Research on Basic Network Protocol of Basic BP Neural Network

Research on Basic Network Protocol of Basic BP Neural Network

연구 분야: Networking

학회: 2023 4th International Conference for Emerging Technology (INCET)

This study aims to explore the basic network protocol research based on BP neural networks. BP neural network is a feedforward network composed of multiple layers of neurons, with learning and generalization abilities. The basic network protocol research based on BP neural network can be used for the design of transmission and routing protocols in various computer networks. Firstly, this study introduces the BP neural network and its training algorithm, including the basic structure and working principle of neurons, Backpropagation and weight update formula. Then, the feasibility and advantages of applying BP neural network to basic network protocol design were explored. BP neural network can learn and adapt to the network environment independently, and optimize the network performance by adjusting the network weight and Activation function and other parameters. Subsequently, this study proposes a transmission protocol design method and a routing protocol design method based on BP neural networks. The design method of transmission protocol focuses on how to select appropriate network parameters and training samples for model training, in order to construct a high-performance transmission protocol. The routing protocol design method focuses on how to use the network topology information and Routing table information in the neural network to build a reliable and efficient network routing protocol. In addition to these three layers, there is a Activation function between I and o. The Activation function can be used to calculate a value indicating the degree to which each node in o is activated by I input. It aims to learn and operate in environments with limited resources, such as on-board computers or mobile devices. The main advantage of this network is that it can reduce the number of parameters to be trained by using only two input variables and one output variable. This makes it easy for users to implement and use on their own systems without any special knowledge... Show More