A Decoding Method For Modulo Operations-Based Fountain Codes Using the Accelerated Hopfield Neural Network

作者：Deng Zaihui^*; Tong Xiaojun; Gan Liangcai

来源：International Conference on Computer Engineering, Information Science & Application Technology (ICCIA), 2016-09-24 to 2016-09-25.

• Fountain codes
• Modulo operation
• Decoding
• Chinese remainder theorem
• Neural network

This paper describes a decoding method using the accelerated Hopfield neural network, in order to address the high complexity of decoding for modulo operations-based fountain codes. The method constructs a neural network model based on a non-linear differential equation, and runs the model after setting an initial value. During the process, the model's output value first rapidly decreases under the effect of the accelerator resistor, slows down near an equilibrium point, and finally regresses to a unique equilibrium point with an arbitrarily small error. The result is half-adjusted to obtain the source data sequence. Simulated tests indicate the method to be valid, and can potentially bring the modulo fountain codes closer to practical application.

• 出版日期2016
• 单位武汉大学; 武汉纺织大学