Is it possible to achieve 25%+ error correction capability when output cannot be bigger than 175% of input? I was looking for Reed-Solomon code. With 255 output symbols, I can have 145 input symbols. (145 * 1.75 < 255)
That means (110/2)/255 * 100% = 21.5%
Is there any method or other error correction code to achieve 25%+? Thanks
No.
The output being 175% of the input means the input is 57% of the whole data. A RS code, as you know, needs 50% space for 25% error correcting capability, that won't fit in.
And, as long as you want something comparable to what a RS code can do, nothing is better in terms of used space. RS codes are optimal according to the singleton bound.
The only chance to get better than that is to use a unreliable scheme that sometimes corrects more than RS, but other times fails where RS would have worked.
