On Blind Recognition of Channel Codes Within a Candidate Set

摘要

Existing schemes for blind recognition of channel codes make use of the average log-likelihood ratio (LLR) of each code’s parity checks. There are difficulties in setting necessary thresholds and in theoretical analysis of the method. This letter proposes to use the average likelihood difference (LD) of the parity checks. The computational complexity is reduced while the recognition performance is kept comparable to that of the LLR method. Moreover, an optimum decoding-based blind recognition is also described. Simulations show that, for rate 1/2 convolutional codes, this new method notably improves the performance.