codeword that may not have any similarity to the data byte. The modulation codes are carefully chosen to minimize low-frequency or D.C. energy, and include characteristics beneficial to synchronization and merging.
The modulation process for CD-ROMs is known as EFM, for eight-to-fourteen modulation, or simply 8/14. DVDs use a new 8/16 modulation process sometimes called “EFM-Plus”, since DVD modulation expands eight bits to sixteen, two more bits than for CD. The modulation codes themselves do not appear to follow any discernable pattern, so tables are often used to convert bytes to codewords and back to bytes. These tables can be quite lengthy, since a one-to-one mapping is not used.
FIG. 1
illustrates modulation of data written to a DVD disk and demodulation of data read from the DVD disk. The user's data is known as a symbol. The user data stream includes the user's actual data and overhead bytes such as error-correction codes, headers, identifiers, and checksums.
A symbol byte SYM is converted to a 16-bit codeword CW by 8/16 modulation, and the 16-bit codeword is written to DVD disk
18
. The actual writing process may be quite complex, such as writing a pattern of pits to a substrate, then forming a mold from the patterned substrate, and finally using the mold to stamp out thousands of individual DVD disks
18
.
When DVD disk
18
is read, the analog signals are processed to generate a bit stream that is divided into 16-bit codewords. Each codeword is converted by 8/16 demodulation back to an 8-bit symbol SYM, the same symbol that was input to the modulation process when the DVD disk was written.
FIG. 2
shows signals read from a DVD disk. The physical pits on the DVD disk are shown in contour
10
. As a laser is scanned over contour
10
, it is reflected to a photo-sensor when the surface is flat, such as it is initially in
FIG. 2
, but scattered when each of the two pits are rotated under the laser. The analog signal with waveform
12
is generated by the photo-sensor and amplified. When the laser is over the pits of contour
10
, signal
12
is low (little light picked up by the photo-sensor). When the laser is over the flay surface of contour
10
, a high signal
12
is generated by the large amount of laser light reflected to the photo-sensor.
A non-return-to-zero-inverted NRZI encoding is used for the disk data so that signal transitions occur frequently enough to encode a clock with the signal. This clock can be recovered with a phase-locked loop (PLL). Once the data is extracted from the clock from signal
12
, the return-to-zero RZ binary bitstream
14
is produced. Each edge of a pit creates a binary one, while no transitions, such as in the middle of a pit or the flat surface, create zeros. As can be seen, bitstream
14
contains many zeros and relatively few ones. The modulated codewords likewise have longer runs of zeros with single ones when the pits transition to the surface.
Modulation Tables—FIG.
3
FIG. 3
shows that the modulation tables in the DVD specification can be used to generate 16-bit codewords from symbol bytes. The DVD specification is available from Toshiba Corp. and includes tables that define the 8/16 modulation standard for DVD. The specification includes two tables: a main table
20
and a substitute table
22
. Both tables are indexed by the 8-bit symbol. While the main table
20
has one line for each of the 256 possible symbols, substitute table
22
contains only 87 lines.
Each table
20
,
22
is further divided into four sub-tables. One of the four sub-tables is chosen based on a current state: states
1
,
2
,
3
, or
4
. For example, when the current state is 2, the second of the four subtables is chosen and the other three subtables are ignored.
Each line contains one codeword entry in each of the four subtables, one codeword for each of the four states. Thus a symbol input for modulation selects one of the 256 lines of main table
20
. This line has four codewords: one of the codewords is selected based on the current state from flip-flop
26
. For some symbols, a matching line exists in substitute table
22
as well as main table
20
. When substitute table
22
contains a line for the symbol, it also has one of the four codewords in the line selected by the current state from flip-flop
26
.
The codeword from main table
20
and the codeword from substitute table
22
are input to multiplexer or mux
24
, and one of the two codewords is selected and output as the 16-bit codeword. The codeword entry from either the main or the substitute table can be chosen, based on the entry than minimizes the absolute value of a running digital sum value (DSV). The DSV is incremented for each 1 and decremented for each 0 in a codeword. Alternately, the substitute entry can be used when the DSV exceeds a threshold. Minimizing the running DSV minimizes voltage excursions of the read signal and thus improves read accuracy.
A four-state machine is apparently used during modulation, but the rules for sequencing among the four states when demodulating are not given in the DVD spec, nor is any pattern apparent from the tables. Instead, each codeword entry
16
in the tables has a next state field
17
appended. The next state is stored as a 2-bit field and output with the selected 16-bit codewords from tables
20
,
22
. Mux
24
selects the next state and codeword and sends the 2 next-state bits for storage in flip-flop
26
. The next state stored in flip-flop
26
is used as the current state for the next symbol lookup.
De-Modulation Tables—FIG.
4
The DVD spec provides no separate tables for demodulation. Instead, the modulation tables can be used in reverse.
FIG. 4
shows how demodulation can be performed using the modulation tables in the DVD spec.
An exhaustive table of 2
16
lines, (64K lines) could be used for the reverse tables for demodulation, using the 16-bit codeword as the index. However, since a run-length-limited code is used for the codewords, most possible combinations of the 16K possible binary combinations are not used. A smaller reverse table using just the entries in the modulation tables is more economical.
The entries in the modulation tables can be ordered by codeword and an associative lookup performed for matching codewords. The demodulation tables act as an associative-mapped cache. The entries are still segregated into subtables by the next state, and into the main and substitute tables. The entries in main demodulation table
30
and substitute demodulation table
32
each contain a codeword, the symbol m
Auvinen Stuart T.
Cady Albert De
Lamarre Guy
LSI Logic Corp.
LandOfFree
Demodulation of DVD codewords using dependency-sorted tables... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Demodulation of DVD codewords using dependency-sorted tables..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Demodulation of DVD codewords using dependency-sorted tables... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2595399