Information Technology Reference
In-Depth Information
sure whether network flows consumed by these applications are really what they
need, and among the same kind of applications, which is more provincial trac?
Existing researches on 3G network and smart phone applications are always
focus on common metrics in the general wireless network. Netdiff [12] is a bench-
mark used to compare different Internet Service Providers (ISP) performance.
M. C. Chan et. al. [11] proposed an algorithm to evaluate 3G networks TCP/IP
transmission rate and delay. H. Junxian et. al. [9] compared the performance
of different smart phone applications under different operator networks. But
there are almost no efforts on the study of operators charging system, neither
comparison among same kind of smart phone applications.
1.1 Our Contribution
The main contributions of our work are:
1. We have designed and implemented an expandable phone side evaluation
framework TraSt (Trac Statistics). This framework can get more detailed
trac usage than any other existing trac monitor frameworks or applica-
tions on Android. And since it is running on the phone, the result can better
reflect users real experience.
2. We have proposed some methods to utilize TraSt to explore four aspects of
the Chinese three major operators' charging systems (see section IV.A), and
get the following result: a. Operators are charging for the trac of network
layer. b. Some protocol overheads such as ARP and ICMPv6 are not taking
into account. c. Different operators have different minimum charging units,
B or KB. d. the phone will not receive any packet before it initiate a request.
These result will lead 3G users to a clearer understanding of the similarities
and differences among different operators' charging services.
3. We also have used TraSt to evaluate four kinds of popular smart phone
applications (browser, map, online music player and video call). We made
experiments in different scenarios, and accurate comparisons of network traf-
fic between different mobile applications. The results show that even among
same kind of applications existing great differences. Our analysis with rec-
ommendations to software developers can lead to better system design and
network infrastructure support.
The rest of the paper is organized as follows. Section 2 introduces data charg-
ing process of 3G networks and smart phone applications. Section 3 presents an
overview of TraSt. Section 4 proposes the issues and study methodology. Section
5 shows the result and Section 6 draws the conclusion and further discusses fu-
ture work. Since Android is the most widely used smart phone operating system
[1], the ensuing discussions are all made on the Android.
2 Preliminaries
We first give a brief overview of 3G charging architecture for data services in con-
text of UMTS [8]. Then introduce the classification of smart phone applications
