JPS62173820A - Error correcting method - Google Patents

Abstract

PURPOSE:To correct not only a two-symbol error certainly but a three-symbol error by estimating error on a basis of syndromes with respect to an internal code and an external code and correcting error only when both codes are coincide with respect to estimated error and error cannot be corrected. CONSTITUTION:When 4 symbols are fetched, two syndromes of a code Q are completed and are transferred to an error position generating circuit 9 from a RAM 3; and if error exists, operation is performed to check whether it can be corrected or not. If there is not contradiction between internal and external codes with respect to error and error can be corrected, the position and the range of error are operated and held. If there is contradiction and error can be corrected, similar operation is performed. When correction is determined, error is sent to an error correcting circuit 8 and is corrected. If there is contradiction between internal and external codes with respect to error, correction is performed with one code and the validity is discriminated with the other. Thus, not only two-symbol error is corrected certainly but also the probability of erroneous correction for three-symbol error is suppressed.

Description

DETAILED DESCRIPTION OF THE INVENTION A. Field of Industrial Application The present invention relates to an error correction method for a composite error correction code system having an inner code and an outer code.

BACKGROUND OF THE INVENTION In addition to music data, subcode data is recorded on a digital audio disc called a compact disc. This subcode data consists of eight subcode channels from P to W, and the presence or absence of a song,
The performance time of the song and image information about the song are recorded.
The six subcode channels R to W are used to record image information, and 24 symbols (one symbol consists of six channel bits H to W) are divided into one unit (called back). ). Its configuration is as shown in Figure 3, and it has two parities called P parity and Q parity as error correction codes.
Two Reed-Solomon codes are used.

A code with Q parity (referred to as a Q code) consists of four symbols, two of which are parity. On the other hand, a code with P parity (referred to as a CP code) consists of 24 symbols, and 4 symbols serve as parity. Since the Q code is included in the P code (the Q code is data in the P code), it is called an inner code with respect to the P code, and the inverse KP code is an outer code with respect to the Q code. It is called.

In the Q code, errors in up to 2 symbols out of 4 symbols can be detected, and errors in 1 symbol can be corrected.
In the code, the relationship between the estimated number of errors and the actual number of errors is as shown below since the minimum code distance of the Q code is 5 symbols.

Margin table below: 1 In the P code, errors in up to 4 symbols out of 24 symbols can be detected, and errors in 2 symbols can be corrected.
In the P code, the relationship between the estimated number of errors and the actual number of errors is as shown below, since the minimum code distance of the P code is 5 symbols.

Table 2 A conventional decoding method for a subcode code having the above-described configuration will be explained.

(1) The simplest decoding method does not decode the Q code, but
This is a method of decoding using only codes.

This is because the Q code is the inner code button of the P code, and if error correction is done when decoding the P code, all data including the Q code should be correct. If so, it means that the error has been incorrectly corrected due to an error that exceeds the correction ability.

Here, a decoding method that uses only the P code to correct a 1-symbol error when it is estimated, and displays an error when an error of 2 or more symbols is estimated is called the P-decoding method. A decoding method that corrects a two-symbol error when it is estimated and displays an error when three or more symbols are estimated is called the P3 decoding method.The OP decoding method has the ability to correct a one-symbol error. Although it is not reliable, it can correctly detect errors up to 3 symbols, and erroneous correction will not occur unless there are errors of 4 or more symbols.

The V decoding method can correct up to 2 symbol errors, but up to 5 symbol errors can be corrected.
Erroneous corrections may occur if symbol errors occur.

(2) The next method uses the Q code, which is the inner code, after P decoding or This is a method of checking the presence or absence of erroneous corrections, or a method of correcting data symbols of a Q code.

The method of checking the validity of error correction using Q code after P' decoding is called P3q decoding method, in which the data symbol of Q code is corrected after P' decoding (this method is used when a 1-symbol error is estimated). and display an error message when two or more symbols are estimated to be in error.
These plq decoding methods and P
The point that the 1Q decoding method can correct up to 2 symbol errors is P“
It is similar to the decoding method, but if a symbol error occurs, the error can be detected, but there is a possibility of erroneous correction, and in the case of erroneous correction, it may be possible to detect this erroneous correction.

For example, if there is one error symbol in the Q code and two error symbols in the P code at positions that cannot be coded (if there are 3 symbol errors in total),
``As a result of estimating a 2-symbol error during decoding and correcting it (erroneous correction), one error symbol increases in the position of the P code but not the Q code, then in the P''q decoding method, Q
Errors in the code can be detected correctly, and therefore errors can be detected in the entire code (ie, P code). on the other hand,
Although the P1Q decoding method can detect errors in the Q code and correctly correct them, the code as a whole is determined to be in error. If so, after performing No. 21, the error should be 0 or 6 or more (if there was an error of 1 or 2, it should have been correctly corrected and become 0), and the error would be detected in the subsequent Q decoding. is detected, which means that Q
This indicates that there are errors that cannot be corrected by decoding, and even if one error symbol in the Q code is correctly corrected, the code as a whole will be in error.

On the other hand, if there is a 3-symbol error in the P code and an error of 3 or more symbols occurs in the Q code after decoding P1, it may be mistakenly estimated as 0 error when decoding the Q code. (3) The next method is to first perform error correction using the Q code, and then perform error correction using the P code. A method that uses Q code to correct one symbol error and P code to correct one symbol error is called QP decoding method. Corrects one symbol error using Q code and uses P code. The method that corrects the two-symbol error will be referred to as the "QP" decoding method.

In the case of the QP decoding method, a 2-symbol error can be corrected only if there is a 1-symbol error in the Q-code, Kame D, and a 1-symbol error in a position other than the Q-code in the p-code.
For other 2-symbol errors, error detection is only performed correctly. For example, if a 2-symbol error occurs in the Q code jt, it is incorrectly corrected as a 1-symbol error in the Qi code and a 3-symbol error occurs. Even if this happens, a 3-symbol error will not be erroneously estimated as a 1-symbol error and erroneously corrected during P decoding, and only error detection will be performed correctly. In the case of the QP1 decoding method, it is possible to correct 2-symbol errors, but if there are 2-symbol errors in the Q code, there is a possibility of erroneous correction.
Result of erroneously estimating 1 symbol error during decoding and erroneously correcting it 3
This is a case where a symbol error occurs, and upon decoding P'', it is erroneously estimated as a 2-symbol error and erroneously corrected.

Next, considering the 3-symbol error, there are 2 symbols in the Q code.
If there is an error of more than one symbol, in the QP decoding method, the error is incorrectly estimated as one symbol error during Q decoding and corrected (erroneously corrected), resulting in a four symbol error, which is converted into one symbol error during P decoding. In the oop detection method, where there is a possibility of erroneously estimating an error and erroneously correcting it, if there is one symbol error in the Q code or two symbol errors in a position other than the Q code in the P code, the error cannot be corrected correctly. However, in other cases, if there is a four-symbol error that may be incorrectly corrected at the time of qt1 code and P' decoding, Even with decoding method 1c, there is a possibility of erroneous correction.

FIG. 4 shows a diagram for comparison of the decoding methods described above.

In Figure 4, [Q] has one symbol error in the Q code, and [P] has one symbol error in the P code but not in the Q code.
There are 4 symbol errors, (QP, l is 1 symbol error in the Q code, and 1 symbol error in the P code but not in the Q code (total 2 symbol errors), (pp ) indicates that there is a two-symbol error in a position in the P code but not in the Q code [The same applies hereafter. The 00 mark indicates that no error can be detected or corrected, and the - mark indicates that an error can be detected. , an X mark indicates that an error can be detected but there is a possibility of an incorrect correction, a ■ mark indicates that a correct correction may be made, but there may be an incorrect correction, and a Δ mark indicates that an error can be detected but there is a possibility of an incorrect correction. These results indicate that there is a possibility that this error correction may occur, and that there are cases where this erroneous correction can be detected.

Problems to be Solved by the Invention The present invention attempts to propose a decoding method that combines the advantages of the above-mentioned QP" decoding method and P5q decoding method. That is,
A 2-symbol error can always be corrected, a maximum of 6-symbol errors can be corrected, and a 5-symbol error can be completely corrected to suppress the possibility of erroneous correction.

20 Means for solving the problem First, in the first step, the error is estimated based on the syndrome for both the inner code and the outer code.O If the error estimated in the first step is for the same code, If there is no contradiction, the following steps are provided. That is, if it is estimated that there is no error, the code is treated as a valid code as it is (Step 2-1), if the error can be corrected, it is corrected and the code is made valid (Step 2-2), and if the error cannot be corrected, it is treated as an invalid code. (Step 2-3).

If there is a conflict, the next step will be taken.

In other words, if it is estimated that error correction is impossible, the code is invalidated (step 3-1), and if it is correctable, the error with a high probability of occurring is estimated, and after error correction using one code, the other code is A step (step 3-2) is provided to prevent erroneous correction by determining the validity of error correction using the following.

In the present invention, in order to prevent erroneous corrections as much as possible, the correction is postponed and the error is first estimated based on the syndrome for the same code. It is limited to cases where the error symbols of the codes match and on the other hand it becomes uncorrectable.Therefore, compared to the Qpl decoding method, there is no error correction when there is a (QQ) error. It is possible to correct up to 6 symbol errors (in the case of (QPP) errors), P'Q decoding method, Pj
It is superior to the q decoding method, and its error detection ability when there is a 3-symbol error is comparable to the P''Q decoding method and the Pgq decoding method. In the figure, Q+ynd is the syndrome generation and error estimation of the Q code, Paynd is the syndrome generation and error estimation of the P code, and Qlc.

rr is 1 symbol error correction of Q code, Ploorr is 1 symbol error correction of P code, P2corr is 2 symbol error correction of P code, P2 (18-oorr is 2 symbol error correction
Each shows that the P code after symbol error correction is restored to its original state. eQ is the number of errors estimated by Qsynd, e
P indicates the number of errors estimated by Psynd. X
α indicates the error position, Yα indicates the magnitude of the error (the position of the error pit within one error symbol), and α indicates the estimated one error symbol. Note that if only one of the estimated errors XP1 and XP2 is within the Q code, it is XPI.

0xpc to be assigned to the error in the Q code
(q) indicates that the error position of the error symbol P is within the Q code, and Xpc(p) indicates that the error position of the error symbol P is within the P code but not within the Q code. OK indicates that there is no error or the code is a usable valid code that has been correctly corrected; NG indicates that there is an error that cannot be corrected and the code is invalid. (Q), (P), (QP], etc. indicate the number and position of error symbols as in the case explained above. On indicates logical product.0 To explain the flowchart in FIG. 1, Q based on the data received first
Find the syndrome for each code and P code, and if the number of errors is within the error correction capability range, the error position (
XQ, XPI XPl, XP2) and error magnitude YQ.

Yp, YPII Yp2 ) (Step 1
).

If the estimation of the position and magnitude of the error described above does not result in a contradiction between Qs7nd and Psynd, if there is no error according to the estimation of Qgynd and Psynd, the code can be corrected as it is, and route a = step 2-1). If there is an error, please correct it) b, dl h
t jl 0 Pt s step 2-2), and if it is estimated that the error cannot be corrected, it is treated as an invalid code and an error is displayed (route f).

V Nistep 23).

On the other hand, if a contradiction occurs between Qsynd and Psynd,
If it is estimated that the error cannot be corrected, the error is displayed as an invalid code (Route Or 8+f*i+u 2-step 3-1), and if it is estimated that the error is correctable, an error with a high probability of occurring is estimated and one code is After correcting the error using the code, use the other code to determine the validity of the error correction and prevent erroneous correction (roux) ke ni step 5-2
).

Some routes will be explained in more detail below.

[Route c) In Qsynd, the error is 0 (eQ=
0), in Paynd, the error symbol is 1 (eP=1
), there is no error for P code C or lower (denoted as (P)) in a position that is not a Q code", that is, "Q code C or less (denoted as (Q))) K1
This is the case where it is estimated that there is a symbol error, and O,a
There is a contradiction between y n d and Psynd. The reason why such a contradiction occurs is that the error is estimated incorrectly in one or both of Qsynd%P8ynd. The error estimation is due to the large number of error symbols.
From Tables 1 and 2 described above, the actual error in route c can be re-estimated as follows.

(1) (Q) has no error (eQ=0 is correct), and (Pl has an error of 4 or more symbols (ep=1 is an error), or (21(0) has an error of 3 or more symbols ( eo=
Q is an error), and [P] K1 symbol or more error [P
There are a total of 4 or more symbols of errors in the code.
= 1 is also incorrect) Even in the case of the side deviation described above, the number of errors is large and exceeds the correction ability, and the error is displayed as an invalid code (k-) e) In Qsynd, tQ) has no error, In Psynd, this is the case when two symbol errors are estimated in [), and the cause of such contradiction can be estimated as follows: 0 (11(Q) is an error ((eQ=0 is correct), and Error of 5 symbols or more in l (ep=2 is error), or error of 3 or more symbols in (21(Q) is 5b(eQ=
(0 is an error), and (p) has an error of 0 or more symbols. In this case, too, the number of errors is large and cannot be corrected.

[Route P] In Qaynd, (Q) is estimated to have a one-symbol error, and in Psynd, it is estimated that there is no error.The cause of such contradiction can be estimated as follows.

+11 (Q) There is a K1 symbol error (eQ;1
is correct), and (4 or more symbols error in Pl [P
There are errors totaling 5 or more symbols in the code) (ep=0
is an error), or there are errors of (21[Q)K2 symbols or more (eQ=
1 is an error], and (P) has an error of 5 or more symbols (e
p=0 is incorrect) In this case as well, the number of errors is large and cannot be corrected.

[Lou) i) In Qsynd, (Q) was estimated as a 1 symbol error, and in Psynd, 1 symbol error was estimated, but this is the case when the error position or error size does not match.The cause of such discrepancy is as follows. It can be estimated exactly.

(1) (1 symbol error in QJ (eQ = 1 is correct), 5 or more symbols error in lpJ (ep = 1)
is an error), or there is an error of (21[Q)Vc2 symbols or more (ea
= 1 is an error), (2 or more symbols error in PJ [total 4 or more symbols error in P code fig (ep = 1
(Error) In this case as well, the number of errors is large and cannot be corrected.

(Lou) k-1, Lou) k-os) In Qsynd, (Q) is estimated to have one symbol error, and Psynd
In (i), it is estimated that there are two symbol errors, and there is a contradiction between the two.

If there is actually a two-symbol error in [QJ,
Qsynd may cause an erroneous estimation of e Q = 1. On the other hand, if (Q) has only one symbol error, Ps
In ynd, [QJ has a 2-symbol error], which causes an erroneous estimation.
This is a case where there are two or more symbol errors (a total of three or more symbol errors in the p code). Considering the reliability of both, a two symbol error is more likely to occur than a three symbol error, so here we use Psynd. I decided to trust
We estimate that there is a 2-symbol error in (Q)0, so,
These two errors are corrected by p2corr0. Then, to check the reliability of this correction, Qsy
If nd is executed and it is confirmed that there is no error, it is set as a valid code. However, if qsyna estimates that there is still an error, it is assumed that P2cmrr is an incorrect correction and Psynd is an incorrect estimation. . In this case, it may be invalidated as is, but if there is enough time to perform error correction again, the error-corrected code is restored to its original state, and thereafter, the code goes to route 0 and route n.

[Route 0, Roux) n) The error estimation in this case is as follows.

(There is a 1 symbol error in 11(Q) (e.

= 1 is correct), (P) error of K2 symbols or more (e
p = 2 is an error and e P 〉3 is correct), or there is a 2 symbol error in (21[Q) (eq Schiller e
p=2 is wrong, 8P is also correct).

Or if there is 1 symbol error in Ql and 2 symbol errors in (P), that is, in the case of (QPP) error [(11 above)
], which can be corrected by the route p described below.

(Lou) p) First, the 1 symbol error of (Ql is corrected with Ql corr, and then the 2 symbol error of (P) is corrected with p200rr, thereby becoming an effective code.

[Route q1 Route r] This is true if there is no (QPP) error, and the following Scherrer estimation can be made. (21 (Q) has two symbol errors, (P) has one or more symbol errors. In the above cases, the number of errors is large and correction is impossible.

[Lou) a) In Qsyna, errors of two or more symbols are estimated [uncorrectable as a Q code], and Psy
In nd, it is estimated that there is a 2-symbol error in [Q], and there is no contradiction, and it can be corrected using the P code. Furthermore, after correction by the P code, it can be checked with Qsynd that there are no errors in the Q code.

[Route t] This is the case where the error is estimated by (lsynd) for the second time, which indicates that p2cOrr was a miscorrection. Estimating the error in this case, (if there are two or more symbols error in Ql) and t
This is a case where there is an error of one or more symbols in p+ (a total of three or more symbols in the p code) (Bp=2 is an error), and error correction is impossible.

[Route u] This is the case where there are no two error positions obtained by Psynd within the Q code. The error in this case is that [q) has two or more symbol errors, and (P
) is estimated to have an error of one or more symbols (a total of three or more symbols in the p code) (ep=2 is an error), and error correction is impossible.

[Lou) V) In Qsynd, errors of 2 or more symbols are estimated, and in Pgynd, the number of errors ep is 0.1.
This is a case where it is estimated to be 3 or more, and the estimated error is as follows.

(11eQ≧2, eP=00 (Ql has 2 or more errors, IP) has 3 or more symbols error [total 5 or more symbols in P code] (21eq≧2, eP=1) Case (Q) has two or more symbol errors, [P] has two or more symbol errors [4 or more symbols in total in the P code] (31e Q l 2, ep = 3 (2 symbols in Ql) There are the above errors, and (P)K1 symbol or more error [total 3 or more symbol errors in P code] In the above example, there are 5 or more symbols errors, and it is not a [QPP,) error. , error correction is not possible.

In addition, in [Route 8], it is not necessary to determine whether the error position obtained from PsyndK is within the Q code (therefore, there is no need for roux).
This is because error checking is performed using d.

Now, it is preferable to use a general-purpose intelligent circuit such as a microcomputer as a decoding circuit for realizing each step of the present invention shown in the above-described flowchart.

FIG. 2 shows an embodiment of such a decoding circuit. The interface circuit +11 is a circuit for passing the subcode signal output from the compact disc player and its clock signal to each subsequent circuit. The write address generation circuit (2) includes a counter that counts up each time one symbol is received and generates an address for storing the received symbol in the buffer RAM (3), and symbol counters for each of the Q and P codes. The de-interleaving circuit (4) is an interleaving decoding circuit, which modifies the write address and regularly intersects the store addresses of the received symbols, thereby changing the order of each symbol originally recorded on the disk in the interleaved order. This is a circuit that rearranges the circuits as before. The read address generation circuit (5) sends symbols to the application circuit of the subcode by using the buffer RA.
It consists of a counter that generates an address for reading symbols from M (31).A correction address generation circuit (
6) is composed of a register for generating the RAM address where the symbol to be corrected is stored when an error symbol is found. Buffer RAM (31
is used to store the received symbol until it is corrected and the symbol is passed to the application circuit side.
Stores symbols. Syndrome generation circuit (7
) is a circuit for generating syndromes, which are parameters necessary for error estimation and correction. When an error symbol is found and the error correction circuit (8) is to be corrected, the error correction circuit (8) latches data indicating the magnitude of the error on the local data bus generated by the error magnitude/error position generation circuit (9). , buffer RAM on the data bus (
Exclusive OR is performed bit by bit with the error symbol data read from 31, and this data is written to the same address of RAM (31).The error size/error position generation circuit (9) determines the position of the error symbol. (indicating which symbol in the code word) and the difference between the correct symbol and the error symbol (the difference between the correct symbol and the error symbol).

Microprogram FIOM (lα is a ROM in which the error estimation/correction sequence is programmed, and the FIO
The data of M is output to the control bus and becomes various control signals to control each circuit. The program counter συ is a counter that generates an address to be given to the microprogram FIOM (FIOM 11), and includes a circuit for resetting, jumping, etc. in response to a control bus signal.

The clock/timing signal generation circuit U generates clock signals and various timing signals for each circuit.

Next, the operation of the above-mentioned circuit will be explained.

(11 When a compact disc player reads subcode data from the disc, it outputs a clock and 1 symbol (6-bit subcode). Each bit of 1 symbol (R18+T+UlvtW) is called a channel. , these 6 channels of subcode data are first output to the data bus by the interface circuit (11).
M (address 31 is the write address generation circuit (2).
It is connected to the de-interleave circuit (4), and one symbol data on the data bus is sent to the write address generation circuit (2).
- 0 stored at the address indicated by the de-interleave circuit (4) Note that the counter of the write address generation circuit (2) is incremented. Furthermore, each time the decoding circuit receives one symbol, the symbol on the data bus is
(2) Data for four symbols is taken in from the compact disc player, and two syndromes for the Q code are completed. Syndrome data is transferred to the error magnitude/error position generation circuit (9) by the local bus, and this circuit (9) determines whether both syndromes are "0" (i.e., no error). A judgment is made.

(3) If there is an error, perform calculations based on the syndrome to check the possibility of correction (whether the error is within the correctable range).

(4) If it is determined that it is possible to correct the error,
The magnitude of the error is calculated and the result of this calculation is held in the circuit (9).

(5) At this point, the error size/error location generation circuit (9) determines whether it is an error, whether it is an error, whether it can be corrected, and if it is correctable, whether the correct error location has been found, etc. Each of the seven lags is latched and connected to the control bus output from the circuit (9).

(6) Now, when the data for 24 symbols from the compact disc player is taken into the buffer RAM (31), as in (2) to 5) above, the four syndrome data of the P code are transferred to the a-cal data. It is transferred from the bus to the error magnitude/error location generation circuit (9), and determines whether there is an error (are all four syndromes vlOI')?
If it is an error, it is latched in the circuit (91) and output to the control bus. Flags such as whether it can be corrected, whether it is one error or two errors if it is correctable, and whether the correct error position has been found are output to the control bus. If so, the error location and error magnitude are maintained in the circuit (9).

(71 Now, according to the algorithm of the present invention, each flag on the control bus is examined, and if necessary, the circuit (9)
The syndrome can be changed using the error position, error size, etc. held within the system, or the error position and error size can be calculated again.

(8) Next, when it is decided to perform correction according to the algorithm of the present invention, the error size/error position generation circuit (
The error position data held in 9) is transferred to the correction address generation circuit (6) via the local data bus, and is added to the counter value from the write address generation circuit (2) to store the symbol to be corrected. buffer RAM
(0 becomes the address of 31. Here, first, the buffer RAM (
31 enters the read mode, and an error symbol is output from the RAM (31) and placed on the data bus.
Data regarding the magnitude of the error is transferred from the circuit (9) to the error correction circuit (8K) via the local data bus.

In the error correction circuit (8), the data of the symbol with an error (error symbol) outputted from the buffer RAM (31) and the data indicating the magnitude of the error are subjected to an exclusive OR operation and latched.This data is corrected. Next, this data is placed on the data bus and transferred to the buffer 7 RAM (which is ready for writing).
Contains 311C%.

(9) In order to show the corrected subcode data to the application circuit side, the read address is given to the buffer RAM (31K and sent out one symbol at a time on the data bus Vc) in synchronization with the subcode clock signal output from the compact disc player. In this case, the error flag on the control bus is also passed to the application circuit side as necessary.

6. Effects of the invention The effects of the decoding method according to the invention are as shown in the last column of FIG. In other words, up to 2 symbol errors can be corrected without fail.
And (QPP) 3 symbol errors can be corrected 0 And the possibility of miscorrecting 3 symbol errors can be suppressed 0 Therefore, this is a decoding method that combines the advantages of the QP1 decoding method and the P″q decoding method. It turns out that 0

[Brief explanation of drawings]

FIG. 1 is a flow chart diagram showing a decoding method according to the present invention;
FIG. 2 is a diagram showing a decoding circuit according to the present invention, FIG. 3 is a diagram showing a format of a subcode signal, and FIG. 4 is a diagram showing a comparison of each decoding circuit. 0(3) is a buffer 7RAM,
(71 is a syndrome generation circuit, (8) is an error correction circuit, (9) is an error size/error position generation circuit,
The residence α is a micro program ROM

Claims (1)

[Claims] (1) A composite error correction method having an inner code and an outer code, which includes a first step of estimating errors based on syndromes for each of the inner code and outer code, and an error estimated in the first step. If there is no contradiction in the inner code and outer code, and it is estimated that there is no error, the code is treated as a valid code (step 2-1).
, if the error can be corrected, correct it and make it a valid code (step 2-2); if the error cannot be corrected, make it an invalid code (step 2-3); and the estimation in the first step above. If the detected error is inconsistent between the inner code and outer code, and it is estimated that the error cannot be corrected, it is treated as an invalid code (step 3-2), and if it is correctable, the error with a high probability of occurring is estimated. An error correction method comprising the steps of: (3-2) performing error correction using one code and then determining the validity of the error correction using the other code to prevent erroneous correction.