Academic journal article Informatica Economica

Tools for Empirical and Operational Analysis of Mobile Offloading in Loop-Based Applications

Academic journal article Informatica Economica

Tools for Empirical and Operational Analysis of Mobile Offloading in Loop-Based Applications

Article excerpt

Offloading for mobile devices is an increasingly popular research topic, matching the popularity mobile devices have in the general population. Studying mobile offloading is challenging because of device and application heterogeneity. However, we believe that focusing on a specific type of application can bring advances in offloading for mobile devices, while still keeping a wide range of applicability. In this paper we focus on loop-based applications, in which most of the functionality is given by iterating an execution loop. We model the main loop of the application with a graph that consists of a cycle and propose an operational analysis to study offloading on this model. We also propose a testbed based on a real-world application to empirically evaluate offloading. We conduct performance evaluation using both tools and compare the analytical and empirical results.

Keywords: Mobile Computing, Offloading, Cloud Computing, Performance Evaluation

(ProQuest: ... denotes formulae omitted.)

1 Introduction

Modern handheld devices, such as smartphones and tablets, offer portability, in- creased computational power, and communi- cation capabilities. Thus, they are becoming an attractive option for users to interact with each other and with their environment.

The convergence of mobile and distributed computing has been studied for a number of years, with results in system design [1] [2], job scheduling [3] [4], resource discovery [5] [6] and so on. Mobile integration with vari- ous other types of computing takes many forms, such as mobile cloud computing, of- floading, delegation, cyber foraging, and data staging. Offloading is a form of transferring tasks of various granularities to remote re- sources. Offloading and delegation are very similar approaches to use remote resources and sometimes they are considered to be complementary, as in the work published by Flores [7]. Cyber Foraging is an opportunis- tic approach of using remote resources from mobiles. Satyanarayanan [8] introduced this concept in 2001 as a pervasive computing technique, and work within the same research team [9] led to a scripting language for cyber foraging. Verbelen et al. [2] introduced AI- OLOS, a middleware to improve mobile ap- plication performance through cyber foraging and Kristensen [10] introduced scheduling concepts in the topic.

Offloading for mobile devices is an increas- ingly popular research topic, matching the popularity mobile devices have in the general population. With the first research efforts targeted specifically on mobile devices da- ting back in the 1990's, in the past couple of years a vast material on offloading for mobile devices has been published.

Several offloading systems have been pro- posed, as middleware, frameworks, or ser- vices. However, we find that few solutions reach the point to have a big impact on live systems and applications. Studying mobile of- floading is challenging because of device and application heterogeneity. However, we be- lieve that focusing on a specific type of ap- plication can bring advances in offloading for mobile devices, while still keeping a wide range of applicability. Thus, our approach is to conduct an application domain exploration and select a family of applications on which to conduct analysis and evaluation of various offloading mechanisms.

In this paper, we conduct experimental and analytical evaluation for mobile offloading, as a step towards an integrated offloading system [11]. We also compare the two sets of results to study the trade-offs of the two methods.

2 Related Work

Analyzing power consumption is an interest- ing research topic, as shown by various pro- jects that can be found in the literature. Caroll and Heiser [12] design multiple mi- cro-benchmarks to associate the power costs to modules of a mobile system. They try to determine the power consumption of differ- ent parts like CPU, GSM and WiFi. Zhang et al. [13] describe a power module construc- tion technique that they use to characterize 3G, GSM, WiFi, CPU and screen, and to in- troduce PowerTutor, an Android application that can use these models for power con- sumption estimation on any device. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.