The Gauss elimination method to recognize Turbo code frame structure has the shortcomings of poor fault tolerance and high complex computation

which directly leads to the inapplicability of subsequent algorithms to recognize the component coder and interleaver. For this reason

a novel method based on correlation between columns of the matrix is proposed. Firstly

analysis matrix is constructed under the undetermined frame length

with the neighboring columns being differentiated. Then statistics about column weight is established

and frame length is confirmed by the statistical values

which can pass the decision threshold when the column is frame length and its multiple. Finally

frame staring point is synchronized according to the relative position of statistical values which pass the threshold. The simulation results show that the correct recognition rate of parameters of Turbo frame structure can reach 100%

when the bit error rate is as high as 0.1. Compared with the existing methods

the proposed algorithm has lower computational complexity and better recognition performance

and is very suitable for non-cooperative communication fields such as cognitive radio.