z-logo
Premium
Joint allocation of uplink and downlink resources for interactive mobile cloud applications
Author(s) -
Chen Jiadi,
Zheng Kan,
Long Hang,
Wang Wenbo
Publication year - 2016
Publication title -
transactions on emerging telecommunications technologies
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.366
H-Index - 47
ISSN - 2161-3915
DOI - 10.1002/ett.2867
Subject(s) - telecommunications link , computer science , resource allocation , cloud computing , mathematical optimization , computer network , markov decision process , distributed computing , real time computing , markov process , mathematics , statistics , operating system
Cloud‐based mobile applications are interactive in nature. Due to the strong correlation between uplink request and downlink response, it is irrational to optimize the uplink and downlink resource allocation problem separately and independently. In this paper, the joint uplink and downlink resource allocation problem for interactive mobile cloud applications is considered. The data exchange characteristics between mobile terminal and cloud server is investigated and a novel traffic model including both uplink and downlink transmission queues is constructed. Based on the new traffic model, the resource allocation problem can be formulated as a constrained Markov Decision Process. The objective is to minimize the application response time, that is, the roundtrip delay of causally related packet group (PG) pairs, with a constraint on PG drop rate. The Q‐learning algorithm and Value Iteration Algorithm are employed to obtain a solution that converges to the optimal one. To combat the curse of dimensionality, a sub‐optimal solution is proposed to solve the problem with an acceptable complexity. Numerical results indicate that the reduced complexity solution can achieve a delay performance approximate to that of the optimal solution and outperform other allocation schemes. Copyright © 2014 John Wiley & Sons, Ltd.

This content is not available in your region!

Continue researching here.

Having issues? You can contact us here