This paper proposes a novel intrusion detection algorithm that aims to identify malicious CAN messages injected by attackers in the CAN bus of modern vehicles. The proposed algorithm identifies anomalies in the sequence of messages that flow in the CAN bus and is characterized by small memory and computational footprints, that make it applicable to current ECUs. Its detection performance are demonstrated through experiments carried out on real CAN traffic gathered from an unmodified licensed vehicle.

Advanced Persistent Threats (APTs) are the most critical menaces to modern organizations and the most challenging attacks to detect. They span over long periods of time, use encrypted connections and mimic normal behaviors in order to evade detection based on traditional defensive solutions. We propose an innovative approach that is able to analyze efficiently high volumes of network traffic to reveal weak signals related to data exfiltrations and other suspect APT activities. The final result is a ranking of the most suspicious internal hosts; this rank allows security specialists to focus their analyses on a small set of hosts out of the thousands of machines that typically characterize large organizations. Experimental evaluations in a network environment consisting of about 10K hosts show the feasibility and effectiveness of the proposed approach. Our proposal based on security analytics paves the way to novel forms of automatic defense aimed at early detection of APTs in large and continuously varying networked systems

Security analytics and forensics applied to in-vehicle networks are growing research areas that gained relevance after recent reports of cyber-attacks against unmodified licensed vehicles. However, the application of security analytics algorithms and tools to the automotive domain is hindered by the lack of public specifications about proprietary data exchanged over in-vehicle networks. Since Controller Area Network (CAN) bus is the defacto standard for the interconnection of automotive electronic control units, the lack of public specifications for CAN messages is a key issue. This paper strives to solve this problem by proposing READ: a novel algorithm for the automatic Reverse Engineering of Automotive Dataframes. READ has been designed to analyze traffic traces containing unknown CAN bus messages in order to automatically identify and label different types of signals encoded in the payload of their data frames. Experimental results based on CAN traffic gathered from a licensed unmodified vehicle and validated against its complete formal specifications demonstrate that the proposed algorithm can extract and classify more than twice the signals with respect to previous related work. Moreover, the execution time of signal extraction and classification is reduced by two orders of magnitude. Applications of READ to CAN messages generated by real vehicles demonstrate its usefulness in the analysis of CAN traffic.

This paper evaluates the effectiveness of information-theoretic anomaly detection algorithms applied to networks included in modern vehicles. In particular, we focus on providing an experimental evaluation of anomaly detectors based on entropy. Attacks to in-vehicle networks were simulated by injecting different classes of forged CAN messages in traces captured from a modern licensed vehicle. Experimental results show that if entropy-based anomaly detection is applied to all CAN messages it is only possible to detect attacks that comprise a high volume of forged CAN messages. On the other hand, attacks characterized by the injection of few forged CAN messages attacks can be detected only by applying several independent instances of the entropy based anomaly detector, one for each class of CAN messages.

As cybersecurity detectors increasingly rely on machine learning mechanisms, attacks to these defenses escalate as well. Supervised classifiers are prone to adversarial evasion, and existing countermeasures suffer from many limitations. Most solutions degrade performance in the absence of adversarial perturbations; they are unable to face novel attack variants; they are applicable only to specific machine learning algorithms. We propose the first framework that can protect botnet detectors from adversarial attacks through Deep Reinforcement Learning mechanisms. It automatically generates realistic attack samples that can evade detection, and it uses these samples to produce an augmented training set for producing hardened detectors. In such a way, we obtain more resilient detectors that can work even against unforeseen evasion attacks with the great merit of not penalizing their performance in the absence of specific attacks. We validate our proposal through an extensive experimental campaign that considers multiple machine learning algorithms and public datasets. The results highlight the improvements of the proposed solution over the state-of-the-art. Our method paves the way to novel and more robust cybersecurity detectors based on machine learning applied to network traffic analytics.