Abstract—Regarding to the problems of low rate of convergence and fault saturation for neural network classifier based on the algorithm of error back propagation during the signal recognition, bee colony algorithm is applied in this paper so as to extract combined feature module of signal and suggest three different algorithms including algorithm with rapidly support, super self-adaption error back propagation and conjugate gradient. These three algorithms are respectively applied in multilayer perception neural network classifier, and help achieve automatic recognition for communication signals and higher recognition rate compared with error back propagation. The simulation result shows that the algorithms put forward in this paper can overcome the drawbacks of error back propagation algorithm. Meanwhile, under the condition that nerve cell has only 20, SNR is 4dB in the hidden layer, the recognition rate of three algorithms are all higher than 95%, the system is easy to implement and has wide range of application prospect in the signal recognition.

Index Terms—Combined feature module, bee colony algorithm, multi-layer perceptron neural network, modulation recognition

Cite: Faquan Yang, Jie Zheng, Haishu Tan, and Yun Fan, “A Specific Combination Scheme for Communication Modulation Recognition Based on the Bees Algorithm and Neural Network," Journal of Communications, vol. 10, no. 10, pp. 797-803, 2015. Doi: 10.12720/jcm.10.10.797-803