DocumentCode :
566177
Title :
Online learning to optimize transmission over an unknown Gilbert-Elliott Channel
Author :
Wu, Yanting ; Krishnamachari, Bhaskar
Author_Institution :
Dept. of Electrical Engineering, University of Southern California, USA
fYear :
2012
fDate :
14-18 May 2012
Firstpage :
27
Lastpage :
32
Abstract :
This paper studies the optimal transmission policy for a Gilbert-Elliott Channel. The transmitter has two actions: sending aggressively or sending conservatively, with rewards depending on the action chosen and the underlying channel state. The aim is to compute the scheduling policy to determine which actions to choose at each time slot in order to maximize the expected total discounted reward. We first establish the threshold structure of the optimal policy when the underlying channel statistics are known. We then consider the more challenging case when the statistics are unknown. For this problem, we map different threshold policies to arms of a suitably defined multiarmed bandit problem. To tractably handle the complexity introduced by countably infinite arms and the infinite time horizon, we weaken our objective a little: finding a (OPT − (∈ + δ))-approximate policy instead. We present the UCB-P algorithm, which can achieve this objective with logarithmic-time regret.
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Modeling and Optimization in Mobile, Ad Hoc and Wireless Networks (WiOpt), 2012 10th International Symposium on
Conference_Location :
Paderborn, Germany
Print_ISBN :
978-1-4673-2294-2
Electronic_ISBN :
978-3-901882-47-0
Type :
conf
Filename :
6260468
Link To Document :
بازگشت