Open AccessHotspots Prediction Based on LSTM Neural Network for Cellular Networks
Author(s) -
Lixia Zhou,
Xia Chen,
Runsha Dong,
Shan Yang
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1624/5/052016
Subject(s) - computer science , artificial neural network , data mining , cellular network , hotspot (geology) , quality of service , time series , long short term memory , artificial intelligence , machine learning , recurrent neural network , real time computing , computer network , geophysics , geology
The tremendous growth in data traffic usage is a prominent challenge to cellular network operators. The aggregation of a small number of users’ services may lead to considerably high loads on base stations, thus generating traffic hotspots in the networks. To provide better quality of service in dynamic network scenarios efficiently, it is necessary to accurately distinguish and predict potential hotspots, so as to adjust resource configuration and allocation in advance and keep the network running smoothly. In this paper, a network traffic hotspots prediction method is proposed based on the Long Short-Term Memory (LSTM) neural network framework. To verify its effectiveness, real-world data are utilized in our experiments. Numerical results demonstrate that a time series model based on the LSTM framework can achieve 92.2% success rate for future hotspots prediction.