Automotive software needs to incorporate automation, connectivity, electrification, and sharing (ACES) to ensure safety and security are addressed.
A key challenge in the modern automobile is dealing with the increasing software. Today, there are more lines of code in the connected car than other highly sophisticated machines, including the U.S. Air Force F-35 Joint Strike Fighter, Boeing 787 Dreamliner, or the U.S. Space Shuttle.
The automotive industry is driven by a group of megatrends: automation, connectivity, electrification, and sharing, commonly referred to as ACES. ACES represents a new opportunity for the automotive industry to meet an entirely new set of challenges, but together they point to increasing software complexity.
Hardware today is more powerful and, as a result, millions of lines of code can be executed through a multitude of systems to perform complex functions inside the connected car. Soon these vehicles will communicate externally by way of vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communications. Safety and security are paramount concerns, so all onboard systems must be secure while the vehicle is in motion — or sitting idle.
Cybersecurity threats are ever-increasing
The “2020 Automotive Cybersecurity Report” (figure 1) from Upstream Security depicts a six-fold increase over a nine-year time period with numbers doubling from 2018 to 2019. The graph depicts a 94% year-over-year growth in cyberattacks since 2016. New business models will have to evolve as complexity, reliability, risk, and liability become primary drivers.
The increased effectiveness and proliferation of automotive cyberattacks have created a new urgency for security solutions, driving new regulations by lawmakers to prevent cyberattacks globally. The U.S. Security and Privacy in Your Car Act, or the “Spy Car Act of 2017,” defines requirements for protection against unauthorized data access and reporting. The bill directs the National Highway Traffic Safety Administration (NHTSA) to issue vehicle cybersecurity guidelines that require motor vehicles manufactured for sale in the United States to build in protection against unauthorized access to electronic controls and driving data.
Also in 2017, the U.S. House of Representatives passed H.R. 33886, “The SELF DRIVE Act,” a first-of-its kind legislation to ensure the safe and innovative development, testing, and deployment of self-driving automobiles. China established an automotive cybersecurity committee to ensure the safe operation of intelligent, connected and electric cars, including research, standards, policies, laws, and regulations. Other data regulations are beginning to emerge, such as the EU’s GDPR, Canada’s Digital Privacy Law (PIPEDA), and the European Parliament Transport Committee’s call for EU regulation on access to car data.
NHTSA’s Automotive Cybersecurity Research Program takes a threat analysis approach to cybersecurity, placing threats into six different categories:
Connected car attack surfaces
By understanding these threats, OEMs can look at four potential attack surfaces of the connected car:
Mapping attack surfaces to a vehicle’s architecture (Figure 2) depicts attack surfaces corresponding to a vehicle’s architecture. This basic schematic highlights connectivity within the car, including the use of automotive gateways and multiple vehicle buses, and different types of domains: infotainment, active safety (containing cameras and radar), and body. Chassis and powertrain ECUs utilize a controller area network (CAN) bus that can be easily exploited. Also shown are a variety of buses to communicate data within the central gateway. The central gateway ECU is a focal point of attack because of its direct exposure to the outside world.
It is quite clear that modern connected cars have multiple entry points, which hackers view as both a challenge and opportunity. To prevent any type of cyberattack, all entry points must maintain an appropriate level of security.
Security can be broken down into three aspects. The first aspect includes authentication and access control. Authentication means who is allowed to do things inside a vehicle. Access control is what the individual or system is allowed to do once inside. The second aspect to security is protection against illegitimate access, data leakages, or harmful software or Trojans from being installed. The final aspect to defining security is to detect and report security incidents.
A multi-layered security approach is needed today
Knowing the attack surfaces within the connected automobile provides the foundation for a multi-layered security approach. Automotive OEMs must secure all internal and external communications. An embedded firewall to protect the vehicle from accepting unauthorized traffic, data, or signals sent by a malicious IP address must be part of the mix. The following are critical components to secure a connected car:
Building a firewall into a vehicle is a highly specialized process tailored exclusively to the automotive environment. To build the firewall, a software development kit is integrated directly into the communications stack, whether TCP/IP, CAN, or other connected solution. The embedded firewall must be highly configurable with built-in flexibility, operate across a range of vehicle ECUs, and work with a real-time operating system (RTOS) or even in the AUTOSAR environment. Many cyber-attacks begin by sending packets to the connected car, seeking weaknesses, so if the firewall can detect this activity early and ensure certain packets are not allowed to be received or forwarded, a potential attack will be thwarted before it even begins. Controlling what ports and protocols are used to receive messages for the vehicle is crucial to protect and report suspicious activity.
Embedded firewalls for ECUs
Adding a firewall to a central gateway requires portable source code that can be integrated and configured into the ECU. Filtering rules built into the firewall block specific IP addresses and recognize unwanted activity with quick response to prevent an attack — firewall support of different types of filtering capabilities (CAN bus, rules-based, threshold-based, static) is critical, including stateful packet inspection.
Logging and reporting attacks enable intrusion detection, which is knowing when something unusual is happening. The connected vehicle must be able to report nefarious activity back to a vehicle operations center allowing security operations teams to take the necessary action and share that information across the security network.
A firewall on an external gateway ECU manages communication with all outside entities, serving as the bullseye for attack by enabling filtering rules for all vehicle communications. Its job is to detect and block attacks before they reach the target ECUs. A firewall on an internal gateway ECU is another option. With multiple networks within the car, an internal gateway ECU allows communication between different networks to isolate safety-critical functionality— the more critical internal systems are protected from potentially malicious network traffic. Finally, a firewall can be on an endpoint ECU, the actual control ECU that manages critical functionality in the vehicle. Control ECUs include anti-lock brakes, airbags, steering control, etc., so it is advisable to deploy a firewall on multiple endpoint ECUs.
There are numerous use cases for secure communication, including communication between the car and external systems, V2V communication, and communication within the car. V2V communication is more common and critical today, so it must be protected. To achieve secure communication within the car, all ECUs must be protected. As a communication session begins, the origin of that communication is known, so encryption is recommended. Encrypted communication uses IP protocols such as TLS, DTLS, and SSH. If running over a CAN bus, CANcrypt can be used. All data encrypted using strong cryptography is required to ward off cyberattacks.
During a communication session (Figure 3), authentication verifies that who you are communicating with is actually who they say they are, i.e., is the other device or process really who it claims to be? For authentication, a public key infrastructure (PKI) to manage and issue digital certificates is crucial. Every ECU must be identifiable and PKI-based certificates provide strong authentication for machine-to-machine communication. Another aspect of PKI security is code signing to enable secure boot and secure updates. With V2I communications, high-speed automated certificate issuance is mandatory since hosting and managing the entire process securely is essential. Where is the certificate authority hosted? How is certificate issuance performed? Is it automated? Is it secure? How are private keys protected?
Finally, an OEM may have its own internal strategy for securing the connected car with a proprietary safety ecosystem. But when considering V2I or V2V communications, where vehicles from multiple OEMs travel the same road, vehicle manufacturers must construct a shared ecosystem with the same requirements for security, management capabilities, and other safety-related capabilities to ensure interoperability among all vehicles on the road.
To protect today’s connected cars, multiple layers of security are required, and all attack surfaces must be taken into consideration. Most cyber-attacks remain undetected until it’s too late, so early detection is a must. As the connected car evolves, it is recommended that cybersecurity configuration be performed remotely with an enterprise security management system. This integration provides centralized management of security policies, situational awareness and device data monitoring, event management, and log file analysis for data analytics. Security needs to be a shared common resource. Embedded firewalls, secure communication, and strong authentication techniques are vital elements that constitute a multi-layered security approach.
This article was originally published on EE Times.
Ahmed Majeed Khan is an engineering leader, working with cross-functional groups to push the envelope of technology implemented in diverse automotive and consumer electronic domains. He holds a senior engineering management position at Siemens Digital Industries Software, where he assisted in the creation of a market-leading automotive-grade product portfolio. Khan is also Siemens’ focal point toward the international automotive software consortium, AUTOSAR.