Home // AICT 2017, The Thirteenth Advanced International Conference on Telecommunications // View article

Machine Learning Regression-Based Approach for Dynamic Wireless Network Interface Selection

Authors:
Lucas M F Harada
Daniel C Cunha

Keywords: Network selection; energy consumption; wireless interface; machine learning; regression

Abstract:
Battery consumption is a general problem in any portable wireless device and it depends directly on the transmission technology (cellular, Wi-Fi or short-range wireless networks) that is used to send and receive data. When various networks are available, mobile devices should be able to choose which network interface to use based on a variety of factors, such as required bandwidth or energy efficiency. This work proposes a dynamic wireless network interface-selection mechanism focused on minimizing the energy consumption of the mobile device, allowing an increase in battery life. In doing so, Machine Learning (ML) regression-based algorithms are used to predict the energy cost per transferred byte for each type of available network interface using field data. A comparison of the energy consumptions for both the proposed mechanism and the Android native method is performed. Numerical results show that our proposal helps save energy.

Pages: 7 to 12

Copyright: Copyright (c) IARIA, 2017

Publication date: June 25, 2017

Published in: conference

ISSN: 2308-4030

ISBN: 978-1-61208-562-3

Location: Venice, Italy

Dates: from June 25, 2017 to June 29, 2017