Abstract
Traffic-aware cellular networks dynamically turn on/off some base stations (BSs) according to the predicted traffic variation pattern and thus are able to improve the energy efficiency while providing plenty of network capacity. In this paper, instead of depending on the predicted traffic knowledge, we formulate the traffic variations as a Markov chain and design an intelligent BS management scheme with the aid of reinforcement learning framework. Specifically, we propose a Transfer Actor-CriTic (TACT) algorithm, which leverages the concept of transfer learning and exploits the transferred learning expertise from historical periods or neighboring regions to obtain better energy saving performance.
Original language | English |
---|---|
Title of host publication | 2014 XXXIth URSI General Assembly and Scientific Symposium (URSI GASS) |
Publisher | IEEE Institute of Electrical and Electronic Engineers |
ISBN (Electronic) | 978-1-4673-5225-3 |
DOIs | |
Publication status | Published - 17 Oct 2014 |
MoE publication type | Not Eligible |
Event | 31st General Assembly and Scientific Symposium of the International Union of Radio Science, URSI GASS 2014 - Beijing, China Duration: 16 Aug 2014 → 23 Aug 2014 |
Conference
Conference | 31st General Assembly and Scientific Symposium of the International Union of Radio Science, URSI GASS 2014 |
---|---|
Country/Territory | China |
City | Beijing |
Period | 16/08/14 → 23/08/14 |