Multi-User Computation Partitioning for Latency Sensitive Mobile Cloud Applications

被引:213
作者
Yang, Lei [1 ]
Cao, Jiannong [1 ]
Cheng, Hui [2 ]
Ji, Yusheng [3 ]
机构
[1] Hong Kong Polytech Univ, Dept Comp, Hong Kong, Hong Kong, Peoples R China
[2] Liverpool John Moores Univ, Sch Comp & Math Sci, Liverpool L3 5UX, Merseyside, England
[3] Natl Inst Informat, Informat Syst Architecture Res Div, Tokyo, Japan
关键词
Mobile cloud computing; offloading; computation partitioning; job scheduling;
D O I
10.1109/TC.2014.2366735
中图分类号
TP3 [计算技术、计算机技术];
学科分类号
0812 ;
摘要
Elastic partitioning of computations between mobile devices and cloud is an important and challenging research topic for mobile cloud computing. Existing works focus on the single-user computation partitioning, which aims to optimize the application completion time for one particular single user. These works assume that the cloud always has enough resources to execute the computations immediately when they are offloaded to the cloud. However, this assumption does not hold for large scale mobile cloud applications. In these applications, due to the competition for cloud resources among a large number of users, the offloaded computations may be executed with certain scheduling delay on the cloud. Single user partitioning that does not take into account the scheduling delay on the cloud may yield significant performance degradation. In this paper, we study, for the first time, multi-user computation partitioning problem (MCPP), which considers the partitioning of multiple users' computations together with the scheduling of offloaded computations on the cloud resources. Instead of pursuing the minimum application completion time for every single user, we aim to achieve minimum average completion time for all the users, based on the number of provisioned resources on the cloud. We show that MCPP is different from and more difficult than the classical job scheduling problems. We design an offline heuristic algorithm, namely SearchAdjust, to solve MCPP. We demonstrate through benchmarks that SearchAdjust outperforms both the single user partitioning approaches and classical job scheduling approaches by 10 percent on average in terms of application delay. Based on SearchAdjust, we also design an online algorithm for MCPP that can be easily deployed in practical systems. We validate the effectiveness of our online algorithm using real world load traces.
引用
收藏
页码:2253 / 2266
页数:14
相关论文
共 23 条
[1]  
[Anonymous], 2010, P 8 INT C MOB SYST A, DOI [10.1145/1814433.1814441, DOI 10.1145/1814433.1814441]
[2]  
Canepa G., 2010, P 1 ACM WORKSH MOB C, P37
[3]  
Chun BG, 2011, EUROSYS 11: PROCEEDINGS OF THE EUROSYS 2011 CONFERENCE, P301
[4]   Optimal scheduling algorithm for distributed-memory machines [J].
Darbha, S ;
Agrawal, DP .
IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS, 1998, 9 (01) :87-95
[5]  
Giurgiu I, 2009, LECT NOTES COMPUT SC, V5896, P83, DOI 10.1007/978-3-642-10445-9_5
[6]  
Hall LA, 1996, PROCEEDINGS OF THE SEVENTH ANNUAL ACM-SIAM SYMPOSIUM ON DISCRETE ALGORITHMS, P142
[7]  
Kosta S, 2012, IEEE INFOCOM SER, P945, DOI 10.1109/INFCOM.2012.6195845
[8]   CLOUD COMPUTING FOR MOBILE USERS: CAN OFFLOADING COMPUTATION SAVE ENERGY? [J].
Kumar, Karthik ;
Lu, Yung-Hsiang .
COMPUTER, 2010, 43 (04) :51-56
[9]   Dynamic critical-path scheduling: An effective technique for allocating task graphs to multiprocessors [J].
Kwok, YK ;
Ahmad, I .
IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS, 1996, 7 (05) :506-521
[10]   BRANCH-AND-BOUND METHODS - A SURVEY [J].
LAWLER, EL ;
WOOD, DE .
OPERATIONS RESEARCH, 1966, 14 (04) :699-+