A modern vehicle is no longer only a mechanical system.
Dozens of electronic control units can exchange information across multiple communication networks. Engine control, braking, steering, body electronics, transmission, diagnostics and other functions may depend on messages moving between these systems.
One of the most important technologies used for this communication is Controller Area Network, commonly known as CAN.
CAN was designed to provide reliable communication between electronic controllers in electrically noisy environments. It was not originally designed as a modern cybersecurity protocol with built in cryptographic authentication and encryption.
That distinction matters.
If an attacker gains an appropriate path to a vehicle network, the ability to observe, inject, replay or disrupt CAN traffic can create security risks.
This article explains the CAN security problem, common attack classes, practical investigation methods and how an authorized CAN security assessment can be performed without treating every unusual frame as an attack.
What Is CAN?
Controller Area Network is a communication protocol originally developed for reliable communication between electronic controllers without requiring a conventional host computer to manage every communication exchange.
In automotive systems, CAN became widely used because it is comparatively simple, robust and well suited to distributed electronic control systems.
Multiple electronic control units can share a CAN network and transmit messages identified by CAN identifiers.
The identifier also participates in the bus arbitration process, allowing higher priority messages to gain access to the bus when multiple nodes attempt to transmit.
This architecture provides important engineering advantages, but the same properties can create security challenges when an unauthorized device gains access to the network.
Why Is CAN Security a Problem?
The original CAN communication model does not inherently provide cryptographic authentication that allows a receiving ECU to verify the true identity of the sender.
It also does not inherently encrypt the payload.
SAE research has specifically described the need for additional measures to provide security properties such as integrity, authenticity, confidentiality and message freshness for automotive CAN communication. ([SAE International](https://saemobilus.sae.org/papers/study-secure-communication-automotive-applications-2017-01-1658))
More recent automotive research continues to describe CAN as intrinsically limited from a cybersecurity perspective because authentication and encryption are not built into the original protocol. ([SAE International](https://saemobilus.sae.org/papers/a-dynamic-cybersecurity-benchmarking-framework-automotive-cyber-physical-systems-2026-26-0612))
This does not mean that every CAN network is automatically vulnerable to remote attack.
The practical risk depends on the vehicle architecture, gateways, segmentation, diagnostic interfaces, wireless systems, ECU security controls and the attacker’s ability to obtain a path to the relevant network.
CAN Security in One Diagram
A simplified model looks like this:
External Interface
|
v
Telematics / Infotainment / Diagnostic Interface
|
v
Gateway
|
+-------------------+
| |
v v
CAN Network Other Networks
|
+---------+---------+
| | |
v v v
ECU A ECU B ECU C
The security question is therefore not simply “Is CAN secure?”
The more useful engineering question is:
“What paths can reach this CAN network, what messages can be injected or observed, and what controls prevent unauthorized communication from affecting safety or vehicle functions?”
What Makes a CAN Network Different From a Normal IP Network?
CAN should not be evaluated using exactly the same assumptions used for a conventional Ethernet or IP network.
CAN is a message oriented bus with arbitration based on identifiers and priority.
| Characteristic | CAN | Typical IP Network |
|---|---|---|
| Communication model | Message oriented bus | Packet based network |
| Addressing | CAN identifiers | IP and transport addresses |
| Bus access | Arbitration | Network switching/routing |
| Payload size | Limited | Much larger packets possible |
| Authentication | Not inherent to classic CAN | Usually provided by higher layer protocols |
| Encryption | Not inherent to classic CAN | Commonly implemented at higher layers |
This difference is important when designing automotive security controls.
What Is a CAN Identifier?
A CAN frame contains an identifier that helps determine the meaning and priority of the message within the network.
The identifier is not equivalent to a cryptographic identity.
This is a critical security distinction.
Knowing a legitimate identifier does not automatically prove that a legitimate ECU generated the frame.
Therefore, a security assessment should examine not only which identifiers exist, but also which ECUs are expected to transmit them, their timing, payload behavior and the controls protecting the network.
How CAN Bus Attacks Can Occur
A CAN attack generally requires some form of access to the communication path or to a component capable of influencing it.
Potential access paths can include:
- Diagnostic interfaces
- Compromised ECUs
- Telematics systems
- Infotainment systems
- Wireless interfaces
- Connected services
- Compromised gateways
- Physical access to the vehicle
Recent research has identified OBD-II and wireless-connected systems such as Bluetooth, Wi-Fi and cellular interfaces as potential paths through which an attacker may eventually reach vehicle communication networks, depending on the architecture. ([Nature Scientific Reports](https://www.nature.com/articles/s41598-025-98433-x))
The presence of an interface does not by itself prove exploitability. The complete attack chain must be evaluated.
Four Important CAN Attack Classes
CAN security research commonly discusses several attack categories.
- Message injection
- Replay
- Flooding or denial of service
- Masquerading or spoofing
Fuzzing and malfunction style attacks are also studied in automotive CAN intrusion detection research. ([Scientific Reports](https://pmc.ncbi.nlm.nih.gov/articles/PMC11002018/))
1. CAN Message Injection
Message injection occurs when an unauthorized device or compromised ECU introduces additional CAN frames into the network.
The security impact depends on which messages are injected, how often they are transmitted and how receiving ECUs interpret them.
For example, an attacker who knows that a particular identifier is associated with an important function may attempt to transmit frames using that identifier.
Without an appropriate authentication mechanism, the receiving ECU cannot rely on the identifier alone as proof of origin.
2. CAN Replay Attacks
A replay attack involves recording legitimate communication and transmitting it again later.
The problem is particularly interesting because the replayed frame can look structurally legitimate.
Research on secure CAN communication has identified message freshness as an important security property because a receiver needs a way to determine whether an apparently valid message is current or simply an old message being reused. ([SAE International](https://saemobilus.sae.org/papers/study-secure-communication-automotive-applications-2017-01-1658))
Frame counters and freshness mechanisms are among the approaches studied to address this problem.
3. CAN Flooding and Denial of Service
CAN arbitration gives priority to messages with higher bus priority.
An attacker who can repeatedly transmit high priority frames can consume bus capacity and interfere with legitimate communication.
This can create a denial of service condition.
Automotive security research describes flooding attacks using high priority identifiers as a mechanism for delaying or preventing legitimate ECU communication. ([Scientific Reports](https://www.nature.com/articles/s41598-025-98433-x))
4. CAN Spoofing and Masquerading
Masquerading occurs when an unauthorized source attempts to make a message appear as though it originated from another legitimate ECU.
This illustrates a fundamental weakness of relying only on message identifiers.
The receiver may recognize the identifier as legitimate without having cryptographic proof of which node actually produced the message.
Security research has specifically studied masquerading attacks in which an attacker observes message identifiers and transmission patterns and then attempts to impersonate another node. ([Nature Scientific Reports](https://www.nature.com/articles/s41598-025-98433-x))
What Is CAN Fuzzing?
CAN fuzzing involves sending systematically generated or malformed test frames to evaluate how ECUs and network components respond.
In a controlled laboratory assessment, fuzzing can help identify unexpected behavior, robustness problems and potentially exploitable protocol handling weaknesses.
It should only be performed against systems where the tester has explicit authorization and an appropriate safety-controlled environment.
CAN Security Is Not Just About Attacks
A professional assessment should not begin with the assumption that an attack must be performed.
The first objective should be understanding the architecture and determining where security controls are missing, weak or incorrectly implemented.
A useful assessment can answer questions such as:
- Which ECUs communicate on the network?
- Which CAN identifiers are observed?
- Which ECU appears to transmit each identifier?
- What are the normal transmission intervals?
- Are there unexpected identifiers?
- Are there unexpected changes in timing?
- Can unauthorized frames be introduced?
- Are message freshness controls implemented?
- Are security gateways separating network domains?
- Is intrusion detection available?
CAN Traffic Analysis
CAN traffic analysis is one of the most practical starting points for an authorized security assessment.
The objective is to understand normal communication before attempting to identify abnormal behavior.
A captured dataset may contain:
- Timestamp
- CAN identifier
- Data length
- Payload bytes
- Transmission frequency
- Bus load information where available
The exact information depends on the capture tool and format.
Why Establishing a Baseline Matters
An anomaly can only be identified meaningfully when normal behavior is understood.
For example, a CAN identifier that appears once during a diagnostic procedure may be perfectly legitimate but look suspicious if the analyst has no knowledge of the operating context.
A useful baseline should therefore include different operating conditions where available.
- Vehicle idle
- Normal operation
- Startup
- Shutdown
- Diagnostic sessions
- Relevant system state changes
What Should Be Measured During CAN Analysis?
| Metric | Why It Matters |
|---|---|
| CAN ID frequency | Helps identify normal message patterns. |
| Inter-arrival time | Helps identify timing anomalies. |
| Payload changes | Helps identify unexpected data behavior. |
| Bus load | Helps identify congestion or flooding. |
| Unknown IDs | May indicate unexpected network activity. |
| Message bursts | May indicate abnormal transmission behavior. |
| Direction/source information | Helps establish expected communication paths where available. |
CAN Intrusion Detection
Intrusion detection systems can monitor vehicle network traffic and attempt to identify abnormal communication.
NHTSA has specifically investigated methods for evaluating anomaly-based intrusion detection systems designed for vehicle data buses using CAN. Its work highlights the importance of objective performance assessment rather than assuming that an IDS is effective simply because it detects some attacks. ([NHTSA](https://www.nhtsa.gov/vehicle-safety-research))
Possible detection features include:
- Message frequency
- Timing patterns
- Identifier behavior
- Payload characteristics
- Bus load
- Known attack signatures
Research continues to investigate machine learning and other approaches to CAN intrusion detection, but detection performance must be evaluated in the context of false positives, computational cost and real vehicle conditions. ([USENIX VehicleSec ’26](https://www.usenix.org/conference/vehiclesec26/technical-sessions))
CAN IDS Is Not a Complete Security Architecture
An intrusion detection system can help identify suspicious activity, but detection is different from prevention.
A vehicle security architecture may require multiple layers.
- Network segmentation
- Secure gateways
- ECU security
- Message authentication
- Freshness protection
- Secure diagnostics
- Intrusion detection
- Secure software updates
The appropriate controls depend on the vehicle architecture and security requirements.
What Is CAN Message Authentication?
Message authentication provides a mechanism for a receiver to determine whether a message was generated by an authorized source and whether the protected information has been altered.
Cryptographic message authentication can use mechanisms such as message authentication codes.
SAE research has examined approaches using CMAC and other mechanisms to improve authenticity and integrity of CAN communication. ([SAE International](https://saemobilus.sae.org/papers/study-secure-communication-automotive-applications-2017-01-1658))
Why Message Freshness Matters
Authentication alone does not necessarily solve replay.
Suppose a message was legitimate when it was originally transmitted.
If an attacker records that valid message and sends it again later, the message may still contain a valid authentication value unless the security design also provides a mechanism to identify whether the message is current.
This is why freshness values, counters or timestamps can become important components of secure automotive communication designs.
CAN Security and AUTOSAR SecOC
Automotive security does not have to rely on the original CAN protocol alone.
Modern automotive architectures can introduce security mechanisms at higher layers.
AUTOSAR Secure Onboard Communication, commonly called SecOC, provides authenticated in-vehicle communication with replay and spoofing protection use cases, including support for low-bandwidth bus systems such as CAN. ([AUTOSAR](https://www.autosar.org/fileadmin/standards/R25-11/FO/AUTOSAR_FO_EXP_SecurityOverview.pdf))
SecOC does not mean that every CAN frame in every vehicle is automatically protected. Security architecture, configuration, key management and ECU implementation remain important.
CAN Security and ISO/SAE 21434
ISO/SAE 21434 should also be distinguished from a specific CAN security mechanism.
ISO/SAE 21434 defines engineering requirements for cybersecurity risk management across the lifecycle of road vehicle electrical and electronic systems, including components and interfaces. ([SAE International](https://saemobilus.sae.org/standards/isosae21434-road-vehicles-cybersecurity-engineering))
It is therefore a cybersecurity engineering framework, not a CAN encryption protocol.
A CAN security assessment can contribute evidence to a wider automotive cybersecurity engineering process, but the two should not be treated as interchangeable.
What Does a Practical CAN Security Assessment Look Like?
A professional assessment should begin with scope and authorization.
The assessment can then be structured into several phases.
Phase 1: Architecture Review
- Identify relevant ECUs.
- Identify CAN networks.
- Identify gateways.
- Identify diagnostic interfaces.
- Identify external communication paths.
- Understand network segmentation.
Phase 2: Traffic Collection
Collect authorized CAN traffic from a laboratory setup, ECU bench, development vehicle or customer-provided capture.
Record the operating conditions associated with each capture.
Phase 3: Baseline Analysis
- Identify CAN identifiers.
- Calculate message frequencies.
- Analyze timing.
- Identify payload patterns.
- Identify expected and unexpected traffic.
Phase 4: Security Analysis
- Evaluate injection exposure.
- Evaluate replay exposure.
- Evaluate flooding exposure.
- Review spoofing risks.
- Review diagnostic interfaces.
- Review gateway controls.
- Review authentication mechanisms.
Phase 5: Controlled Validation
Where explicitly authorized and technically safe, controlled testing can be performed against a laboratory or test environment to validate the identified security assumptions.
Live vehicle testing should require appropriate authorization, safety controls and an environment where unexpected ECU behavior cannot create unacceptable risk.
Phase 6: Reporting
Findings should be documented with evidence, affected components, security impact, severity, limitations and recommended remediation.
What Tools Can Be Used for CAN Security Analysis?
The correct tool depends on the assessment environment, CAN implementation and customer requirements.
| Tool Category | Example Use |
|---|---|
| CAN interface | Capture and transmit authorized CAN traffic. |
| CAN analysis software | Decode, filter and visualize frames. |
| Wireshark | Analyze supported CAN-related captures and higher-layer protocols. |
| Python | Automate CAN log analysis and anomaly detection. |
| SocketCAN | Linux-based CAN interface and analysis. |
| CAN utilities | Capture, inspect and generate controlled test traffic. |
| Oscilloscope | Investigate physical-layer behavior where required. |
The tool should follow the assessment objective. A large toolset does not automatically produce a better security assessment.
Can Wireshark Be Used for CAN Security?
Wireshark can be useful when the relevant CAN traffic is available in a supported capture or interface environment.
It can help analysts inspect traffic, timestamps, identifiers and protocol information depending on the capture format and protocol layers involved.
However, Wireshark is an analysis tool, not a complete automotive security assessment platform.
A professional workflow may combine packet analysis with CAN-specific tooling, scripts, ECU documentation and controlled test equipment.
Can CAN Security Be Assessed Remotely?
Yes, but the scope must be defined carefully.
A remote assessment does not necessarily mean remotely attacking a customer’s vehicle.
Many useful activities can be performed remotely using customer-provided data and controlled test environments.
- CAN log analysis
- CAN identifier analysis
- Message frequency analysis
- Timing analysis
- Anomaly analysis
- CAN architecture review
- Gateway security review
- CAN IDS evaluation
- Security documentation review
- Security report preparation
This model is particularly useful for organizations that already have a development vehicle, ECU bench or captured CAN datasets but need additional cybersecurity analysis.
What a Remote CAN Security Assessment Could Deliver
| Deliverable | Purpose |
|---|---|
| CAN communication map | Document observed communication relationships. |
| CAN ID inventory | Identify observed identifiers and patterns. |
| Traffic baseline | Document expected message behavior. |
| Anomaly findings | Identify unusual traffic requiring investigation. |
| Attack surface review | Identify potential communication entry points. |
| Security control review | Evaluate authentication, segmentation and detection controls. |
| Risk register | Prioritize security findings. |
| Remediation plan | Provide engineering recommendations. |
What a CAN Security Report Should Contain
A professional report should be useful to both engineering and management teams.
- Executive summary
- Assessment scope
- System assumptions
- Architecture overview
- Data sources
- Testing methodology
- Observed CAN traffic
- Security findings
- Evidence
- Risk rating
- Technical impact
- Business or safety relevance
- Recommended controls
- Limitations
- Next steps
Do Not Confuse CAN Analysis With Vehicle Hacking
CAN traffic analysis is one part of automotive cybersecurity.
A complete vehicle security assessment may also need to consider:
- Telematics
- Bluetooth
- Wi-Fi
- Cellular connectivity
- Mobile applications
- Cloud APIs
- Diagnostic protocols
- Automotive Ethernet
- OTA update systems
- ECU firmware
- Gateway architecture
Recent automotive cybersecurity research emphasizes that the vehicle attack surface extends beyond the CAN bus itself as connected services, OTA updates and other interfaces become part of the vehicle’s cyber-physical architecture. ([SAE International](https://saemobilus.sae.org/papers/a-dynamic-cybersecurity-benchmarking-framework-automotive-cyber-physical-systems-2026-26-0612))
CAN Security Assessment vs Full Automotive Cybersecurity Assessment
| Assessment | Primary Focus |
|---|---|
| CAN security assessment | In-vehicle CAN communication. |
| CAN-FD assessment | CAN-FD communication and associated security controls. |
| ECU security assessment | Individual ECU interfaces, software and security mechanisms. |
| Gateway assessment | Network segmentation and communication control. |
| Automotive network assessment | Multiple in-vehicle communication technologies. |
| Vehicle cybersecurity assessment | Broader vehicle attack surface and lifecycle security. |
How to Reduce CAN Security Risk
There is no universal control that makes every CAN network secure.
Controls should be selected according to the vehicle architecture and risk assessment.
- Segment critical networks.
- Use secure gateways.
- Restrict diagnostic access.
- Protect external communication paths.
- Use message authentication where appropriate.
- Implement freshness protection where required.
- Monitor for abnormal traffic.
- Secure ECU software.
- Protect keys and cryptographic material.
- Secure software update mechanisms.
AUTOSAR’s security architecture documentation specifically identifies authenticated in-vehicle communication with replay and spoofing protection as a SecOC use case and includes support for low-bandwidth systems such as CAN. ([AUTOSAR](https://www.autosar.org/fileadmin/standards/R25-11/FO/AUTOSAR_FO_EXP_SecurityOverview.pdf))
A Practical CAN Security Assessment Workflow
1. Define scope
↓
2. Understand vehicle architecture
↓
3. Identify CAN interfaces
↓
4. Collect authorized traffic
↓
5. Establish baseline
↓
6. Analyze IDs and timing
↓
7. Identify anomalies
↓
8. Evaluate attack exposure
↓
9. Review existing controls
↓
10. Validate safely in a test environment
↓
11. Prioritize findings
↓
12. Produce remediation report
What Should Engineers Ask Before Testing CAN Security?
- What CAN networks are in scope?
- Which ECUs are connected?
- Which identifiers are documented?
- Which identifiers are safety or security relevant?
- What external systems can reach the network?
- Are gateways enforcing communication policies?
- Are diagnostic interfaces protected?
- Are authentication mechanisms implemented?
- How is message freshness handled?
- Is CAN traffic monitored?
- What happens if unexpected traffic appears?
CAN Security Assessment Checklist
| Area | Assessment Question |
|---|---|
| Architecture | Are CAN networks properly segmented? |
| ECUs | Are communicating nodes identified? |
| CAN IDs | Are expected identifiers documented? |
| Traffic | Is normal traffic behavior understood? |
| Injection | Can unauthorized frames influence the network? |
| Replay | Is message freshness protected? |
| Spoofing | Can a node impersonate another source? |
| DoS | Can excessive traffic disrupt communication? |
| Diagnostics | Are diagnostic interfaces appropriately protected? |
| Gateway | Are network boundaries enforced? |
| IDS | Can abnormal traffic be detected? |
| Response | Is there an appropriate response to detected attacks? |
Final Perspective
CAN remains an important part of automotive electronic architecture, but its original communication model was not designed to provide all of the security properties expected from modern connected vehicles.
That does not make every CAN network inherently unsafe.
Security depends on the complete architecture surrounding the bus.
The relevant questions are where CAN can be reached, which messages matter, which ECUs are authorized to transmit them, what prevents unauthorized communication and how abnormal behavior is detected.
A mature automotive cybersecurity program therefore treats CAN as one part of a larger vehicle security architecture.
Understand the network first. Establish normal behavior. Identify the attack surface. Validate the security controls. Then make engineering decisions based on evidence.
Need a CAN Security Assessment?
Cautivra provides remote and laboratory-oriented automotive cybersecurity analysis for organizations that need help evaluating CAN communication, CAN traffic, ECU communication patterns, network architecture and security controls.
Depending on the project scope, an assessment can work with customer-provided CAN logs, ECU documentation, architecture diagrams, laboratory captures or an authorized test environment.
The objective is to produce evidence-based findings rather than simply generate a list of theoretical vulnerabilities.
Request a CAN Security Assessment
Frequently Asked Questions
Is CAN bus secure?
Classic CAN does not inherently provide modern cryptographic authentication or encryption. Security therefore depends on additional controls implemented at the ECU, gateway, communication or system architecture level.
Can CAN messages be spoofed?
Depending on the architecture and available access, unauthorized nodes may be able to transmit frames using identifiers associated with legitimate communication. This is why message authentication and other security controls can be important.
What are common CAN bus attacks?
Commonly studied attack classes include message injection, replay, spoofing or masquerading, flooding and denial of service, as well as fuzzing and other malformed traffic techniques.
Can CAN traffic be monitored for attacks?
Yes. CAN intrusion detection systems can analyze traffic characteristics such as message frequency, timing, identifiers and payload behavior to identify anomalies. NHTSA has researched methods for objectively evaluating CAN-based automotive IDS performance.
What is CAN bus penetration testing?
CAN penetration testing is an authorized security assessment of CAN-connected systems designed to identify and validate security weaknesses. It may involve traffic analysis, controlled injection, replay testing, fuzzing or other techniques in a suitable laboratory or test environment.
Can CAN security be assessed remotely?
Yes. Remote work can include CAN log analysis, communication mapping, anomaly analysis, architecture review, security control assessment and report preparation. Physical testing of a live vehicle requires appropriate equipment, authorization and a controlled environment.
What is the relationship between CAN security and AUTOSAR SecOC?
CAN itself does not provide the complete cryptographic security architecture required for modern vehicles. AUTOSAR SecOC is one mechanism that can provide authenticated in-vehicle communication with replay and spoofing protection, including use cases for CAN-based systems.
Is CAN security part of ISO/SAE 21434?
CAN security can be relevant to an automotive cybersecurity engineering process conducted under ISO/SAE 21434. However, ISO/SAE 21434 is a lifecycle cybersecurity engineering standard, not a CAN security protocol.
Sources and Further Reading
- SAE International: ISO/SAE 21434 Road Vehicles — Cybersecurity Engineering
- AUTOSAR: Security Overview
- NHTSA: Vehicle Cybersecurity Research
- SAE International: The Study of Secure CAN Communication for Automotive Applications
- SAE International: Automotive Cybersecurity Research
- Scientific Reports: CAN Bus Intrusion Detection Research