JPS6290723A - Encoding and decoding devices for postscript type memory - Google Patents

Abstract

PURPOSE:To obtain an encoding and decoding devices for postscript type memory which is large in total writing information quantity per one storage unit as compared with conventional ones, by specifying effective code words selected against information symbols. CONSTITUTION:The code words against information symbols (x) of the (T+1)-th generation (T-0, 1,...) to be written in a block form a binary sequence of N bits in length and are not the continuous sequence of '1' of the length N. Moreover, the code words can be transformed from the code words written in the block at the T-th generation and set under a condition where the information symbols can clearly be decoded from the code words of the (T+1)-th generation. The code words read by a block reading section 6 are inspected whether they are effective code words or noneffective ones and, when they are effective code words, decoded symbols corresponding to the effective code words are stored in a decoded symbol buffer. When they are noneffective code words, a decoded block counter 10 is recurrently actuated after the content of the counter 10 is increased by one.

Description

[Detailed Description of the Invention] Industrial Application Field This device 8A rewrites information on a write-once memory such as a punched card, paper tape, or write-once optical disk, which has an irreversible property that it cannot be erased once information is written. The present invention relates to an information encoding/decoding device that makes it possible to read rewritten information.

Conventional technology Write-once memory such as optical disks stores binary information by irreversibly changing the physical properties of the surface, such as by drilling holes in which the recording medium is partially melted. Normally, it is not considered to write new information again to a storage medium that has been previously written. However, for efficient and economical use of storage media, it is desirable to have a storage medium that allows information to be rewritten. For this purpose, codes have been devised that allow information to be rewritten several times in write-once memory. In the following, we will refer to this code as WOM (Write
once memory) code. Further, when the code length of the WOM code is N, the storage area of the write-once memory is divided into N bits each, which is called a block. The following is information and control 0-/l
/ (Information and control
), Vol, 55. This is an encoding/decoding method proposed by Rivest and Shamir in 1982.

Table 1 shows the first generation of WOM code (code subscriber) where the number of information symbols M is 4, the code length N is 3, and the number of written generations is 2.
, and a code word of the second generation (code word B). It is assumed that all 0s are recorded on the recording medium in the initial state.

In the first generation, a code requestor (X) is written for the information symbol X. Here, writing 0 means not causing any change to the storage medium, and writing 1 means causing some irreversible physical change to the storage medium. In the second generation, the information symbol y is
If different, code word B(y) is written, and if y=x, no writing is performed.

In this encoding/decoding method, the T+1st generation (T=0.1) codeword to be written in a certain pool is (
1) It is a binary sequence of length N bits, and (II
) It is possible to transition from the code word written in the 7th generation to that block, and (1)
It is set so as to satisfy the condition that the information symbol of the T+1th generation can be uniquely decoded from the code word of the T-)-1th generation.

Let us show a concrete example of Table 1 encoding. In the first generation writing, the write code word string of 4 blocks for the information symbol string 00 011°11 is 000 100010 0
01. After this, when 111°1000 is input as the second generation information symbol string, the written code word string for that block is 110 101 010111
becomes.

On the other hand, it is clear from Table 1 that the information symbol is uniquely decoded from this written code word string.

Problems to be Solved by the Invention However, in the above encoding/decoding method, there is something that should be improved in terms of the total amount of written information.

In other words, in the above encoding/decoding method, the amount of information written per memory rate is (1o
g 24 )/3 bits, and the total amount of written information is only % bits per storage rate.

Furthermore, in other WOM codes cited in the above-mentioned literature, the total amount of written information for one memory rate is at most 1.6.
It's only 04 bits.

In view of this point, it is an object of the present invention to provide an encoding/decoding device for a write-once memory in which the total amount of written information per storage rate is larger than that of the conventional device.

Means for Solving the Problems In the encoding/decoding device for write-once memory of the present invention, the T+1 generation (T=o, 1.
), the code word for the information symbol X is (1) a binary sequence of length N bits, and (11) 1 of length N
(III) It is possible to transition from the codeword written in the 7th generation in the block), and (IV) The information symbol is uniquely derived from the codeword of the T+1th generation. is set under the condition that it can be decrypted. In the following, a codeword that satisfies this condition will be referred to as an effective codeword.

Then, the encoding device reads the contents of the write block of the write-once memory when reading the information symbol χ,
If a valid codeword exists, it is written in that block; if no valid codeword exists, the block is made an invalid block and a continuous sequence of 1s of length H is written as an invalid codeword, and code word,
This is to write to a block that can be written after that block. To be sure, the decoding device first checks whether a certain block is an invalid block or not, and if it is a valid block, it uniquely determines the information symbol from the code word written in that block; If it is a block, the above decoding process is recursively performed for blocks after that block.

Operation: The write-once memory encoding/decoding device according to the present invention relies on the existence of the invalid code word as described above (TI), (1).
The selection range of code words that satisfy the condition is expanded, and as a result, the total amount of information written per one storage unit becomes larger than that of the conventional method.

Example Hereinafter, the present invention will be explained in detail based on examples.

Table 2 shows the correspondence between information symbols and code words of the encoding/decoding apparatus according to one embodiment of the present invention.

First, an outline of the encoding process will be explained. It is assumed that all 0s are recorded on the recording medium in the initial state.

In the first generation, an initial code word A(XI) is written for an information symbol X in a certain block S1. In the second generation, if the information symbol x2 to be written in block S1 is different from xl, update code word B(χ2) is written,
If =XI, no writing is performed. In the third generation, if the information symbol x3 to be written into the block is equal to X2, no writing is performed. On the other hand, if Xs is X2
If the codeword being written is the initial codeword, an updated codeword B(Xs) is written. If an updated codeword has been written, an invalid codeword 111 is written in that block S1, and from the next block S2 onwards, a block S1 in which an initial codeword has been written is searched for, and Sl
The updated code word B (xs) is written in . At this time, an invalid code word 111 is written from Sl to 51-1. Fourth
Information can be written even after the generation if there is a block to be written to the recording medium.

On the other hand, the outline of the decoding process is as follows. First, it is checked whether a certain block is an invalid codeword or not, and if it is a valid codeword, a decoded symbol is obtained from the block codeword-information symbol correspondence table, and this is output. On the other hand, if it is an invalid codeword, the block counter is incremented by one and the above decoding process is recursively performed in the next block.

Table 2 FIG. 1 is a diagram showing the configuration of a write-once memory encoding device in one embodiment of the present invention.

In FIG. 1, 1 is an information symbol input section which inputs a ternary information symbol consisting of 0, 1°2. 2 is an information symbol buffer that temporarily stores the information symbol string input from the information symbol input section 1;

3 is the encoding buffer 7 counter and the information symbol buffer 2
The number of the information symbol currently being written is stored. Reference numeral 4 denotes a write-once memory, which can be written and recalled in units of focus.

Further, in this embodiment, it is assumed that writing and reading are possible in units of blocks each consisting of 3 bits. 5 is an encoded block counter that stores the block number of the write-once memory 4 that is currently being written or read. A block reading unit e reads the stored contents of the block specified by the encoded block counter 6 in the write-once memory 4. 7 is a code word selection unit which selects a code word to be written from the code word read by the block reading unit 6 and the information symbol designated by the encoding buffer counter 3 from the information symbol buffer 2.

The relationship between the read code word, information symbol, and write code word is the same as shown in the Markov chain diagram of FIG. In the figure, the base point of the arrow is the code word read by the block reading unit 6, the information symbol shown next to the arrow, and the end point of the arrow is the code word to be written. If the codeword to be written is an effective codeword, an effective codeword writing unit, which will be described later, is activated. On the other hand, if the codeword to be written is an invalid codeword, it recursively activates itself after activating an invalid codeword writing unit, which will be described later. Reference numeral 8 denotes an effective code word writing unit which writes the effective code word selected by the code word selection unit 7 into the block specified by the coding block counter 6 in the write-once memory 4, and writes the effective code word selected by the code word selection unit 7 into the block designated by the coding block counter 6 in the write-once memory 4. Increase the contents of 6 by 1.

Reference numeral 9 denotes an invalid code word writing unit which writes an invalid code word into the block designated by the encoded block counter 5 in the write-once memory 4 and increments the contents of the encoded block counter 6 by one.

FIG. 3 is a diagram showing the configuration of a write-once memory decoding device in Embodiment 1 of the present invention.

In the figure, 10 is a decoding block counter, which stores the block number of the write-once memory 4 that is currently being read. Reference numeral 11 denotes a decoded symbol conversion unit, which has a codeword and decoded symbol conversion table shown in FIG. 4, and checks whether the codeword read by the block reading unit 6 is a valid codeword or an invalid codeword. In the case of a code word, the corresponding decoded symbol is stored in the decoded symbol buffer 7, which will be described later, and then the contents of the decoded buffer counter, which will be described later, are set to 1.
decoded block counter 1 in case of invalid codeword.
After incrementing the contents of 0 by 1, recursively activate the program. 1
2 is a decoded symbol buffer, which is a decoded symbol converter 11;
The decoded symbols set by are stored sequentially. 13
is the decoding buffer counter, decoding symbol buffer 1
The second one stores the storage position number of the decoded symbol currently being written.

I will explain according to the first one.

[Encoding device]

(1) The information symbol string input from the information symbol input section 1 is temporarily stored in the information symbol buffer 2. At this point, the contents of the encoding buffer counter 3 and the encoding block counter 5 are each ○.

(2) If the content of the encoding buffer counter 3 is larger than the effective length of the information symbol buffer 2, the encoding process is terminated. Otherwise, proceed to (3).

(3) If the contents of the encoding block 07 counter 5 are larger than the effective capacity of the write-once memory 4, the encoding process is terminated. Otherwise, the block reading unit 6 reads the block specified by the encoded block counter 5 from the write-once memory 4.

(4) The code word selection unit 7 selects a write code word from the information symbol specified by the encoding buffer counter 3 in the information symbol buffer 2 and the code word read by the block reading unit 6.

If the written codeword is a valid codeword, proceed to (4); if it is an invalid codeword, proceed to (6).

(5) The valid code word writing unit 8 writes the valid code word selected by the code word selection unit 7 into the block specified by the coding block counter 6 in the write-once memory 4, and writes the valid code word selected by the code word selection unit 7 into the block specified by the coding block counter 6 in the coding buffer 7 counter 3 and the coding block. The contents of block counter 5 are each incremented by 1. After this, proceed to (2).

(6) The invalid code word writing unit 9 writes the invalid code word into the block specified by the encoded block counter 5 in the write-once memory 4, and increases the contents of the encoded block counter 6 by one. After this, proceed to (3).

[Decoding device]

('7) Set the contents of the decoding banff 7 counter 13 and the decoding block counter 10 to 0, respectively.

(8) If the content of the decoding buffer counter 13 is larger than the effective length of the decoding symbol buffer 12, the decoding process is terminated. Otherwise, proceed to (9).

(9) If the content of the decoding block counter 1Q is larger than the effective capacity of the write-once memory 4, the decoding process is terminated. Otherwise, the block reading unit 6 reads the block specified by the decoding block counter 10 from the write-once memory 4.

(10) The decoded symbol converter 11 checks whether the codeword read by the block reader 6 is a valid codeword or an invalid codeword. If the read codeword is a valid codeword (
Proceed to step 11), and if it is an invalid codeword, proceed to step (12).

(11) The decoded symbol conversion unit 11 searches for a code word that matches the read code word in the column (a) of the conversion table in FIG. It is stored in the position specified by the decoding buffer counter 13. and a decoding buffer counter 13 and a decoding block counter 10
Increase the contents of each by 1. After this, proceed to (8).

(12) The decoded symbol converter 11 increments the contents of the decoded block counter 10 by one. After this, proceed to (9).

Next, a series of specific operations of the write-once memory encoding device of this embodiment will be described. It is now assumed that the write-once memory 4 consists of eight blocks, and its storage state is as shown in Table 3 (IL). It is assumed that the information symbols shown in Table 3 (b) are input from the information symbol input unit 1 as the first generation input and are temporarily stored in the information symbol buffer 2. It is also assumed that the contents of the encoding buffer counter 3 and the encoding block counter 6 are each 0. First, in this encoding device, the block reading unit 6 reads the contents of the block specified by the encoding block counter 6 from the write-once memory 4 . This content is 000. Then, the code word selection unit selects the effective code word ○ according to FIG. Select 00. Then, the effective code word writing unit 8 writes the effective code word 000 into the block ◯ indicated by the encoded block counter 5 in the write-once memory 4 . Then, the contents of the encoding buffer counter 3 and the encoding block counter 6 are each incremented by 1.

The above-described process is repeated, and when all eight information symbols input from the information symbol input section 1 are written into the write-once memory 4, the first generation encoding process ends. The storage state of the write-once memory 4 at this time is as shown in Table 3 (C).

Next, it is assumed that the input shown in Table 3 (d) is received from the information symbol input section 1 as the second generation information symbol. At this time, by performing the processes (1) to (6) above, eight symbols are written, and the storage state of the write-once memory 4 becomes as shown in Table 3 (61).

Furthermore, it is assumed that the input of Table 3 (f) as a third generation information symbol is received from the information symbol input section 1. At this time, by performing the above-mentioned processing, six valid codewords and three invalid codewords are written, and the storage state of the write-once memory 4 becomes as shown in Table 3).

Furthermore, it is assumed that the input of Table 3 (h) as a fourth generation information symbol is received from the information symbol input unit 1. At this time, by performing the above processing, four symbols of valid codewords and four symbols of invalid codewords are written, and the storage state of the write-once memory 4 becomes as shown in Table 3 (i>).

Table 3 (a) ooo○oo ooo ooo ooo o
oo ooo oo.

(b) 01201120 (C) 00000101000000100101
00o○ (d) 02211201 (e) 00010101000100110101
1001 (f) 1 1 1 2 0 (g) ○01 111 110001101 11
1 111011 (h) 2110 (i) 101111 110001111111
111011 Furthermore, the write once memory decoding device of this embodiment performs the processes from V) to (12) described above.
When the generation is as shown in Table 3 (C), the decoded symbol string of (b) in Table 3 is stored in the decoded symbol buffer 712.
Store in. In addition, when the storage contents of the write-once memory 4 are as shown in Table 3 (e) in the second generation, (
0) is stored in the decoded symbol buffer 12. Further, when the storage contents of the write-once memory 4 are as shown in Table 3 (g) in the third generation, the decoded symbol string "1112o" is stored in the decoded symbol buffer 12. Further, when the storage contents of the write-once memory 4 are as shown in Table 3 (i) in the fourth generation, the decoded symbol string "21
1o'' are stored in the entire decoded symbol buffer 12.

As described above, in the above-mentioned operation example, the write-once memory encoding device of this embodiment stores the data in the storage area of the 8-block write-once memory over four generations as shown in (b) in Table 3.
A total of 26 information symbols of d), (f), and (h) can be written. 1 information symbol is Log23
Since it has an information amount of bits, the total amount of written information per memory unit is 10g23.25/(3.8, that is, 1.651 bits).

The above sequence of operations is just an example, so in order to demonstrate the superiority of the encoder/decoder for write-once memory of this embodiment over the conventional method, the average total amount of written information will be determined using probability calculation. must be evaluated. Below, the average total amount of written information per storage unit of the encoding/decoding device of this embodiment will be derived. Here, it is assumed that the ternary input information symbols occur with equal probability and statistically independently.

The Markov chain diagram in FIG. 2 shows how the write state of a certain block changes from the initial state 5o=000 after multiple generations have been written using this code. Hereinafter, the state in which the code word W is written will be referred to as state Sw. If we consider that the probability of occurrence of each ternary input symbol is K, the following state transition probability matrix P can be obtained.

Here, the initial state is Sa. In states SQ to 82, writing is possible with probability 1. Further, in states S3 to S85, writing is possible with a high probability. Probability Q1 that a certain block is writable in the first generation
is clearly 1, and the above probability Q2 in the second generation is as follows. Generally, the above probability Q in the th generation is t〉2
In , it becomes .

Given the above state transition probability matrix P, the probability that the first generation will be in the state S1 and the second generation will be in the state Sj is as follows.

is obtained. Here, a=t (t-1), b=t (t-1)/2,
c=3t-2t 2-1゜a==3t-2t-1,6
= 3t-1, f = 3t0 According to these formulas, the average total amount of written information r for one memory unit is r = 1og23/3- Σ Qt t = = -1 = 1.534 (b-t) becomes. Therefore, it is shown that the write-once memory encoding/decoding device of this embodiment has a larger total amount of written information than the conventional system.

FIG. 6 shows a Markov chain diagram showing the relationship between information symbols and written code words in a write-once memory encoding/decoding device according to another embodiment of the present invention.

This code has a code length N of 4 and a number M of information symbols of 4. The average total amount of written information per storage unit of this code can be calculated by performing the same probability calculation as described above.
It is proven that the number of bits is 1.926 bits. In this implementation sequence, as well as the above-mentioned implementation sequence, W
The average total amount of written information per storage unit is also large compared to the OM code.

As described in detail, according to the present invention, it is possible to realize an encoding/decoding device for a write-once memory in which the total amount of information written per one storage unit is larger than that of the conventional one, and its practical effects are significant. Hey.

[Brief explanation of drawings]

FIG. 1 is a block diagram of a write-once memory encoding device for "-implementation sequences" according to the present invention, FIG. 2 is a Markov chain diagram showing the relationship between information symbols and code words in the same embodiment, and FIG. 3 is a diagram according to the present invention. A configuration diagram of a write-once memory decoding device according to an embodiment of the present invention;
FIG. 4 is a content diagram of a conversion table between code words and decoded symbols, and FIG. 6 is a flowchart showing a series of operations of the write-once memory encoding/decoding apparatus according to an embodiment of the present invention.
FIG. 6 shows an encoding method for write-once memory according to another embodiment of the present invention.
It is a Smarkov chain diagram showing the relationship between information symbols and code words of a decoding device. 1...Information symbol input unit, 4...Writable memory, 5...Encoding block counter, 6...
...Block reading unit, 7...Code word selection unit, 8...Valid code word writing unit, 9...
- Invalid code word writing unit, 10...Decoded block counter, 11...-Decoded symbol conversion unit. Name of agent: Patent attorney Toshio Nakao and 1 other person -ψ
Figure 2 Figure 3 Figure 4 (Kyu) (b) Figure 5 A (Encoding table L) Figure 5 B (Decoding table)

Claims (2)

[Claims] (1) an information symbol input section for inputting information symbols;
A write-once memory in which N bits are one block and write and read in blocks are possible; block designation means for specifying a block in the write-once memory; and designation by the block designation means in the write-once memory. a block reading unit that reads the memory contents of the block to be processed, a valid codeword writing unit, an invalid codeword writing unit, and the codeword read by the block reading unit and the input information symbol from the information symbol input unit. Select a write codeword that is a valid codeword or an invalid codeword, and if the codeword to be written is a valid codeword, start the valid codeword writing section, and if the codeword to be written is an invalid codeword, start the invalid codeword. and a code word selection section that starts itself recursively after starting the word writing section, and the effective code word writing section writes the effective code word selected by the code word selection section to the block in the write-once memory. The invalid codeword writing unit writes the invalid codeword to the block specified by the block specifying means in the write-once memory and writes the contents specified by the block specifying means to the block next to the block. The effective code word selected for the information symbol x by the code word selection unit is (I) length N
is a binary sequence of bits, and (II) is not a continuous sequence of ones of length N, and (III) is transitionable from the codeword written in the block, and (I
V) It is set under the condition that the information symbol can be uniquely decoded from the effective code word, while the invalid code word is set to be a continuous sequence of 1's of length N. A write-once memory encoding device. (2) a write-once memory that has N bits as one block and can be read in blocks; a block designation means for specifying a block number of the write-once memory that is currently being read; a block reading section that reads the stored contents of the block specified by the block specifying means; a block reading section that stores the relationship between a code word that satisfies a predetermined condition and a decoded symbol that corresponds to the code word that is read by the block reading section; is a valid codeword or an invalid codeword, and if it is a valid codeword, the corresponding decoded symbol is output, and if it is an invalid codeword, the specified content of the block designation means is changed to the next block of the block. It has a decoding symbol converter that recursively activates itself after forming a block, and as the predetermined conditions, the effective code word is (I) a binary sequence of length N bits, and (
II) is not a continuous sequence of 1s of length N, and (III) is capable of transition from the codeword written in the block, and (IV) is a unique information symbol from the effective codeword. 1. A decoding device for a write-once memory, characterized in that the condition is set such that the code word can be decoded, and the invalid code word is set to be a continuous sequence of 1's of length N.