Blackdroid: A black-box way for Android plaintext
and ciphertext privacy leaks detecting and guarding
Yan Zhang
1
,Yazhe Wang
4
,Dan Wang
5
State Key Laboratory of Information Security
Institute of Information Engineering, CAS
Beijing, China
Ruoding Zhang
2
Institute of Information Engineering, CAS
Beijing, China
Qihui Zhou
3
Ocean University of China
Qingdao, China
Abstract—Constantly emerging third-party apps bring huge
convenience and enjoyment to Android users. However,
vulnerabilities of Android’s own permission management
mechanism allow apps to read user’s privacy data and send them
outside without the user’s consent. In this paper, we provide
Blackdroid, a light weight plaintext and ciphertext privacy leaks
detecting and controlling solution for Android. Based on preset
labels, Blackdroid tracks the flow of text privacy data in third-
party apps via a black-box way, and drops the bad packages
containing privacy if necessary. We tested 4428 popular apps
from Chinese Android market Gfan using Blackdroid. Of the
tested apps, 21.7% send out packages carrying sensitive contents.
In performance experiments, we observed only 4.5% time
overload gain in internet package sending out phase.
Keywords—Privacy guard; Dataflow tracing; Data label;
Black-box detecting
I. INTRODUCTION
Android smartphone provides third-party app installation
platform to users
[1] [2]
. Until 2013 October 10, official Android
Market already published more than 863,575 apps
[3]
, while
unofficial markets possess thousands of apps too. Android
provides a coarse granularity permission management
mechanism to restrict behaviors of these apps
[4] [5] [6]
. In
original Android, if an app needs access to a feature protected
by a kind of permission, it must declare that it requires that
permission in the manifest. Then, when the app is installed,
the installer checks the authorities that signed the app's
certificates and asks the user to determine whether to grant the
requested permission or not. If the permission is granted, the
app is able to use the features.
However, these permissions, IMEI reading, contacts
reading, internet accessing, bluetooth opening, and so on, are
easy to be misused to cause privacy leak on Android. That’s
because once an app gained both privacy data reading and
internet connecting permissions, during its whole running time,
its reading and sending out privacy behaviors (or we can say
its privacy leaking behaviors) can’t be monitored and defensed
by Android. Hence, if a user allows a contacts searching
optimization app to access his contacts list for speeding his
searching operations, and in the meantime allows it to access
the internet for fetching update info, he has no way of
knowing whether or not the app will send his contacts info to
advertisers or to any other entities.
To handle this problem, we proposed Blackdroid in this
paper, which is a text privacy leaks detecting solution for
Android that tracks the flow of text privacy data in third-party
apps via a black-box way. Blackdroid is a light weight security
extension to Android, which only requires setting data feature
labels for private text data (such as IMEI device number,
contacts list, SMS messages, SD card files, and app list) in the
source, and detect plaintext and ciphertext privacy leaks in the
internet outlet point, while leaving the whole privacy data
disposal processes in apps as a black box.
Using Blackdroid, we tested 4428 popular apps from one of
Chinese Android markets Gfan
[7]
, and found that 21.7% of
them sent out privacy text data. Furthermore, by evaluating
Blackdroid’s performance in experiments, we observed only
4.5% time overload in the internet package sending out phase
of an android smartphone deployed with it. No extra cost was
observed in other Android system behaviors.
The rest of this paper is organized as follows: Section 2
describes the design and implementation of Balckdroid .
Section 3 characterizes the evaluation of Blackdroid, Section 4
describes related works, and Section 5 concluded the paper.
II. B
LACKDROID DESIGN
In this section, we discuss the design of Blackdroid which
is an Android text privacy leak detection solution based on
black-box label tracking. In this solution, we adopt data
feature label (DFL for short) to identify user privacy data. An
internet outlet monitor (IOMonitor for short) is employed to
intercept data packages sent out by the apps and check
whether they contain labeled privacy data.
A. Data feature label
We define DFL as a quadrinominal <n, class, features, m>,
where n is the unique number to identify a label, class is the
name of a user text data category, features are a set of {f
1
,…,f
n
}
Supported by the National Natural Science Foundation of China under Grant
o.61202476, the Strategic Priority Research Program of the Chinese Academy
of Sciences, Grant No. XDA06010701,XDA06040502.