ChapterPDF Available

Abstract and Figures

Runtime enforcement is an effective method to ensure the compliance of program with user-defined security policies. In this paper we show how the stream event processor tool BeepBeep can be used to monitor the security properties of Java programs. The proposed approach relies on AspectJ to generate a trace capturing the program’s runtime behavior. This trace is then processed by BeepBeep, a complex event processing tool that allows complex data-driven policies to be stated and verified with ease. Depending on the result returned by BeepBeep, AspectJ can then be used to halt the execution or take other corrective action. The proposed method offers multiple advantages, notable flexibility in devising and stating expressive user-defined security policies.
Content may be subject to copyright.
Monitoring of Security Properties Using
Mohamed Recem Boussaha, Raphaël Khoury, and Sylvain Hallé
Laboratoire d’informatique formelle
Université du Québec à Chicoutimi, Canada
Runtime enforcement is an effective method to ensure the
compliance of program with user-defined security policies. In this paper
we show how the stream event processor tool BeepBeep can be used to
monitor the security properties of Java programs. The proposed approach
relies on AspectJ to generate a trace capturing the program’s runtime
behavior. This trace is then processed by BeepBeep, a complex event
processing tool that allows complex data-driven policies to be stated and
verified with ease. Depending on the result returned by BeepBeep, AspectJ
can then be used to halt the execution or take other corrective action.
The proposed method offers multiple advantages, notable flexibility in
devising and stating expressive user-defined security policies.
1 Introduction
Mobile code has emerged as an effective solution to the challenges of computing
in distributed systems. Nonetheless, security concerns remain omnipresent, and
may act as a break to the adoption of this technology, in part because of the need
for each user to tailor the security policy governing his system to his own need.
In this paper, we show how BeepBeep [5], a complex event processor can be
used as a runtime monitor for enforcing a wide array of user-defined security
policies. This study also serves to illustrate BeepBeep’s capabilities of as a log
trace analyzer. BeepBeep takes as input a data stream, in this case an execution
trace capturing the method calls and parameters values. This information can
be generated using any number of mechanism. BeepBeep has the capacity to
efficiently analyse this information in real-time to determine if it conforms
with a user defined specification. BeepBeep can also generate useful diagnostic
information about the program’s runtime behavior, which can in turn be used
for further security analysis or debugging.
We rely upon AspectJ [7] to generate the input trace which allows BeepBeep
to perform the enforcement. However, the monitoring using BeepBeep is agnostic
of the mechanism used to generate the traces, and while AspectJ also exhibits
some capabilities to operate as a security policies enforcement mechanism on
its own, that is not necessarily the case for other tracers. The use of BeepBeep
allows the security enforcement mechanism to be independent from the tracer.
Like other runtime security enforcement mechanism, the approach presented
in this paper is precise, in the sense that it reject only those executions that
violate the security policy, permitting safe executions of the program to proceed,
and it results in no false-positives or false-negatives. It is late, in the sense that
the execution is not halted until a violation is about to occur, thus allowing as
much of the execution to take place as is permissible given the security policy in
place. The main advantage of the proposed approach over other monitoring tools
is the flexibility and expressivity of the policy specification language.
The remainder of this paper is organised as follows. Section 2 surveys related
works. In section 3, we give an overview of the architecture of the security
enforcement mechanism proposed in this paper. Section 4 describes some of the
security properties we can enforce and Section 5 presents experimental results.
Concluding remarks are given in Section 6.
2 Related Work
The Naccio project [4] provides a library of Java security policies that are enforced
at runtime. Each policy replaces certain Java Virtual Machine (JVM) classes
as needed to allow enforcement, and the JVM must be modified to ensure that
the correct (security policy specific) class is used. Any policy part of Naccio’s
library, as well as a multitude of policies unavailable on that platform, can easily
be stated and enforced using BeepBeep.
Several tools leverages machine learning techniques and static analysis to
categorize Android applications (written in Java) as either malicious or benign.
The tool ANDRANA [3] relies on static analysis of the applications’s code to
create a vector of features for each application. Classification is then performed
to determine if the observed features a typical to those previously observed in
malware. Other classifier rely upon the app’s manifest file [9], it’s service life
cycle [6], API calls [1] or a combination of API calls and other statically detected
features [2]. Like other methods based on static analysis, these exhibits a risk of
false-positive and false-negatives, and are vulnerable to obfuscation.
Another countermeasure in the face of malicious mobile code is the reliance
upon of code certification. Code signing utilizes cryptographic keys to guarantee
the authorship of code. While a useful security tool, code signing only serves to
authenticate the author of a given code, but provides no guarantees as to its
actual behavior. The author may be wrongly trusted by a user, and even code
from reputable sources can exhibit an exploitable vulnerability.
The approach proposed in this paper is precise and thus risks neither false-
positives nor false-negatives. It can be applied to code of unknown origin, and
allows the user to easily customize the security policy to his needs. Indeed, as we
will show in the next section, it can be used not only to enforce a wide variety of
security policies but also to ensure the respect of resource-usage constraints or to
generate diagnostic reports about the program’s runtime behavior.
3 Architecture
The trace is generated using AspectJ, a tool that allows adding executable blocks
to the source code without explicitly changing it. AspectJ allows programmers
to set points in the source, known as pointcuts, to where the execution is to
temporarily halt and while the newly added code blocks are executed. In our
case, we used AspectJ to insert code before and after every method call to record
the information needed to perform monitoring. As mentioned above, this is only
one of several methods that could be used to generate the trace.
Figure 1 shows a sample of the trace of a simple Java tutorial program. Each
line correspond to either a single method call or method return. The former
begin with the keyword the keyword ‘call’ contains the following information:
the method return type, the method’s containing class, the method’s name, each
of the methods parameters type and value, and finally the method’s call level
on the stack. The later begin with the keyword ‘Return’ and contain the return
value. Values of literals, string and elementary types are provided explicitly but
those of objets are provided by references. Arrays are prefixed with ‘[’.
Call: return type : void // class: MonitoredProgram // method: main //
type param 1: class [Ljava.lang.String;// value param 1:[Ljava.lang.String;@72ea2f77//level: 0
Call: return type : void // class: // method: <init> //
type param 1: class java.lang.String // type param 2: int //
value param 1: // value param 2: 80 // level: 1
Return: Socket[,port=80,localport=51706]
Call: return type : class //
class: // method: getInetAddress // no parameter // level: 1
Fig. 1: A fragment of a trace
BeepBeep [5] is a complex event processing tool that can perform complex
manipulations on large data streams efficiently. Internally, BeepBeep decomposes
the desired data-processing task into a number of atomic processors, each of
which takes as input one (or more) event streams, and in turn, outputs one or
more event streams. These processor are chained together with the output of one
(or more) processor being piped to the input of the next one in such a manner
that, feeding BeepBeep’s input stream through this chain produces the desired
computation. Part of the contribution of this paper is to show how complex,
data-driven security properties of programs can be stated in terms of a small
number of BeepBeep processors.
A benefit of the approach under consideration is the ease with which the
desired security property can be stated. Each BeepBeep processors consists in an
average 20 lines of Java code, contained in a single class. Users can reuse these
elementary blocs, chaining them together to easily compose complex policies.
4 Security Properties
We began by replicating several of the security properties present in Naccio’s li-
brary. Most of these are safety policies and can be enforced with as little as one or
two BeepBeep Processors. Such properties include: NoExec, NoJavaClassLoader,
NoNetReceiveing, NoNetSending, NoPrinting, NoReadingFiles, NoListingFiles
LimitBytesWritten and LimitBytesRead, LimitCreatedFiles, LimitObservedFile.
The first 7 of these properties simply halt the execution upon encountering a
specific forbidden method call. The latter 4 are only slightly more involved. Limit-
BytesWritten and LimitBytesRead limit to total number of bytes that are written
(resp. read) to files or to the network. LimitCreatedFiles and LimitObservedFile
limit the number of files that can be created (resp. read).
Figure 2 gives an example of the BeepBeep processors for the property
LimitBytesWritten. It consists of only 3 processors: the first extracts from the
trace those method calls that perform write operation and passes those method
calls to the second processor. The second extracts number of bytes written
by from these method calls, and again passes this information on to the next
processors. The final processor computes the sum of the values it receives as input
and aborts the execution (through AspectJ) if this sum surpasses a customizable
value recorded in the property.
Fig. 2: The BeepBeep processors for property LimitBytesWritten
Schneider introduced the property [10] ‘no send after read’ as a typical example
of a safety property. This property states that after having read from a protected
file, the program is no longer allowed to access the network. This property is also
part of the Naccio library.
The expressiveness of the approach under consideration is illustrated by the
following pair of properties: the property ‘a is a key’, states that a given piece
of information provided in the trace, such a parameter to a specific method or
its return value, never take the same value twice. Conversely, it’s negation, the
property ’a is not a key’ requires that this value be unique. These properties are
interesting since, as observed by Segoufin [11], several widely used data models
can express one of these properties, but not the other, and neither of these
properties is part of the Naccio library. Both can be stated using relatively simple
processors, that stores the values that have appeared so far in the execution in a
list, and consult this list before allowing the execution to proceed.
Deserialization attacks [8] have recently emerged as important vector of
attack against Java programs. Any program that relies upon serialized objects to
exchange information with a distant party may be susceptible to a serialization
attack, even if the data is validated after having been received. The attack can be
performed in any one of several ways, notably by sending data of an unexpected
type, by sending an object of the correct type but with a high order nested
structure, leading to resource exhaustion when it is deserialized or by sending an
object whose fields values are not consistent with the normal execution of the
program. Beep Beep processors offer a simple and effective counter-measure in
all cases. Since the trace contains return values and their type, validation can
be performed with a processor similar to the ones described above. To protect
against a deserialization bomb, we developed a processor that bounds the number
of consecutive nested calls of the
method. BeepBeep can also
ensure important data secrecy properties by preventing data read in sensitive
files from being included in the serialized object.
BeepBeep allows us to state more complex policies that relate the values
present in different parts of the trace to one another. For example:
The parameter values of a given function are always increasing/decreasing in
consecutive calls. This property ensures the correctness of recursive function.
After being created, a given data object is not modified (data integrity).
No thread is frozen for more than 100 milliseconds before resuming its
execution (starvation freedom).
Whether two specific methods work on the same object, or alternatively
provide a list of objects that are manipulated by both of these methods.
Since the output of a processor can be of any format, BeepBeep can also
provide profiling information about the ongoing execution, such as maximal,
minimal and average stacks depth, the number of objects created for each
object type, etc. We present two final processors that illustrate this capability of
BeepBeep: processor
lists, for each method call in the
trace, the number of times it directly calls every other method. This information
is provided in the form of a directed weighed graph, in which each vertex is
labelled with a method name, and a vertex of weight
is present between vertexes
iff method
calls method
times in the trace. We give a schematic
representation of this processor in Figure 3. The processor
provides a list of every method that manipulates each data object. Recall that
data objects are identified by reference in the trace. These two processors, while
not security property enforcers, provide crucial information on data flow analysis
that is essential to debugging and to the enforcement of data flow policies.
action call
Constants TuplesMethods
3 4 5
8 9
Fig. 3: The BeepBeep processor chain for property CallSequenceProfiling
Figure 3 shows the chain of processors required to compute the call graph
from an execution trace. The core of this chain is the
processor, depicted as
number 8 in the figure. It takes two event streams as its input: the first (left-hand
side) is a stream of events of an arbitrary type; the second (top side) is a stream
of Boolean values. This processor internally maintains a stack of received events.
To this end, the Boolean stream acts as a push flag. When an event
and a
Boolean value
arrive at the processor’s inputs, two situations may occur. If
, the top of the stack (if not empty) is output but not removed, and
then pushed onto the internal stack if
. If
is ignored, and the
element at the top of the internal stack is popped and discarded.
The original stream of method events is first split in three (1); one of these
copies is given as the input to the
processor (8), while another is sent to a
processor (2). This processor evaluates the function that compares the
field of the method event with the constant
; the result is a stream of
Boolean values, indicating whether the incoming event is a method call (
) or a
method return (
). This stream itself is split in three (7), and one of these copies
is given as the push flag of the
processor. The stack is hence instructed to
push an incoming event when it is a method call, and to pop the top of the stack
when it is a method return. As a result, the output of the
processor is the
method event corresponding to the current method in the program’s execution.
A third copy of the original stream of events is sent to a
(3). This processor receives two inputs: an arbitrary event
and a Boolean value
called the filter flag. Event
is output if
, otherwise
is discarded. The
filter flag, in this case, is the result of applying the function
which returns a Boolean value; in other words, the processor keeps only method
call events, and filters out method returns. The same filtering condition is applied
to the output of the Stack processor (9).
The end result of this first part of the chain is that processors 3 and 9
synchronously output method call events; events at matching positions in the
streams represent the caller of a method (9) and the method being called (3).
From this point on, the rest of the processing is straightforward. Both events
are processed so that only the value of their
field is kept (4, 10); these two
values are then joined in a tuple (5), and these tuples are then accumulated into
a multiset using a CumulativeProcessor (6).
The output stream resulting from 6 is a sequence of multisets, each of the
m0, n0
mk, nk
; each tuple (
mi, ni
)is a caller-callee pair of method
names. The number of times each distinct pair occurs in the multiset corresponds
to the number of times
was called from
in the trace. From then on, it is
easy to take the multiset of tuples and covert it into a directed graph that shows
the weighted dependencies between methods in the observed execution.
It is worth mentioning that, in this whole graph, only the
(used in processors 2, 4 and 10) and
processor (8) are specific to our use
case. This amounts to 35 lines of custom code. All the remaining processors and
functions are generic, and already come in BeepBeep’s core or one of its existing
Figure 4 shows a more informative variant of the
It computes the number of bytes written by each function that does so, and
expresses this information in the form of a plot. The processor chain begins by
Constants Tuples
2 3 6
4 5
8 9
Fig. 4: The BeepBeep processor chain for property BytesWrittenGraph
extracting from the trace the methods that perform write operations, discarding
all other lines and pairs containing the method name and the number of bytes
written are joined in a tuple (processor 6). Processor 7 splits its input stream
into multiple distinct streams, each of which contains tuples originating from a
single method. This allows the computation of the number of bytes written to be
aggregated separately for each method (processors 9 and 10). The remainder of
the processor chain aggregates this information with a timestamp, and updates
a hash table accordingly. This hash table can then serve as the basis of a plot
generated on demand.
5 Experimental Results
We tested this method on traces of length 1 000 000, generated in the manner
described above on a Java calculator. Figure 5 plots the execution times (in ms.)
for four representative processor chains, namely
. As can be seen in these results, execution
times are largely proportional to the number of processors in each processor
chain. Since most security properties require only linear sequencing of processors,
their operation can easily be streamlined by merging the operations of multiple
monitor in a single class.
Figure 1 details the number of processors, number of custom lines of code
(not counting code already present in BeepBeep’s template library) and execution
time several of the processors mentioned in this paper. This table illustrates the
ease with with BeepBeep processors can be composed, often necessitating only a
minimal amount of custom code. Once these processor chains are implemented,
Fig. 5: Experimental Results
they can in turn be included as components of processor chains for more complex
properties with the addition of a single line of code.
The only processor chain whose execution time is not inconsiderable is
, described in the previous section. Much of its execution
time is incurred in the final processor, which updates a hash table linking each
method to the number of bytes that have been written by that method during
the program’s current execution. Since a BeepBeep processor chain manipulates
events sequentially, and each processor feeds its output to a successor, the final
processor of the
processor chain performs the update by
copying the current hash table before updating it with the information it has
received in its last input event. However, since in our particular case, the hash
table update is performed in the final processor of the processor chain. As a
result, it is possible to edit this processor so that is simply updates the hash table,
without making a copy. This revision brings
’s execution
time in line with that of other monitors of its size.
6 Conclusion
In this paper, we showed how the event processor BeepBeep can be used for
runtime enforcement. The approach is agnostic to the tracer used to generate the
trac and itself by the ease by which properties can be stated using BeepBeep’s
processor chain structure. BeepBeep is useful not only for stating and enforcing
security properties, but also ro generate useful diagnostic information about
the trace, as we also illustrated using examples. One avenue of further research
which we are currently exploring is to draw on BeepBeep’s capabilities to allow
us to express a more informative verdict that simply a boolean indication of
the respect\violation of the security property. For instance, the monitor could
provide indications as to which parts of the trace caused the violations, rate
its severity, and suggest weaker properties that are respected. This information
Property nb of size exec. time
processors (lines) (ms)
NoExec 2 6 1590
NoClassLoader 2 6 1636
NoNetwork 2 6 1654
NoReadingFiles 2 6 1699
IsKey 2 8 1883
LimitBytesWritten 3 11 1900
CallSequenceProfiling 8 35 3701
BytesWrittenGraph 13 53 14633
Table 1: Description of tested Monitors
could, in turn, serve as the basis for a more corrective reaction to a potential
violation than simply aborting the execution.
Aafer, Y., Du, W., Yin, H.: DroidAPIMiner: Mining API-Level Features for Robust
Malware Detection in Android, pp. 86–103. Springer International Publishing,
Cham (2013), 1_6
Arp, D., Spreitzenbarth, M., Hubner, M., Gascon, H., Rieck, K.: Drebin: Effective
and explainable detection of android malware in your pocket. In: NDSS. The
Internet Society (2014)
Bedford, A., Garvin, S., Desharnais, J., Tawbi, N., Ajakan, H., Audet, F., Lebel,
B.: Andrana: Quick and accurate malware detection for android. In: Foundations
and Practice of Security - 9th International Symposium, FPS, Québec City, QC,
Canada, October 24-25, 2016,. pp. 20–35 (2016)
Evans, D., Twyman, A.: Flexible policy-directed code safety. In: 1999 IEEE Sympo-
sium on Security and Privacy, Oakland, California, USA, May 9-12, 1999. pp. 32–45.
IEEE Computer Society (1999),
Hallé, S.: When RV meets CEP. In: Runtime Verification - 16th International
Conference, RV 2016, Madrid, Spain, September 23-30. pp. 68–91 (2016)
6. Khanmohammadi, K., Rejali, M.R., Hamou-Lhadj, A.: Understanding the service
life cycle of android apps: An exploratory study. In: Proceedings of the 5th Annual
ACM CCS Workshop on Security and Privacy in Smartphones and Mobile Devices.
pp. 81–86. SPSM ’15, ACM, New York, NY, USA (2015)
Kiczales, G., Hilsdale, E., Hugunin, J., Kersten, M., Palm, J., Griswold, W.G.: An
Overview of AspectJ, pp. 327–354. Springer Berlin Heidelberg (2001)
Lai, C., Microsystems, S.: Java insecurity: Accounting for subtleties that can
compromise code. IEEE Software (2008)
Sato, R., Chiba, D., Goto, S.: Detecting android malware by analyzing manifest
files. In: Proceedings of the Asia-Pacific Advanced Network
Schneider, F.B.: Enforceable security policies. ACM Trans. Inf. Syst. Secur. 3(1),
30–50 (2000),
Segoufin, L.: Automata and logics for words and trees over an infinite alphabet. In:
Computer Science Logic, 20th International Workshop, CSL 2006, 15th Annual
Conference of the EACSL, Szeged, Hungary, Sept. 25-29,. pp. 41–57 (2006)
... The constructor of the contract expects a string (lines [12][13][14] to initialize the myString variable, which will store the string persistently. Then, a method getMyString is provided to access the currently stored string (lines [16][17][18][19], as well as a setMyString method to modify it (21-24). ...
... Extensions of BeepBeep with predefined custom objects are called palettes; there exist palettes for various purposes, such as signal processing, XML manipulation, plotting, and finite-state machines. Over the past few years, BeepBeep has been involved in a variety of case studies [12,29,30,31,39,68], and provides built-in support for writing domain-specific languages [32]. A complete description of BeepBeep is out of the scope of this paper; the reader is referred to a recent tutorial [28] or to a recent textbook about the system [35]. ...
... These tuples are then forwarded to a Pump (lines 9-10), which forces the input source to consume its events (i.e., each line of the CSV file is read and consumed by the chain of processors until the end of the file). Afterwards, the pump is connected to the processor ParcelLifecycle (lines [12][13], which actually depicts the behaviour of the processor chain in Fig. 8. For the sake of simplicity, the content of this processor is not detailed here, but can be found in the corresponding Java project10, as well as the code of the other properties of this chapter. ...
Full-text available
The combination of the Internet of Things and blockchain-based technologies represents a real opportunity for supply chain and logistics protagonists, who need more dynamic, trustworthy and transparent tracking systems in order to improve their efficiency and strengthen customer confidence. In parallel, hyperconnected logistics promise more efficient and sustainable goods handling and delivery. This chapter shows how the Ethereum blockchain and smart contracts can be used to implement a shareable and secured tracking system for hyperconnected logistics. A simulation using the well-known AnyLogic software tool provides insights on the monitoring of properties depicting shipment lifecycle constraints through a stream of blockchain log events processed by BeepBeep 3, an open source stream processing engine.
... Although these workflows are, in principle, independent of any specific software tool or library, in this paper we focus on their representation and implementation as chains of objects from BeepBeep, an open source system 3 for processing logs and realtime event streams. The library is written in Java, and has become a mature system that has already been used in the past, among other things, for the runtime detection of bugs in software [111] and the monitoring of security policies [29]. ...
Full-text available
Information systems produce different types of event logs; in many situations, it may be desirable to look for trends inside these logs. We show how trends of various kinds can be computed over such logs in real time, using a generic framework called the trend distance workflow. Many common computations on event streams turn out to be special cases of this workflow, depending on how a handful of workflow parameters are defined. This process has been implemented and tested in a real-world event stream processing tool, called BeepBeep. Experimental results show that deviations from a reference trend can be detected in realtime for streams producing up to thousands of events per second.
Full-text available
Event logs and event streams can be found in software systems of very diverse kinds. For instance, workflow management systems and ERP platforms produce event logs in some common format based on XML. Financial transaction systems also keep a log of their operations in some standardized and documented format, as is the case for web servers such as Apache and Microsoft IIS. Network monitors also receive streams of packets whose various headers and fields can be analyzed. Recently, even the world of video games has seen an increasing trend towards the logging of players’ realtime activities. Analyzing the wealth of information contained in these logs can serve multiple purposes. Business process logs can be used to reconstruct a workflow based on a sample of its possible executions; financial database logs can be audited for compliance to regulations; suspicious or malicious activity can be detected by studying patterns in network or server logs. However, the available tools to process logs or streams of events are often large systems that are hard to setup, and even simple examples seem needlessly complicated. In this book, you will learn about BeepBeep, a versatile Java library intended to make the processing of event streams both fun and simple. Through more than a hundred simple, illustrated code examples, you will see how running event processing tasks can be done in just a few lines of code—and what is more, code that you actually understand. From generating plots to computing statistics and evaluating temporal logic specifications, BeepBeep can prove a handy addition to a developer’s toolbox.
Full-text available
This paper describes a plug-in extension of the BeepBeep 3 event stream processing engine. The extension allows one to write a custom grammar defining a particular specification language on event traces. A built-in interpreter can then convert expressions of the language into chains of BeepBeep processors through just a few lines of code, making it easy for users to create their own domain-specific languages.
Conference Paper
Full-text available
This paper is an introduction to Complex Event Processing (CEP) intended for an practicioners of Runtime Verification. It first describes typical CEP problems, popular tools and their query languages. It then presents BeepBeep 3, an event stream processor that attempts to bridge the gap between RV and CEP. Thanks to BeepBeep’s generic architecture and flexible input language, queries and properties from both fields can be efficiently processed.
Conference Paper
Full-text available
The fast growing use of the Android platform has been accompanied with an increase of malwares in Android applications. A popular way in distributing malwares in the mobile world is through repackaging legitimate apps, embedding malicious code in them, and publishing them in app stores. Therefore, examining the similarity between the behavior of malicious and normal apps can help detect malwares due to repacking. Malicious apps operate by keeping their operations invisible to the user. They also run long enough to perform their malicious tasks. One way to detect malicious apps is to examine their service life cycle. In this paper, we examine the service life cycle of apps. We extract various features of app services. We use these features to classify over 250 normal and malicious apps. Our findings show that malicious apps tend to use services to do their malicious operation and have no communication with the other components of the app, whereas the services in normal apps are usually bound to other components and send messages to notify users about the operations they perform. The results of this exploratory study can be used in the future to design techniques for detecting malicious apps using the classification of their service features.
Conference Paper
Full-text available
The threat of Android malware has increased owing to the increasing popularity of smartphones. Once an Android smartphone is infected with malware, the user suffers from various damages, such as the theft of personal information stored in the smartphones, the unintentional sending of short messages to premium-rate numbers without the user’s knowledge, and the ability for the infected smartphones to be remotely operated and used for other malicious attacks. However, there are currently insufficient defense mechanisms against Android malware. This study proposes a new method to detect Android malware. The new method analyzes only manifest files that are required in Android applications. It realizes a lightweight approach for detection, and its effectiveness is experimentally confirmed by employing real samples of Android malware. The result shows that the new method can effectively detect Android malware, even when the sample is unknown.
Conference Paper
Full-text available
Malicious applications pose a threat to the security of the Android platform. The growing amount and diversity of these applications render conventional defenses largely ineffective and thus Android smartphones often remain un-protected from novel malware. In this paper, we propose DREBIN, a lightweight method for detection of Android malware that enables identifying malicious applications di-rectly on the smartphone. As the limited resources impede monitoring applications at run-time, DREBIN performs a broad static analysis, gathering as many features of an ap-plication as possible. These features are embedded in a joint vector space, such that typical patterns indicative for malware can be automatically identified and used for ex-plaining the decisions of our method. In an evaluation with 123,453 applications and 5,560 malware samples DREBIN outperforms several related approaches and detects 94% of the malware with few false alarms, where the explana-tions provided for each detection reveal relevant properties of the detected malware. On five popular smartphones, the method requires 10 seconds for an analysis on average, ren-dering it suitable for checking downloaded applications di-rectly on the device.
Conference Paper
In order to protect Android users and their information, we have developed a lightweight malware detection tool for Android called Andrana. It leverages machine learning techniques and static analysis to determine, with an accuracy of 94.90%, if an application is malicious. Its analysis can be performed directly on a mobile device in less than a second and using only 12 MB of memory.
Conference Paper
The increasing popularity of Android apps makes them the target of malware authors. To defend against this severe increase of Android malwares and help users make a better evaluation of apps at install time, several approaches have been proposed. However, most of these solutions suffer from some shortcomings; computationally expensive, not general or not robust enough. In this paper, we aim to mitigate Android malware installation through providing robust and lightweight classifiers. We have conducted a thorough analysis to extract relevant features to malware behavior captured at API level, and evaluated different classifiers using the generated feature set. Our results show that we are able to achieve an accuracy as high as 99% and a false positive rate as low as 2.2% using KNN classifier.
Conference Paper
In a data word or a data tree each position carries a label from a finite alphabet and a data value from some infinite domain. These models have been considered in the realm of semistructured data, timed automata and extended temporal logics. This paper survey several know results on automata and logics manipulating data words and data trees, the focus being on their relative expressive power and decidability.
Conference Paper
The article introduces a new approach to code safety. We present Naccio, a system architecture that allows a large class of safety policies to be expressed in a general and platform-independent way. Policies are defined in terms of abstract resource manipulations. We describe mechanisms that can be used to efficiently and conveniently enforce these safety policies by transforming programs. We are developing implementations of Naccio that enforce policies on JavaVM classes and Win32 executables. We report on results using the JavaVM prototype
Conscientious Java developers are typically aware of the numerous coding guidelines that they should follow when writing code, such as validating inputs, minimizing accessibility to classes and members, and avoiding public static nonfinal fields. Java developers follow such guidelines to avoid common programming pitfalls (often called antipatterns), thereby reducing the likelihood of bugs or security vulnerabilities in their programs.
This paper addresses those questions for the class of enforcement mechanisms that work by monitoring execution steps of some system, herein called the target, and terminating the target's execution if it is about to violate the security policy being enforced. We call this class EM, for Execution Monitoring. EM includes security kernels, reference monitors, firewalls, and most other operating system and hardware -based enforcement mechanisms that have appeared in the literature. Our targets may be objects, modules, processes, subsystems, or entire systems; the execution steps monitored may range from fine-grained actions (such as memory accesses) to higher-level operations (such as method calls) to operations that change the security-configuration and thus restrict subsequent execution