DocumentCode :
86926
Title :
Markov Decision Processes with Threshold Based Piecewise Linear Optimal Policies
Author :
Erseghe, Tomaso ; Zanella, A. ; Codemo, Claudio G.
Author_Institution :
Dipt. di Ing. dell´Inf., Univ. di Padova, Padua, Italy
Volume :
2
Issue :
4
fYear :
2013
fDate :
Aug-13
Firstpage :
459
Lastpage :
462
Abstract :
This letter investigates the structure of the optimal policy for a class of Markov decision processes (MDPs), having convex and piecewise linear cost function. The optimal policy is proved to have a piecewise linear structure that alternates flat and constant-slope pieces, resembling a staircase with tilted rises and as many steps (thresholds) as the breakpoints of the cost function. This particular structure makes it possible to express the policy in a very compact manner, particularly suitable to be stored in low-end devices. More importantly, the threshold-based form of the optimal policy can be exploited to reduce the computational complexity of the iterative dynamic programming (DP) algorithm used to solve the problem. These results apply to a rather wide set of optimization problems, typically involving the management of a resource buffer such as the energy stored in a battery, or the packets queued in a wireless node.
Keywords :
Markov processes; MDP; Markov decision processes; computational complexity; constant slope pieces; iterative dynamic programming algorithm; optimal policy; piecewise linear cost function; piecewise linear structure; resource buffer; threshold based form; threshold based piecewise linear optimal policies; wireless node; Dynamic programming; Markov decision processes; low complexity; optimization; piecewise linear policy;
fLanguage :
English
Journal_Title :
Wireless Communications Letters, IEEE
Publisher :
ieee
ISSN :
2162-2337
Type :
jour
DOI :
10.1109/WCL.2013.052813.130213
Filename :
6523044
Link To Document :
بازگشت