Information Technology Reference
In-Depth Information
1. Which layer in the protocol stack does each operator charging for?
2. Whether some protocol overhead is not taking into account?
3. What is the smallest units of charging for each operator?
4. Will smart phone receive packets passively without initiate a request? If so,
will this part of trac be charging?
The first two issues concern what to charge, the third concerns whether the
charging result is accurate, and the last concerns whether there exist some irra-
tional case in the charging process.
We need to describe in detail the last issue. In general, an application sends
a request to the server, the server reply to it. This trac should be charging
of course. But what if the application doesn't send any request, and the server
send packets to the device, will it receive the packets and will operators charge
for this trac? Obviously, it is unfair if the operators charging for this trac.
Even more frightening is that if the server is running a malicious program which
continue to send packets to a user, will bring huge losses to the user.
4.2 Issues in Smart Phone Applications
There are many same types of applications on Android markets. The main prob-
lem is:
1. How great the difference is among the same kind applications?
2. What makes the difference?
These issues are what users and developers most concerned separately.
4.3 Methodology
We first study the issues in the operators' charging system. We conduct a series
of experiments, get the data volume at the end device with TraSt, and compare
it with that recorded by operators. We run tests with the three major mobile
operators in China, which are China Mobile, China Unicom and China Telecom.
For the first issue, we open the network to run commonly used applications
such as browsers and maps and get the data volume of each layer in the protocol
stack (
V
UE
). Then close the network and get the operators' charging result (
V
Op
).
Calculate the RE (Relative Error) of each layer:
RE
=
|
V
UE
−
V
Op
|
×
100%
(1)
V
Op
We repeat the experiment three times, to see which layer's RE is the smallest.
This layer's trac is what the operators charging for. But at this moment, we
don't know whether operators are charging for some special trac, such as the
trac produce by TCP handshake. In order to make the ratio of special trac
as small as possible, we let the applications run enough time (10 minutes) so
that they can receive/send enough data.
