Premium
Deep‐DRX: A framework for deep learning–based discontinuous reception in 5G wireless networks
Author(s) -
Memon Mudasar Latif,
Maheshwari Mukesh Kumar,
Shin Dong Ryeol,
Roy Abhishek,
Saxevrati
Publication year - 2019
Publication title -
transactions on emerging telecommunications technologies
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.366
H-Index - 47
ISSN - 2161-3915
DOI - 10.1002/ett.3579
Subject(s) - computer science , deep learning , trace (psycholinguistics) , latency (audio) , real time computing , low latency (capital markets) , wireless network , recurrent neural network , artificial intelligence , network packet , artificial neural network , wireless , computer network , telecommunications , philosophy , linguistics
The tremendous advancement in various types of mobile devices with distinct services demands emerging 5G networks to deal with different kinds of traffic. The high data rates along with beam searching operation in 5G increase user equipment (UE) energy expenses. Discontinuous reception (DRX) is used in long‐term evolution network to save the power of UE. The DRX conserves the energy at the cost of increased latency. On the other hand, long short‐term memory (LSTM), a deep neural network, has shown incredible results in learning time‐varying sequences from data sets. In this article, we propose a novel idea to train LSTM network for prediction of next packet arrival time based on real wireless traffic trace (dataset). Then, we use the trained LSTM model to predict dynamic sleep time in DRX for 5G networks. Our proposed algorithm, deep learning–based DRX (Deep‐DRX) is able to make dynamic sleep cycle in 5G communications. Deep‐DRX achieves 10 % and 30 % of power saving with a mean delay of 1.006 ms and 1.05 ms for Trace 1 and Trace 2, respectively.