|
I need some error correction for an integer of 32 bits.
I've found several solutions, like 'reed solomon', 'FEC', 'parity bit'. But all of those are for a lot of bits.
I just need error correction for 32 bits and the error correction should be maximum 24 bits.
Does any of you have any idea of how to clear this problem and which of the error corrections I should use (or even create one myself)?
|
|
|
|
|
Hello,
http://en.wikipedia.org/wiki/Hamming_code - that one is pretty easy, scalable (you decide of length of data chunk, longer chunk - lesser bits will be taken for correction purposes).
Why parity bit is for lot of bits? You can use 1 parity bit for 2 bits of data, for 20 bits, for 200... (of course it will be less effective).
Btw - Hamming codes can correct error bits, not only check if data is not corrupted.
|
|
|
|
|
I see, hamming_code would be great indeed (thank you). But it's only possible for single error correction. In the best case I would like to correct every bit, but that's impossible, but I would like to have the highest efficiency.
I'm now busy with trying parity, horizontal + vertical + diagonal. And then check if the parity's are ok.
But is there another way than the hamming? Because I think this one is not really efficient.
|
|
|
|
|
Hello,
Actually, I think that Hamming is very efficient for what it offers (especially for longer streams, you add 1 bit for every *2 bits of data), if you need more bits corrected - use shorter version etc.
If I have understand you well and by
Deresen wrote: trying parity, horizontal + vertical + diagonal
you mean putting input into matrix and adding parity bits to every row/column, then you won't be able to fix any bits if you will have 2 errors (this is sometimes true, sometimes not, but still, you can't guarantee correcting 2 bits) and the cost of parity bits is much higher then using hamming.
You can also google for Convolutional code, but I don't know much about them, so I can't guarantee it is what you have been lookng for.
|
|
|
|
|
whats wrong with Reed-Solomon?
If I understand your requirements, you need an RS(4,1) code over GF(2^8) or RS(8,2) over GF(2^4), and because the codeword size is small you can use Euclid's algorithm to efficiently compute the key equation rather than Berlekamp-Massey (modified).
check this link out:
http://en.wikipedia.org/wiki/Reed%E2%80%93Solomon_error_correction[^]
|
|
|
|
|
To be honest, I did not really understand the reed solomon error correction.
And I've read this '2 byte errors per 32-byte block', this also means 2 bits per 32 bits. And that's to less for me.
The big problem is that I also have to check the error correction, if that is right. So I have to correct that stream also. Is this a possibility with reed solomon? And could you please give a small example of how the reed solomon works, for instance with a byte?
|
|
|
|
|
Reed-Solomon does provide the ability to detect errors and also correct them, RS is actually quite trivial to understand and implement, RS operates off-of a basis type which is called a symbol. A symbol can be any size from 2-bits and up.
My suggestions were based on two symbol sizes, either an 8-bit or 4-bit symbol size. The 4-bit symbol size would be recommended in your case, as it means 8 symbols per your data block (32-bits). The proposed code of RS(8,2) over GF(2^4) means any two of the 8 symbols can have any number of bit errors (1-4 bits of error/burst error), from this upto both symbols of error can be accurately DETECTED and CORRECTED.
Further more if you happen to know which bits are in error you can increase the correction capabilities 2 fold via erasure correction methods.
If you're familiar with C++ the following library has RS code examples for various bit sizes: http://www.schifra.com/downloads.html[^]
|
|
|
|
|
Thank you very much for your time.
This will give me enough homework for a while.
Let's dive into C++ again.
|
|
|
|
|
What is the expected nature of errors? Is there a significant likelihood of having multiple independent bit errors, or would bit errors more likely be concentrated?
Correcting multiple independent bit errors is very hard; it is clearly impossible to do two-bit correction on 32 bits of data with less than 11 bits of ECC data (using 42 bits to code a 32-bit word, allowing for zero, one, or two errors, there would be 1,765 ways each code word could be represented; a 10-bit ECC would only allow for 1,024).
On the other hand, if all bit errors will be localized in some particular fashion, error detection and correction becomes much simpler. For example, if all errors will be within a single byte, one can store a parity byte along with enough parity information to detect errors in each byte. If a problem is indicated with any byte, the parity byte will allow it to be corrected.
|
|
|
|
|
Hi everyone!
I'm working on a project and i need to do what follows:
i have an array of strings,example:
house3
house2
green23
green.5
H6
H01
G19
G78
..and so on.
I have to extract only the recurrent strings: house,green,H,G.
Any idea about it? (the language is java but doesn't matter, i just need the idea!)
Thanks in advance,
Enrico.
Program your life ^^
|
|
|
|
|
The following should give you some ideas:
http://en.wikipedia.org/wiki/Patricia_tree[^]
...cmk
The idea that I can be presented with a problem, set out to logically solve it with the tools at hand, and wind up with a program that could not be legally used because someone else followed the same logical steps some years ago and filed for a patent on it is horrifying.
- John Carmack
|
|
|
|
|
What are you gonna do exactly ?
Do you want to identify the strings that contain house,green,H,G ?
|
|
|
|
|
not exactly. I have some strings and i have no idea which they are.
I have to identify the recurrent substrings such as house, green...but i don't have them as input..
Program your life ^^
|
|
|
|
|
Some of the problem parameters aren't clear: Do the recurrent strings always start at the beginning of the line? Do you just ignore non-letters (numbers and punctuation)?
If the answer to both of these questions is yes, you can do this:
1. For each line
2. Extract the string of just the letters from the beginning of the line.
3. See if this string is in a hash table. If so, it's a recurrent string.
4. Else insert this string into the hash table.
If the strings don't always start at the beginning of the line, you have to do multiple insertions for each string: the whole string, the substring starting at the 2nd letter, 3rd letter, etc...
|
|
|
|
|
You need to look at using a data structure like a Trie[^], or as was already pointed out Patrcia Trie[^]. The Wikipedia articles should give you enough information to get started on an actual implementation.
Scott Dorman Microsoft® MVP - Visual C# | MCPD
President - Tampa Bay IASA
[ Blog][ Articles][ Forum Guidelines] Hey, hey, hey. Don't be mean. We don't have to be mean because, remember, no matter where you go, there you are. - Buckaroo Banzai
|
|
|
|
|
When you aren't a mathematician, it's a pain. When you know what its called, there are pages and pages of info. But people rarely put "The DooDingle algorithm calculates the surface area of a dodecahedron cut by a given XYZ plane" on their webpages.
So half the battle is trying to find out what you should be Googling for.
(Incidentally, I'm trying to find a method of locating the bounding triangle in a Delaunay triangulation for any given coord XY, where the vertices of the triangles are a random set of XYZ coords. Then I can work out the Z at any XY, given my grid).
|
|
|
|
|
Try Googling for Delaunay. I have seen it there as well as under "Convex Hull".
Dave.
|
|
|
|
|
Hi! I am working on a graphing application which plots math functions (for example sin(x)). At this stage the program works but in some cases the graph isn't correct. The program calculates the points of the graph and finds the point (x,f(x)). But some calculations return very small numbers an the graph is not correct. In order to fix this I decided also to calculate the limits of the function in the non defined points (for example lim 1/x, x->0). If someone has stumbled upon some articles or knows something about the problem, let me know. (Google only showed me the math definition of the limit, but I couldn't find an algorithm). Thanks!
Still learning...
|
|
|
|
|
It might be a good idea to identify which inputs return wrong answers, then try to find a pattern in the errors. Fixing the underlying bug will produce a more reliable program than adding a patch involving limits.
|
|
|
|
|
Hi......i need help in ASP.NET for vs 2005 to develop an e commerce website for my final year B.Tech. project....plz guide me...........specially how to develop credit cart verification transaction and shopping cart module??????????
|
|
|
|
|
Well, you'd normally start out by defining your requirements. Possibly you'll want to draw some high level use cases out of these, and then you'll want to break these down into a more detailed design. From there, you can pull together your basic code.
|
|
|
|
|
Pete O'Hanlon wrote: Well, you'd normally start out by defining your requirements. Possibly you'll want to draw some high level use cases out of these, and then you'll want to break these down into a more detailed design. From there, you can pull together your basic code.
Very good advise...
If you want to try doing some Data Flow Diagrams(DFD's). Could help with how you want to the project to flow.
|
|
|
|
|
My objective is to made an online shopping website based on ASP.NET 2.0 ,for which i m using visual studion IDE and prograaming in C#............i need a good ebook on ASP.NET which includes session management as well as database handling......I have made modules.......I m facing difficulty in shopping cart implementation and credit card verification modules..........
|
|
|
|
|
Dear shriyangika,
First of all you should be knowing how the data is flowing.take paper and pencil note down everything what ever coming in your mind about the project ..
as you told
shriyangika wrote: I have made modules.......I m facing difficulty in shopping cart implementation and credit card verification modules..........
whats the module.. how it looks like...and what exact problem you are facing... if you can tell all these it will helpful for the one who is trying to help you here..
vikas da
|
|
|
|
|
You're tackling the problem from the wrong side here. You're talking the technology end when, what you should be doing, is actually working out what the problem really is.
|
|
|
|