Abstract
Atmospheric ducts are horizontal layers that occur under certain weather conditions in the lower atmosphere. Radio signals guided in atmospheric ducts tend to experience less attenuation and spread much farther, i.e, hundreds of kilometers. In a large-scale deployed TD-LTE (Time Division Long Term Evolution) network, atmospheric ducts cause faraway downlink wireless signals to propagate beyond the designed protection distance and interfere with local uplink signals, thus resulting in a large outage probability. In this paper, we analyze the characteristics of ADI atmospheric duct interference (Atmospheric Duct Interference) by the use of real network-side big data from the current operated TD-LTE network owned by China Mobile. The analysis results yield the time varying and directional characteristics of ADI. In addition, we proposed an SVM (Support Vector Machine)-classifier based spacial prediction method of ADI by machine learning over combination of real network-side big data and real meteorological data. Furthermore, an implementation of ADMM (Alternating Direction Methods of Multipliers) framework is proposed to implement a distributed SVM prediction scheme, which reduces data exchange among different regions/cities, maintains similar prediction accuracy and is thus of a more practical use to operators.
Original language | English |
---|---|
Pages (from-to) | 66-80 |
Number of pages | 15 |
Journal | Journal of Communications and Information Networks |
Volume | 2 |
Issue number | 1 |
DOIs | |
Publication status | Published - 1 Mar 2017 |
Publication type | A1 Journal article-refereed |
Publication forum classification
- No publication forum level