Subscriber dynamic characteristics-based wireless network accessing bandwidth prediction.

Autor: Han, Di, Liang, Haili, Shen, Xianmin, Yang, Lei
Zdroj: International Journal of Machine Learning & Cybernetics; Dec2014, Vol. 5 Issue 6, p875-885, 11p
Abstrakt: The future of mobile computing increases the urgent need for predictions of network bandwidth to guide the users' operating habits. For instance it's assumed that servers, which are used for listening online music and watching video (streaming service) in our life, and the users of the network are immobile and will not change within a certain time. But the actual network is filled with regular changes. If we can predict the amount of network bandwidth, which users use in the next period of time, we can guide the users to transmittal and cache data in good transmission of the network status. At the same time, we do not transfer data in the bad network status or even when we can not receive the signal. It can improve the efficiency of network transmission in order to save the energy consumption of the data transmission. The real-time measuring network bandwidth value need upload files from client to server, and then to download files from the server to the client. We must get the uploading or downloading time. This approach generally takes a long time; therefore the test will not be able to proceed frequently in the limited time. In addition, the measurement of periodic real-time network bandwidth leads to the larger energy consumption in the mobile terminal. In this paper, we design an accurate and low cost forecasting method. First, we get a mapping relationship between human and environmental factors through large-scale data collection and analysis. Secondly, we use the method of trajectory match to predict the mobile network status in order to get the results which the users wish to get. At the end of the paper, we make our designed model simulation predictions and compare to the real measurement data to validate our predictive analysis. [ABSTRACT FROM AUTHOR]
Databáze: Complementary Index
Externí odkaz: