Mobile Performance Metrics for Resource
Management
Kriszti´ an P´ andi
Department of Automation and Applied Informatics
Budapest University of Technology and Economics
Budapest, Hungary
Email: Pandi.Krisztian@aut.bme.hu
Hassan Charaf
Department of Automation and Applied Informatics
Budapest University of Technology and Economics
Budapest, Hungary
Email: Charaf.Hassan@aut.bme.hu
Abstract—In information technology two recent trends can
be observed. Mobile computer technology has greatly evolved
providing various rich functionalities. Capabilities of mobile
terminal is given, more performance can be obtained with adding
additional resources. Cloud computing is recognized to be a
new area for solving performance issues. These technologies
used together can enhance the user experience of the mobile
terminal. Sophisticated mobile resource management layer needs
to be used, that can utilize extra performance of the cloud. The
basis of effective resource management is a precise knowledge of
the hardware and software capabilities, furthermore, the usage
of mobile terminal. Performance metrics serve as an input for
resource management. This study will present architecture of
mobile performance measurement and profile creation layer. Key
performance metrics are collected and investigated. Attributes
of profile creation are also gathered. Architecture is examined,
whether it can be later used as an input for mobile resource
management
I. I NTRODUCTION
The aim of the study is to determine the performance of
the mobile terminal. Performance measurement and usable
metrics are mandatory for our later research, what is resource
management in mobile terminal. The goal of the resource
management is to decide where is the optimal place for a
certain service/application to run; on the mobile terminal itself,
or on public cloud computing server. Hence a performance and
usage of the mobile terminal should be determined.
Cloud computing promises [1] to provide high performance,
flexible and low cost on-demand computing services. Emerg-
ing complexity of the application used in mobile terminals
implicate harnessing these extra performance resources. Ap-
plications with distributed components differ from traditional
non distributed applications in numerous attributes, such as
communication type and overhead, latency, concurrency etc.
The task of mobile terminal resource and service manage-
ment is to decide where an application or service should be ex-
ecuted. To effectively fulfill this complex task a sophisticated
and dedicated decision formula is needed. As such a formula
amplifies adapted and dedicated software and performance
metrics. Mobile terminal coupled with distributed system can
be dynamic, changing over time, resulting CPU and network
load changing. Therefore mobile terminal as a part of the
distributed hierarchy needs to have very different metrics
than traditional software and performance metrics. Despite
of changing attributes, mobile terminal have a usage history,
where environment can recurrence. Therefore profile creation
is recommended, based on data collected from mobile terminal
usage.
With mobile computer technology progress, the software
and hardware platform becomes more and more complex,
together with the amount of the tasks meant to be processed.
Mobile terminals have some special features in comparison
with traditional computing; small size, dependence on lim-
ited battery lifetime, computing power is changing, possible
presence of 3D hardware, network bandwidth is limited, and
almost exclusively wireless, relatively small display size and
special user input.
Usually similar applications are used in mobile terminals
and in traditional computers thus similar user experience is
expected. Therefore, with comparably less performance nearly
the same look and feel is required. In consequence of that
capabilities of the mobile hardware should be efficiently har-
nessed with smart resource management and load balancing.
The main contributions of this paper are: (i) discussion on
applicable mobile performance measurement methodology ;
(ii) recommendation to gather the characteristics of mobile
phone usage and creating a user specific profile; (iii) discussion
on the architecture of mobile performance metrics and profile
creation architecture.
The rest of this paper is organized as follows: Section
II presents the related work. In Section III we discuss the
performance measurement methodology, based on that we
recommend a performance measurement and profile creation
layer in Section IV. In Section V conclusions are presented. Fi-
nally, in Section VI questions for the future work is described.
II. RELATED WORK
The need for measuring performance of computing ma-
chines arisen early, and solutions exists for this problem [2].
Early days were dedicated for creating standardized bench-
marking programs, and various of them were created (Standard
Performance Evaluation Corporation (SPEC) [3]) or EEMBC
[4] etc. Common attributes of these benchmarks is the batch
style execution, what is measuring program executing time
and speed, less is the executing time higher is the system
performance. All of these benchmarks are measuring not only
– 329 –
ICSSE 2013 • IEEE International Conference on System Science and Engineering • July 4-6, 2013 • Budapest, Hungary
978-1-4799-0009-1/13/$31.00 ©2013 IEEE