Location
Grand Wailea, Hawaii
Event Website
https://hicss.hawaii.edu/
Start Date
7-1-2020 12:00 AM
End Date
10-1-2020 12:00 AM
Description
In recent years, connected autonomous vehicles (CAVs) feature an increasing number of Ethernet-enabled electronic control units (ECUs), thereby creating more threat vectors that provide access to the Controller Area Network (CAN) Bus. Currently, mitigation techniques to protect the CAN bus from compromised ECU units in vehicle ad hoc networks (VANET) often utilize classical cryptographic techniques. However, ECUs often have temporal signatures that leak internal state information to eavesdropping attackers who can leverage temporal properties for longitudinal attacks. Unfortunately, these types of attacks are difficult to defend against using classical encryption schemes and intrusion detection systems (IDS) due to their high computational demands and ineffectiveness at protecting CAVs throughout the duration of their long lifespans. In order to address these problems, we propose a novel cryptographic framework that protects information embedded in ECU network communications by delivering an encryption system that periodically "salts" the temporal dynamics of individual ECU units with chaotic signals that are difficult to learn. We demonstrate the framework on two datasets, and our results show that the underlying temporal signatures cannot be approximated by state-of-the-art learning algorithms over finite time horizons.
Protecting Temporal Fingerprints with Synchronized Chaotic Circuits
Grand Wailea, Hawaii
In recent years, connected autonomous vehicles (CAVs) feature an increasing number of Ethernet-enabled electronic control units (ECUs), thereby creating more threat vectors that provide access to the Controller Area Network (CAN) Bus. Currently, mitigation techniques to protect the CAN bus from compromised ECU units in vehicle ad hoc networks (VANET) often utilize classical cryptographic techniques. However, ECUs often have temporal signatures that leak internal state information to eavesdropping attackers who can leverage temporal properties for longitudinal attacks. Unfortunately, these types of attacks are difficult to defend against using classical encryption schemes and intrusion detection systems (IDS) due to their high computational demands and ineffectiveness at protecting CAVs throughout the duration of their long lifespans. In order to address these problems, we propose a novel cryptographic framework that protects information embedded in ECU network communications by delivering an encryption system that periodically "salts" the temporal dynamics of individual ECU units with chaotic signals that are difficult to learn. We demonstrate the framework on two datasets, and our results show that the underlying temporal signatures cannot be approximated by state-of-the-art learning algorithms over finite time horizons.
https://aisel.aisnet.org/hicss-53/st/cybersecurity_and_sw_assurance/2