A Method for Constructing Dataset of Campus Network Broadband Access Server

Broadband Access Server (BAS) is a critical device located at the edge of the Internet Service Provider (ISP) network, responsible for centralized management, control, and forwarding of user data traffic. In campus networks, broadband access technologies have played a significant role in providing faster and more stable network connections. Currently, the field lacks relevant datasets, which has led to a shortage of accurate, effective, and high-quality data support for BAS research. This paper proposes a method for constructing a dataset for campus network BAS and uses this method to collect one month of data from a real campus network environment. To validate the usability of the dataset, the device load prediction was taken as an example, and various machine learning algorithms were tested. The experimental results show that the dataset collected by this method is accurate and complete, truly reflecting the actual characteristics of the devices, and capable of distinguishing the generalization capabilities of different models. It can be used for relevant experimental research and has achieved the expected objectives.