Skip to main content
Disciplines
Automotive Business IT + Informatics Construction + Real Estate Electrical Engineering + Electronics Energy + Sustainability Insurance + Risk Finance + Banking Management + Leadership Marketing + Sales Mechanical Engineering + Materials
Events
DE
EN
Books
Journals
Topic Page
Marketing
Start single access now
Access for companies
EXTENDED SEARCH
Log in
Top

2020 | Book

PAC: Privacy-Preserving Arrhythmia Classification with Neural Networks Read chapter Read first chapter

Foundations and Practice of Security

12th International Symposium, FPS 2019, Toulouse, France, November 5–7, 2019, Revised Selected Papers

Editors: Dr. Abdelmalek Benzekri, Prof. Michel Barbeau, Prof. Guang Gong, Dr. Romain Laborde, Joaquin Garcia-Alfaro

Publisher: Springer International Publishing

Book Series : Lecture Notes in Computer Science

Part of: Springer Professional "Wirtschaft+Technik" , Springer Professional "Technik" , Springer Professional "Wirtschaft"

Login to get access
insite
SEARCH

About this book

This book constitutes the revised selected papers of the 12th International Symposium on Foundations and Practice of Security, FPS 2019, held in Toulouse, France, in November 2019.

The 19 full papers and 9 short papers presented in this book were carefully reviewed and selected from 50 submissions. They cover a range of topics such as machine learning approaches; attack prevention and trustworthiness; and access control models and cryptography.

Table of Contents

Frontmatter

Machine Learning Approaches

Frontmatter
PAC: Privacy-Preserving Arrhythmia Classification with Neural Networks
Abstract
In this paper, we propose to study privacy concerns raised by the analysis of Electro CardioGram (ECG) data for arrhythmia classification. We propose a solution named PAC that combines the use of Neural Networks (NN) with secure two-party computation in order to enable an efficient NN prediction of arrhythmia without discovering the actual ECG data. To achieve a good trade-off between privacy, accuracy, and efficiency, we first build a dedicated NN model which consists of two fully connected layers and one activation layer as a square function. The solution is implemented with the ABY framework. PAC also supports classifications in batches. Experimental results show an accuracy of 96.34% which outperforms existing solutions.
Mohamad Mansouri, Beyza Bozdemir, Melek Önen, Orhan Ermis
Ransomware Network Traffic Analysis for Pre-encryption Alert
Abstract
Cyber Security researchers are in an ongoing battle against ransomware attacks. Some exploits begin with social engineering methods to install payloads on victims’ computers, followed by a communication with command and control servers for data exchange. To scale down these attacks, scientists should shed light on the danger of those rising intrusions to prevent permanent data loss. To join this arm race against malware, we propose in this paper an analysis of various ransomware families based on the collected system and network logs from a computer. We delve into malicious network traffic generated by these samples to perform a packet level detection. Our goal is to reconstruct ransomware’s full activity to check if its network communication is distinguishable from benign traffic. Then, we examine if the first packet sent occurs before data’s encryption to alert the administrators or afterwards. We aim to define the first occurrence of the alert raised by malicious network traffic and where it takes place in a ransomware workflow. Logs collected are available at http://​serveur2.​seres.​rennes.​telecom-bretagne.​eu/​data/​RansomwareData/​.
Routa Moussaileb, Nora Cuppens, Jean-Louis Lanet, Hélène Le Bouder
Using Machine Learning to Detect Anomalies in Embedded Networks in Heavy Vehicles
Abstract
Modern automobiles have more than 70 electronic control units (ECUs) and 100 million lines of code to improve safety, fuel economy, performance, durability, user experience, and to reduce emissions. Automobiles are becoming increasingly interconnected with the outside world. Consequently, modern day automobiles are becoming more prone to cyber security attacks. Towards this end, we present an approach that uses machine learning to detect abnormal behavior, including malicious ones, on embedded networks in heavy vehicles. Our modular algorithm uses machine learning approaches on the internal network traffic in heavy vehicles to generate warning alarms in real-time. We tested our hypothesis on five separate data logs that characterize the operations of heavy vehicles having different specifications under varying driving conditions. We report a malicious detection rate of 98–99% and a mean accuracy rate of 96–99% across all experiments using five-fold cross-validation. Our analysis also shows that with a small subset of hand-crafted features, the complex dynamic behavior of heavy vehicle ECUs can be predicted and classified as normal or abnormal.
Hossein Shirazi, Indrakshi Ray, Charles Anderson
Selection and Performance Analysis of CICIDS2017 Features Importance
Abstract
During the last decade network infrastructures have been in a constant evolution. And, at the same time, attacks and attack vectors become increasingly sophisticated. Hence, networks contain a lot of different features that can be used to identify attacks. Machine learning are particularly useful at dealing with large and varied datasets, which are crucial to develop an accurate intrusion detection system. Thus, the huge challenge that intrusion detection represents can be supported by machine learning techniques. In this work, several feature selection and ensemble methods are applied to the recent CICIDS2017 dataset in order to develop valid models to detect intrusions as soon as they occur. Using permutation importance the original 69 features in the dataset have been reduced to only 10 features, which allows the reduction of models execution time, and leads to faster intrusion detection systems. The reduced dataset was evaluated using Random Forest algorithm, and the obtained results show that the optimized dataset maintains a high detection rate performance.
Bruno Reis, Eva Maia, Isabel Praça
Semantic Representation Based on Deep Learning for Spam Detection
Abstract
This paper addresses the email spam filtering problem by proposing an approach based on two levels text semantic analysis. In the first level, a deep learning technique, based on Word2Vec is used to categorize emails by specific domains (e.g., health, education, finance, etc.). This enables a separate conceptual view for spams in each domain. In the second level, we extract a set of latent topics from email contents and represent them by rules to summarize the email content into compact topics discriminating spam from legitimate emails in an efficient way. The experimental study shows promising results in term of the precision of the spam detection.
Nadjate Saidani, Kamel Adi, Mohand Said Allili
Interpreting Machine Learning Malware Detectors Which Leverage N-gram Analysis
Abstract
In cyberattack detection and prevention systems, cybersecurity analysts always prefer solutions that are as interpretable and understandable as rule-based or signature-based detection. This is because of the need to tune and optimize these solutions to mitigate and control the effect of false positives and false negatives. Interpreting machine learning models is a new and open challenge. However, it is expected that an interpretable machine learning solution will be domain specific. For instance, interpretable solutions for machine learning models in healthcare are different than solutions in malware detection. This is because the models are complex, and most of them work as a black-box. Recently, the increased ability for malware authors to bypass antimalware systems has forced security specialists to look to machine learning for creating robust detection systems. If these systems are to be relied on in the industry, then, among other challenges, they must also explain their predictions. The objective of this paper is to evaluate the current state-of-the-art ML models interpretability techniques when applied to ML-based malware detectors. We demonstrate interpretability techniques in practice and evaluate the effectiveness of existing interpretability techniques in the malware analysis domain.
William Briguglio, Sherif Saad
Labelled Network Capture Generation for Anomaly Detection
Abstract
In the race to simplify man-machine interactions and maintenance processes, hardware is increasingly interconnected. With more connected devices than ever, in our homes and workplaces, the attack surface is increasing tremendously. To detect this growing flow of cyber-attacks, machine learning based intrusion detection systems are being deployed at an unprecedented pace. In turn, these require a constant feed of data to learn and differentiate normal traffic from abnormal traffic. Unfortunately, there is a lack of learning datasets available. In this paper, we present a software platform generating fully labelled datasets for data analysis and anomaly detection.
Maël Nogues, David Brosset, Hanan Hindy, Xavier Bellekens, Yvon Kermarrec

Attack Prevention and Trustworthiness

Frontmatter
Lempel-Ziv Compression with Randomized Input-Output for Anti-compression Side-Channel Attacks Under HTTPS/TLS
Abstract
Security experts confront new attacks on TLS/SSL every year. Ever since the compression side-channel attacks CRIME and BREACH were presented during security conferences in 2012 and 2013, online users connecting to HTTP servers that run TLS version 1.2 are susceptible of being impersonated. We set up three Randomized Lempel-Ziv Models, which are built on Lempel-Ziv77, to confront this attack. Our three models change the deterministic characteristic of the compression algorithm: each compression with the same input gives output of different lengths. We implemented SSL/TLS protocol and the Lempel-Ziv77 compression algorithm, and used them as a base for our simulations of compression side-channel attack. After performing the simulations, all three models successfully prevented the attack. However, we demonstrate that our randomized models can still be broken by a stronger version of compression side-channel attack that we created. But this latter attack has a greater time complexity and is easily detectable. Finally, from the results, we conclude that our models couldn’t compress as well as Lempel-Ziv77, but they can be used against compression side-channel attacks.
Meng Yang, Guang Gong
Secure Logging with Security Against Adaptive Crash Attack
Abstract
Logging systems are an essential component of security systems and their security has been widely studied. Recently (2017) it was shown that existing secure logging protocols are vulnerable to crash attack in which the adversary modifies the log file and then crashes the system to make it indistinguishable from a normal system crash. The attacker was assumed to be non-adaptive and not be able to see the file content before modifying and crashing it (which will be immediately after modifying the file). The authors also proposed a system called SLiC that protects against this attacker. In this paper, we consider an (insider) adaptive adversary who can see the file content as new log operations are performed. This is a powerful adversary who can attempt to rewind the system to a past state. We formalize security against this adversary and introduce a scheme with provable security. We show that security against this attacker requires some (small) protected memory that can become accessible to the attacker after the system compromise. We show that existing secure logging schemes are insecure in this setting, even if the system provides some protected memory as above. We propose a novel mechanism that, in its basic form, uses a pair of keys that evolve at different rates, and employ this mechanism in an existing logging scheme that has forward integrity to obtain a system with provable security against adaptive (and hence non-adaptive) crash attack. We implemented our scheme on a desktop computer and a Raspberry Pi, and showed in addition to higher security, a significant efficiency gain over SLiC.
Sepideh Avizheh, Reihaneh Safavi-Naini, Shuai Li
Enroll, and Authentication Will Follow
eID-Based Enrollment for a Customized, Secure, and Frictionless Authentication Experience
Abstract
High-assurance user identification and credentials provisioning are crucial for accessing digital services. Usability, service customization, and security should be carefully balanced to offer an appropriate user experience. We propose an eID-based enrollment approach for tailoring authentication to the particular needs of the service provider and strike a good trade-off between usability and security via the registration of authenticators, artifacts providing identity proofs. We demonstrate the practicality of our approach in the case of patient access to Electronic Health Records (EHR) through an Android application: enrollment is done by using the Italian national eID card to register the mobile authenticator, unlocked by the user’s fingerprint, customized to interact with the identity and access management system of the EHR.
Silvio Ranise, Giada Sciarretta, Alessandro Tomasi
TATIS: Trustworthy APIs for Threat Intelligence Sharing with UMA and CP-ABE
Abstract
Threat intelligence platforms offer cyber emergency teams and security stakeholders access to sightings of cyberthreats and indicators of compromise. Given the sensitivity of the information, access may be restricted to certain members within an organization, offered to the general public, or anything in between. Service providers that host such platforms typically expose APIs for threat event producers and consumers, and to enable interoperability with other threat intelligence platforms. Not only is API security a growing concern, the implied trust by threat event producers and consumers in the platform provider remains a non-trivial challenge. This paper addresses these challenges by offering protection against honest but curious platform providers, and putting the access control back into the hands of the owner or producer of the threat events. We present TATIS, a solution for fine-grained access control to protect threat intelligence APIs using User Managed Access (UMA) and Ciphertext-Policy Attribute-Based Encryption (CP-ABE). We test the feasibility of our solution using the Malware Information Sharing Platform (MISP). We validate our contribution from a security and privacy point of view. Experimental evaluation on a real-world OSINT threat intelligence dataset illustrates our solution imposes an acceptable performance overhead on the latency of API requests.
Davy Preuveneers, Wouter Joosen
Protecting Android Apps from Repackaging Using Native Code
Abstract
Android app repacking allows malicious actors to modify apps, bundle them with malware or steal revenue. Current detection mechanisms of app distribution services are questionable in their effectiveness, and other proposed repackaging protection schemes do not have the necessary protection against circumvention. We propose a repackaging protection architecture that verifies the app’s integrity at runtime. We make use of encrypted sections of bytecode that can be decrypted with a key derived at runtime. The method partially relies on native code, and as such is difficult to circumvent. We show that our implementation provides a practical integration in the workflow of an app developer.
Simon Tanner, Ilian Vogels, Roger Wattenhofer

Access Control Models and Cryptography

Frontmatter
Command Dependencies in Heuristic Safety Analysis of Access Control Models
Abstract
The principle merits of access control models lie in the ability to precisely reason about their security properties in lineage of the safety problem. It formalizes the question if future changes in a model’s protection state may eventually violate a security requirement, thereby falsifying model correctness. One fundamental problem of safety analysis is that, as proven in the seminal HRU model calculus, this property is undecidable for the most expressive class of models. To tackle this problem in practical security engineering, a heuristic approach has proven useful that exploits the fact that model commands share dependencies, which are assumed to be (1) one-dimensional and (2) static. In complex models for modern application domains, such as type enforcement in operating systems, both assumptions cannot be made. This paper studies both problems and provides a heuristic solution approach for the problem of dynamic dependencies. Based on our heuristic, we demonstrate the practical impact of this analysis problem and discuss the general implications on model design and analysis strategies.
Peter Amthor, Martin Rabe
On Attribute Retrieval in ABAC
Abstract
Despite the growing interest in Attribute-Based Access Control (ABAC) and the large amount of research devoted to the specification and evaluation of ABAC policies, to date only little work has addressed the issue of attribute management and retrieval. In many modern systems, the attributes needed for policy evaluation are often retrieved from external sources (e.g., sensors, access points). This poses concerns on the correctness of policy evaluation as the policy decision point can be provided with incorrect attribute values, which can potentially yield incorrect decisions. In this paper, we investigate the problem of selecting mechanisms for attribute retrieval and its relation with the accuracy of policy evaluation. We first introduce the notion of policy evaluation under error rate and use this notion to compute the evaluation accuracy of a policy. We formulate the Attribute Retrieval Mechanism Selection Problem (ARMSP) in terms of evaluation accuracy and show that ARMSP is exponential in the number of attribute values. To overcome this computation limitation, we investigate approaches to estimate the evaluation accuracy of a policy while maintaining the computation feasible.
Charles Morisset, Sowmya Ravidas, Nicola Zannone
Incorporating Off-Line Attribute Delegation into Hierarchical Group and Attribute-Based Access Control
Abstract
Efforts towards incorporating user-to-user delegation into Attribute-Based Access Control (ABAC) is an emerging new direction in ABAC research. A number of potential strategies for integrating delegation have been proposed in recent literature but few have been realized as full ABAC delegation models. This work formalizes one such strategy, entitled User-To-User Attribute Delegation, into a working delegation model by extending the Hierarchical Group and Attribute-Based Access Control (HGABAC) model to support dynamic and “off-line” attribute delegation. A framework to support the proposed delegation model is also presented and gives implementation details including an updated Attribute Certificate format and service protocol based on the Hierarchical Group Attribute Architecture (HGAA).
Daniel Servos, Michael Bauer
U-EPS: An Ultra-small and Efficient Post-quantum Signature Scheme
Abstract
Lamport and Winternitz signature schemes are well known one-time quantum resistant digital signature schemes. Along this line, several new one-time signature schemes are proposed. However, their private key and signature sizes are of \(\mathcal {O}(n^2)\) for \(k<n\)-bit security. Considering the applications in Internet of Things (IoT) and blockchains, \(\mathcal {O}(n^2)\) size is notably high. In this paper, we introduce a new one-time post-quantum signature scheme called U-EPS which achieve \(k=112\)-bit security with private key size 2n and signature sizes 3n bits (for \(n = 256\)), respectively. Our scheme only requires two calls of hash function and a single call of encryption/decryption algorithm for signature generation and verification procedures. We provide a concrete instantiation and implementation of U-EPS using SPIX-256 which is a NIST Lightweight Cryptographic Project Round 2 candidate. Finally, we give the comparison results with existing schemes.
Guang Gong, Morgan He, Raghvendra Rohit, Yunjie Yi
An Efficient Identification Scheme Based on Rank Metric
Abstract
Using random double circulant codes, we design a rank metric version of Aguilar, Gaborit and Schrek (AGS) identification scheme which is resistant to attacks using quantum computers. We achieve optimum results in different scales comparing public key size, secret key size and communication cost with known identification schemes. Moreover, our protocol is more efficient and practical for different resource constrained devices such as smart cards or Radio Frequency Identification (RFID) tags. Furthermore, using the Fiat-Shamir paradigm, we design an efficient signature scheme.
Edoukou Berenger Ayebie, Hafsa Assidi, El Mamoun Souidi
Security Analysis of Auctionity: A Blockchain Based E-Auction
Abstract
Auctions are widely used to sell products between different users. In this paper, we present Auctionity, an English e-auction based on blockchain. We describe the different protocols used in Auctionity. We also define the security models and the associated properties. We formally prove some security properties of this protocol using ProVerif.
Pascal Lafourcade, Mike Nopere, Jérémy Picot, Daniela Pizzuti, Etienne Roudeix
Dynamic Searchable Encryption with Access Control
Abstract
We present a searchable encryption scheme for dynamic document collections in a multi-user scenario. Our scheme features fine-grained access control to search results, as well as access control to operations such as adding documents to the document collection, or changing individual documents. The scheme features verifiability of search results. Our scheme also satisfies the forward privacy notion crucial for the security of dynamic searchable encryption schemes.
Johannes Blömer, Nils Löken

Short Papers

Frontmatter
Get-your-ID: Decentralized Proof of Identity
Abstract
In most systems without a centralised authority, users are free to create as many accounts as they please, without any harmful effect on the system. However, in the case of e-voting, for instance, proof of identity is crucial, as sybil identities can be used to breach the intended role of the system. We explore the conditions under which a decentralised proof of identity system can exist. We also propose such a scheme, called Get-your-ID (GYID), and prove its security. Our system allows a user to generate and revoke keys, via an endorsement mechanism, and we prove that under some conditions which we discuss, no user can have more than one active key. We then show how voting protocols can be adapted on top of our system, thus ensuring that no user is able to cast a valid vote more than once.
Pascal Lafourcade, Marius Lombard-Platet
Towards Secure TMIS Protocols
Abstract
Telecare Medicine Information Systems (TMIS) protocols aim at authenticating a patient in a telecare context, and permitting information exchange between the patient and a distant server through a verifier. In 2019, Safkhani and Vasilakos [10] showed that several protocols of the literature were insecure, and proposed a new protocol. In this paper, we show that their proposal is insecure, mainly due to incorrect use of distance bounding countermeasures, and propose a secure version, resistant to distance bounding related threats.
David Gerault, Pascal Lafourcade
Detecting Ransomware in Encrypted Web Traffic
Abstract
To date, only a small amount of research has focused on detecting ransomware at the network level, and none of the published proposals have addressed the challenges raised by the fact that an increasing number of ransomware are using encrypted channels for communication with the command and control (C&C) server, mainly, over the HTTPS protocol. Despite the limited amount of ransomware-specific data available in network traffic, network-level detection represents a valuable extension of system-level detection as this could provide early indication of ransomware activities and allow disrupting such activities before serious damage can take place. To address the aforementioned gap, we propose, in the current paper, a new approach for detecting ransomware in encrypted network traffic that leverages network connections, certificate information and machine learning. We leverage an existing feature model developed for general malware and develop a robust network flow behavior analysis model using machine learning that separates effectively ransomware traffic from normal traffic. We study three different classifiers: random forest, SVM and logistic regression. Experimental evaluation on a diversified dataset yields a detection rate of 99.9% and a false positive rate of 0% for random forest, the best performing of the three classifiers.
Jaimin Modi, Issa Traore, Asem Ghaleb, Karim Ganame, Sherif Ahmed
Digital Forensics in Vessel Transportation Systems
Abstract
Large vessels are safety-critical systems where operations, performance and component availability are continuously monitored by means of multiple sensors producing large amount of data. Relevant information is preserved in Event Data Recorders that are fundamental for the reconstruction of scenarios related to serious malfunctions and incidents in technical and legal terms. By considering the state-of-the-art and two important naval accidents we evidence some issues related to the exploitation of recorded data in reconstructing the events timeline and the semantics of the scenarios. These studies motivate our proposal that aims to guarantee strong data integrity and availability of all information registered in Event Data Recorders. Our results are fundamental for the precise identification of the sequences of events and for the correct attribution of human and/or machine responsibilities.
Alessandro Cantelli-Forti, Michele Colajanni
A Privacy Protection Layer for Wearable Devices
Abstract
The use of wearable devices is growing exceptionally which helps the individuals to track their daily activities, habits, health status, etc. They are be-coming so powerful and affordable that in some cases replace the use of mobile devices. Users are not aware of the extent and quality of the data being collected by these devices and the inherent risk of data security and privacy violation. This research introduces a privacy aware layer built over a sample smartwatch OS to limit user data access through enforcement of user set privacy settings. Since the nature of wearable devices leaves little space for interaction between the device and users, we develop a interface to capture user’s privacy preference. Two user studies, one using the Tizen platform and one with the privacy protection layer, are performed to compare its effectiveness.
Muhammad Mohzary, Srikanth Tadisetty, Kambiz Ghazinour
Validating the DFA Attack Resistance of AES (Short Paper)
Abstract
Physical attacks are a serious threat to the Internet of Things devices. Differential power analysis attacks are the most well-known physical attacks that exploit physical information leaked from hardware devices to retrieve secret information. Fault analysis attacks, a type of physical attack, are often considered more powerful than side-channel attacks if an attacker can inject the attacker’s intended faults. In fact, a few times of fault injections have enabled the attacker to retrieve the secret key. In this study, we propose a new model to validate the resistance of block ciphers to Differential Fault Analysis (DFA) attacks by assuming an ideal block cipher in which the differential probability is the same for all input and output differences. We show that Advanced Encryption Standard (AES) is near ideal for DFA attack resistance according to the experimental results.
Hakuei Sugimoto, Ryota Hatano, Natsu Shoji, Kazuo Sakiyama
A Rejection-Based Approach for Detecting SQL Injection Vulnerabilities in Web Applications
Abstract
According to OWASP top10 Application Security Risks [8, 9] SQL injection (SQLi) remains the most dangerous and most commonly exploited vulnerability in web applications. Thus, a lot of attentions are devoted by the scientific community for the development of SQLi verification tools. In this paper we focus on the development of an efficient, black box, SQLi vulnerability scanner to achieve an accurate detection. Our new approach is based on the use of structural similarity between rejection pages and their corresponding injection pages. A software prototype has been implemented and showed promising results as compared to well-known web application scanners.
Lalia Saoudi, Kamel Adi, Younes Boudraa
Lightweight IoT Mutual Authentication Scheme Based on Transient Identities and Transactions History
Abstract
Robust authentication of Internet of Things (IoT) is a difficult problem due to the fact that quite often IoT nodes are resource-constrained and lack the storage and compute capability required by conventional security mechanisms. We propose, in the current paper, a new lightweight multi-factor authentication scheme for IoT nodes that combines temporary identities and challenge/response based on transactions history. The approach allows IoT nodes to anonymously and mutually authenticate in an unlinkable manner. We evaluate the efficiency of the proposed scheme, and establish the security of our protocol through informal security analysis and formally by using the automated validation of Internet security protocols and applications (AVISPA) toolkit.
Mohammed Alshahrani, Issa Traore, Sherif Saad
Towards Privacy-Aware Smart Surveillance
Abstract
Cameras are rapidly becoming a daily part of our lives. Their constant streaming of information about people gives rise to different security and privacy concerns. Human analysis using cameras or surveillance footage has been an active field of research. Different methods have been introduced which showed success in both the detection and tracking of pedestrians. Once a human is detected and/or tracked, different motion analyses can be performed in order to better understand and model human behavior. A majority of these methods do not take user privacy or security into account, making security monitoring systems a significant threat to individuals’ privacy. This threat becomes more serious and evident when the security cameras are installed in places where vulnerable people (e.g. elders, children) frequently spend time such as day-cares, schools, retirement homes, or violated to serve independent interests. This work presents a model that is able to understand human motion, and deploys an anonymization technique that facilitates the preservation of an individual’s privacy and security.
Emil Shirima, Kambiz Ghazinour
Backmatter
Metadata
Title
Foundations and Practice of Security
Editors
Dr. Abdelmalek Benzekri
Prof. Michel Barbeau
Prof. Guang Gong
Dr. Romain Laborde
Joaquin Garcia-Alfaro
Copyright Year
2020
Electronic ISBN
978-3-030-45371-8
Print ISBN
978-3-030-45370-1
DOI
https://doi.org/10.1007/978-3-030-45371-8

Premium Partner

    Image Credits