Pulse or digital communications – Receivers – Particular pulse demodulator or detector
Reexamination Certificate
1997-12-12
2001-06-19
Pham, Chi (Department: 2631)
Pulse or digital communications
Receivers
Particular pulse demodulator or detector
Reexamination Certificate
active
06249553
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a device for detecting information in which information recorded on a recording medium at a high density can be detected with no error.
2. Description of the Related Art
Research and development on disk units with high densities and large capacities such as optical and magnetic disks represented by DVDs is widely and energetically being pursued. Signal processing technology supporting high reliability in regenerated information is indispensable for such disk units with the higher density features. To this end, filing devices using PRML (Partial Response Maximum Likelihood) technology have successfully entered the market. This system, in which equalization of a partial response wave form and maximum likelihood detection are combined, is well known by conducting the maximum likelihood retrieval after correcting readout signals by wave form equalization in order to utilize characteristics of a maximum likelihood detector giving full consideration to a regenerative channel. For example, there is a description on PRML in the preliminary paper volume for the 1994 annual convention of the Institute of Television Engineers (ITE '94, pp. 287 to 288).
Intersymbol interference grows larger and regenerative amplitude is reduced in the case where information recorded with a high density is regenerated in either an optical disk or a magnetic disk. Therefore, the SNR of a magnetic disk is smaller and the CNR of the high frequency component of a readout signal in the case of an optical disk is also smaller, so that the error rate of detected information is increased. The maximum likelihood detection system detects information by using characteristics of a regenerative channel having a given state transition and if one time series pattern is screened so as to have the minimum square mean of errors among all the time series patterns to be conceivable from the characteristics of the regenerative channel with respect to an amplitude information string having the number of quantized bits, for example, on the order of four bits, which is input to the detector, information can be detected with a low error rate, even though the SNR or CNR is small. It is difficult, however, to perform the above mentioned processing in an actual circuit in terms of the circuit scale and its processing speed, and thus an algorithm called a Viterbi algorithm is used to select a path in a recursive manner in order to realize the process, wherein the Viterbi algorithm is described in IEEE Transactions on Communications, vol. COM-19, October 1971.
An operation in the case where information recorded in an optical disk medium is detected by being subjected to PRML detection using the most simple PR (1, 1) channel is described with reference to
FIGS. 21
,
22
, and
23
. A signal read by a head is corrected in advance so as to be the PR (1, 1) channel by use of an equalizer, for example a transversal filter. As shown in
FIG. 21
, this channel is distributed around three reference levels E
1
(=−1, 0, 1). In this case, amplitude information X
i
which is digitized in each channel clock is subjected to a two state transition as shown in
FIG. 20. A
maximum likelihood detection finds a string Ei with which a sum of square errors z
n
relative to a reference level shown in the mathematical formula 1 is the minimum with respect to x
i
.
Zn
=
∑
i
n
(
X
i
-
E
i
)
2
(
1
)
However, it is hard in an actual time span to obtain Ei with which z
n
is the minimum by calculating the equation (1) in every case conceivable. Thus, Ei is generally determined using a procedure called a Viterbi algorithm. A graph shown in
FIG. 23
, which is obtained by redrawing the state transition diagram of
FIG. 22
on an abscissa of time, is called a trellis diagram. In a Viterbi algorithm, z
n
values of two paths are calculated from respective sums till a time n−1 and x
n
input at a time point n and then the path which has the smaller value of z
n
is selected, wherein the term “path” means a directive graph in which a time point and the next time point are connected by a line. As time is retroacted from the present to the past, while this path selection is repeatedly conducted at each time point, paths are converged at a time point expressed by an expression indicating that the paths have been merged. This means that it has been determined that a single path is left and the output corresponding to the path is the detection result. Generally, the z
n
is called a path metric, and the path metric at a time point is called a branch metric.
M
n
(S
0
)=min [M
n−1
(S
0
)+(X
n
+1)
2
, M
n−1
(S
1
)+X
n
2
]
M
n
(S
1
)=min [M
n−1
(S
1
)+(X
n
−1)
2
, M
n−1
(S
0
)+X
n
2
] (2)
The mathematical formula (2) is a recurrence formula in the case where the above mentioned algorithm is adapted to the state transition of FIG.
22
. Mn (S
j
) indicates a sum z
n
of square errors until a time point n in a state S
j
at the time point n and min [a, b] is a function indicating the minimum between a and b. Path metric values for two respective paths at an immediately preceding time point when inputs are respectively made to states S
0
and S
1
are calculated, whereupon the smaller one is selected and thereafter the path metric value is renewed.
FIG. 23
illustrates an example in which selection of a path is conducted at each time point using the mathematical formula (2). A thick line indicates a path conceivable at a time point. Two paths, respectively thick and thin, are present between time points 0 to 6 and there is no determined path, but a merged path is present at time points of 7 and 9. After merging, an output value q
l
corresponding to the path is successfully output and thereby the most likelihood detection can be performed.
In order to actually constitute a circuit, (y
n
+1)
2
, y
n
2
and (y
n
−1)
2
are generated in a branch metric calculating circuit
21
, such as shown in FIG.
4
. Subsequently, path metric values Mn (S
0
) and Mn (S
1
), and the branch metric value are added, a comparative calculation is conducted and thereby one of the paths input to a state is selected and further the path metric value of the path is used as the new path metric value. This operation is repeated and thereby paths are merged into a single path to detect the most likely path. A comparator output indicating selection information of a path is stored in a path memory
23
and detection of information can be conducted by outputting bit information corresponding to a determined path before a merging point.
In a rewritable optical disk device in its first generation, a mark position recording system is adopted, wherein data is recorded on a disk medium by being converted to a length between pits and information is detected by a peak detection system. In peak detection, a readout signal is differentiated and subjected to zero-cross and thereby the middle position of a recording pit is detected, so that a level fluctuation has never been problematic. When a high density in recording is used, a regeneration amplitude is reduced in the mark position recording system and thereby information cannot be detected with high reliability. For this reason, the mark edge recording system as used in CDS, in which a length of a pit itself bears information, has begun to be used in optical disks. However, when information recorded in this recording system is regenerated, since a peak detecting system cannot be used, a level detection is generally used in which 0 and 1 is decided with a threshold value as a reference. In this case, level fluctuation directly gives a wrong influence and it can be said that this level fluctuation is one of the significant causes for disturbing the process of achieving a higher density.
Especially in the case of an optical disk, a direct current level included in regenerated information has a chance to fluctuate.
(1) There i
Burd Kevin M
NEC Corporation
Ostrolenk Faber Gerb & Soffen, LLP
Pham Chi
LandOfFree
Device and method for detecting information does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Device and method for detecting information, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Device and method for detecting information will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2454054