Detecting and reconstructing an unknown convolutional code by counting collisions

We suggest in this paper a new method for detecting whether a given binary sequence is a noisy convolutional codeword obtained from an unknown convolutional code. It basically consists in forming blocks of the sequence which are big enough to contain the support of a codeword in the dual of the convolutional code and to count the number of blocks which are equal. This detection process is quite efficient and presents the advantage over all previously known methods to achieve this goal even in the case of an unknown modulation. Moreover, this method can also be used to reconstruct the unknown convolutional code when the modulation is known.