In cognitive radio context, the coding parameters are unknown at the receiver. The design of an intelligent receiver is essentially to identify these parameters from the received data blindly. In this paper, we are interested in the blind identification of binary linear block codes from received noisy data. In order to recognize the code length, the concept of the normalized column weight vector is defined and cosine similarity is used to measure the difference between linear block codes and random codes. Then, the correct code length could be obtained by finding the local minimum of cosine similarity. The proposed code length recognition method needs no prior knowledge about the codes, which results in completely blind identification. To reconstruct the parity check matrix, the concept of association rules mining is introduced to the problem of blind identification of channel codes for the first time. Furthermore, five criteria are proposed to reduce the redundant rules mined by the association rules mining algorithm and to recognize the parity check vectors effectively. Simulations show that the proposed two methods have excellent performance even in a high error rate transport environment. The performance comparisons with existing methods validate the advantages of our two proposed methods.

在认知无线电环境中，接收器处的编码参数未知。智能接收机的设计本质上是从接收到的数据中盲目地识别这些参数。在本文中，我们对从接收到的噪声数据中盲目识别二进制线性块代码感兴趣。为了识别代码长度，定义了标准化列权重向量的概念，并使用余弦相似度来测量线性块代码和随机代码之间的差异。然后，可以通过找到余弦相似度的局部最小值来获得正确的代码长度。所提出的码长识别方法不需要关于码的先验知识，这导致完全盲识别。要重建奇偶校验矩阵，关联规则挖掘的概念首次被引入到信道编码的盲识别问题中。此外，提出了五个准则来减少由关联规则挖掘算法挖掘的冗余规则，并有效地识别奇偶校验向量。仿真表明，提出的两种方法即使在高误码率传输环境下也具有优异的性能。与现有方法的性能比较证明了我们提出的两种方法的优点。仿真表明，提出的两种方法即使在高误码率传输环境下也具有优异的性能。与现有方法的性能比较证明了我们提出的两种方法的优点。仿真表明，提出的两种方法即使在高误码率传输环境下也具有优异的性能。与现有方法的性能比较证明了我们提出的两种方法的优点。