Abstract:
In this paper we propose a passive cellular data rate estimation method, which works with parameters accessible from most Commercial Off-The-Shelf (COTS) modems today. We...Show MoreMetadata
Abstract:
In this paper we propose a passive cellular data rate estimation method, which works with parameters accessible from most Commercial Off-The-Shelf (COTS) modems today. We show that by adjusting the optimization objective we can tweak the estimator to achieve higher accuracy at lower data rates. This is particularly useful in cases where a high data rate cellular link is backed up by a low data rate failover link (e.g. a satellite connection). The way the problem is approached makes this solution easily implementable in a product, for example as an input to a Quality-of-Service (QoS) router. The estimation is performed using both, supervised machine learning algorithms and a simple linear regression. The latter can be considered as a lightweight implementation on systems with low processing power or tight energy restrictions.
Published in: 2018 14th International Conference on Wireless and Mobile Computing, Networking and Communications (WiMob)
Date of Conference: 15-17 October 2018
Date Added to IEEE Xplore: 27 December 2018
ISBN Information:
Print on Demand(PoD) ISSN: 2160-4886