A multidimensional time series data on wireless network to improve the detection performance of unsupervised learning algorithms
Chayan Paul, S. Ranjith, R.S. Vijayashanthi, R. Benazir Begam, R. Sabitha · 2022 International Conference on Innovative Computing, Intelligent Communication and Smart Electrical Systems (ICSES) · 2022
In recent years, only with developments of cellular networking technologies as well as the implementation of 4G LTE networks, the skills and options provided to end-users had expanded to a level that seems to be limitless. Smart phone users who travel by public transportation rely on high-speed internet infrastructure and expect to have a stable, greater connectivity for the length of their journey. Because of the high requirements placed on wireless networks and the regular changes in the fundamental radio channel, customers frequently encounter rapid and surprising alterations in the connectivity reliability of their connections. The old traditional process of evaluating and inspecting infrastructure cellular technology should be no longer sufficient to fulfill the demands of the manufacturing industry. Consequently, an accurate mobile network cellular irregularity diagnostic method is required to monitor for problems that lead and to improving the effectiveness of device construction and operation. It is estimated that there are thousands of mistakes and exclusions in the current single-dimensional abnormality identification technique, and also that the present multifunctional abnormality diagnostic technique has a poor detection accuracy on multifunctional time series analysis. Multidimensional statistics depending on network service device abnormality classification optimization approach described in this work is predicated on machine learning that use limited of information to support in the instruction of tremendous unmonitored techniques, thereby enhancing the detection accuracy of unsupervised classification techniques. This paper conducts experimental studies to validate the functionality of the optimization techniques, and it shows a significant progress over the four generally used unmonitored techniques, that can greatly enhance the abnormality detection performance of the existing techniques, according to the authors.