A Novel Blind Detection Algorithm Based on Adjustable Parameters Activation Function Hopfield Neural Network

Aiming at the shortcomings of the traditional blind detection algorithm, we focus on the flexibility of the activation function. This paper presents an adjustable parameters activation function, which not only showed greater flexibility and nonlinear properties by regulating the steepness, position and mapping range, but improved the performance of the Hopfield Neural Network(HNN) blind detection algorithm. The simulation results demonstrate that the novel algorithm can reduce the error rate significantly and speed up the convergence of HNN on the condition of low signal-noise ratio(SNR) and complex large data environment. Thereby the novel activation function improves the performance of blind detection algorithm.