Root Cause Localization of Wireless Communication Network Faults Based on 1DCNN-Att Network

Locating the root cause of wireless network failures is critical to wireless network operations and maintenance. The introduction of new technology architecture in 5G and the rapid development of 6G technology have led to the emergence of diverse new scene services. Consequently, these services pose challenges in locating the root cause of wireless communication network faults. However, existing root cause localization schemes suffer from issues such as low accuracy and complex models. In this paper, we propose a scheme based on 1DCNN and attention mechanism to address these problems. Our scheme involves selecting, filling, and compressing the original data for efficient time series feature engineering. Then, we utilize the 1DCNN network to capturing the nonlinear relationships of different key performance indicators to overcome the low accuracy of the existing schemes. Additionally, this scheme has the advantages of a simple structure and short processing time. Finally, we conduct experiments and analysis using a dataset from real-world 5G business scenarios, and the results demonstrate the superior performance of our proposed scheme.