Reconstruction of punctured convolutional codes
read more
Citations
Parameter Estimation of a Convolutional Encoder from Noisy Observations
Classification of Error Correcting Codes and Estimation of Interleaver Parameters in a Noisy Transmission Environment
Reconstruction of a Linear Scrambler
Algebraic method for blind recovery of punctured convolutional encoders from an erroneous bitstream
Reconstruction of convolutional codes from noisy observation
References
Algebraic Coding Theory
Shift-register synthesis and BCH decoding
A new algorithm for finding minimum-weight words in a linear code: application to McEliece's cryptosystem and to narrow-sense BCH codes of length 511
Parameter Estimation of a Convolutional Encoder from Noisy Observations
A New Algorithm for Finding Minimum-Weight Words in a Linear Code: Application to Primitive Narrow-Sense BCH Codes of Length~511
Related Papers (5)
Parameter Estimation of a Convolutional Encoder from Noisy Observations
Novel Blind Encoder Parameter Estimation for Turbo Codes
Frequently Asked Questions (11)
Q2. How long does it take to test a punctured code?
Their algorithm runs in a few minutes for a (2, 1) parent code punctured into a (9, 8) code, which corresponds to ( 16 9 ) = 11 440 patterns to test, but takes only a few milliseconds for the same (2, 1) parent code punctured into a (4, 3) code.
Q3. How many dualwords did Valembois generate in a second?
In practice, Valembois’ algorithm outputs dozens of dualwords in a second: during their tests, even with the highest noise levels and the largest codes, the authors were always able to generate the whole orthogonal space G⊥.
Q4. What is the constraint length of a convolutional code?
An important parameter of convolutional codes is their constraint length which corresponds to the total size of their internal memory.
Q5. What is the first algorithm to reconstruct a convolutional code?
The first one applies to generic convolutional codes and enables reconstruction in the presence of high noise levels with a relatively short bitstream (a few thousand bits are enough).
Q6. What is the way to reconstruct a convolutional code?
However knowing any basis of G is enough to correct errors, so the authors first recover the space G and then try to deduce the correct matrix G.In 1995, B. Rice was the first to deal with the problem of reconstructing a convolutional code and proposed an algorithm [12] to solve this problem for (n, 1) convolutional codes when there are no errors.
Q7. What is the value of the constraint length of a code?
This parameter is important as the complexity of the Viterbi decoding algorithmis proportional to 2m+k which means that higher constraint length codes are much more costly to decode.
Q8. What is the way to deduce the parent code of a given matrix?
2) Recovering the Parent Code: for a given transition matrix P on F2[D], from the n last columns of the expanded matrix P × G′ the authors can deduce the n polynomials of the equivalent (n, 1) convolutional code: some of these columns might have been punctured but can easily be deduced from other columns of P × G′ by multiplication by a power of Z−1.
Q9. What is the possible encoding of the vector space G?
In practice, one can recover the vector space G generated by G over F2(D), but then, any polynomial basis of G is a possible encoder.
Q10. How do the authors make the base G′ basic?
This is done in two steps: • first make the basis G′ basic by computing its Smith form(cf. [10] p. 39), • then reduce this basic basis to decrease the degrees of theelements while keeping the basis basic (cf. [10] p. 58).
Q11. What is the technique used to recover a binary dualword?
1) The Reconstruction Algorithm: the technique the authors use is quite simple: the authors know that for block lengths which are sufficiently large multiples of n, Valembois’ algorithm should recover some dualwords.