Two operators at a control-room desk with a row of monitors, a laptop and keyboards

OT/ICS security

ISO 15118 security: Plug & Charge PKI, TLS, and the SLAC MITM gap

Last updated:

CISA's CVE-2025-12357 advisory, based on research from Southwest Research Institute, confirms that the SLAC pairing step used with ISO 15118-2 can be manipulated to insert a man-in-the-middle between an EV and a charging station — exploitable wirelessly, at close range, via electromagnetic induction — while the separate, independently verified Brokenwire attack can abort an active charging session from 47 meters away using less than 1 watt of radio power. ISO 15118 is the vehicle-to-charger protocol (distinct from OCPP, the separate station-to-backend protocol); it defines everything from powerline pairing to the Plug & Charge PKI that lets a car authorize payment with a certificate instead of an RFID card. Both documented weaknesses above are standards-level behaviors present across implementations, not one vendor's bug, and the industry's own stated fix for the MITM gap is migrating from the optional, one-way TLS of ISO 15118-2 to the mandatory mutual TLS of ISO 15118-20.

Key takeaways

  • ISO 15118 is the vehicle-to-charger communication standard, and it secures only the later steps of a charging session.
  • Plug & Charge (PnC) authorizes payment with a certificate the car presents, with no card or app.
  • Through a multi-branch PKI under a V2G root, with separate certificates for the charger, the vehicle and the contract.
  • Through the charger's OCPP connection to its backend (CSMS), so the backend is part of the PnC trust chain.

What is ISO 15118, and what does it actually secure?

ISO 15118 is the vehicle-to-charger communication standard, and it secures only the later steps of a charging session. It links the Electric Vehicle Communication Controller (EVCC) in the car with the Supply Equipment Communication Controller (SECC) in the charger, and the gaps sit in the steps before TLS starts. The free ISO 15118-3:2015 preview defines SLAC (Signal Level Attenuation Characterization) and the EV-to-EVSE "matching" process; the ISO 15118-20:2022 preview lists "ISO 15118-2:2014" and "ISO 15118-3:2015" among its normative references and scopes -20 to the use of "V2GTP, TLS, TCP and IPv6".

Session step Defined in Protection
Basic signalling over the control pilot IEC 61851-1 None
HomePlug Green PHY link and SLAC matching ISO 15118-3 Network key handed over in the clear during matching
SECC Discovery (SDP): IP address, port, TLS or not ISO 15118-2 / -20 None; plain UDP
SupportedAppProtocol: DIN 70121, -2 or -20 ISO 15118-2 / -20 Inside TLS only if TLS was chosen
Charging and payment messages ISO 15118-2 or -20 TLS optional in -2, mandatory in -20

This guide covers only the vehicle-to-charger link. The charger-to-backend link is OCPP, and the two meet in certificate handling: the Open Charge Alliance's OCPP 2.0.1 white paper states that "OCPP 2.0.1 fully supports the ISO 15118-2 protocol" and that Plug & Charge authorization "is supported, as well as all functions to install and update ISO 15118 certificates". Backend threats are covered in our OCPP protocol security guide.

What is Plug & Charge, and how does it differ from Autocharge or EIM?

Plug & Charge (PnC) authorizes payment with a certificate the car presents, with no card or app. That contract certificate is tied to an EMAID (e-Mobility Account Identifier), and the charger validates it against a V2G root. Two other identification modes exist in real deployments:

  • External Identification Means (EIM): RFID card, app or QR code. Authorization happens outside the vehicle-to-charger link, so PnC-specific PKI risks don't apply.
  • Autocharge: a non-standardized practice built on DIN SPEC 70121 that identifies the vehicle by its MAC address, with no cryptographic binding. Cloning a MAC address is trivial.
  • Plug & Charge: the only one built on signed, revocable credentials, at the cost of a far more complex trust chain.

How does the Plug & Charge PKI and contract-certificate chain actually work?

Through a multi-branch PKI under a V2G root, with separate certificates for the charger, the vehicle and the contract. The CharIN PKI page publishes Certificate Policies for a "first-generation PKI compliant to ISO 15118-2" and a "second-generation PKI compliant to ISO 15118-20". The second-generation Certificate Policy (v1.0, 23 November 2022) states that "There SHALL be only one V2G Root CA in the V2G PKI", appointed by the CharIN PnC Europe Governance Body, and defines these branches:

Certificate Held by Used for
CSO (SECC) leaf Charging station TLS server authentication to the vehicle
Vehicle certificate EVCC TLS client authentication to the charger (ISO 15118-20)
OEM Provisioning certificate EVCC, from manufacturing Lets the e-MSP encrypt new contract credentials for this vehicle
Contract certificate EVCC, issued by the e-MSP Payment authorization (carries the EMAID)
Certificate Provisioning Service (CPS) Provisioning service Signs contract credentials for delivery via a contract certificate pool

The policy requires the V2G Root CA and the Tier-1 e-MSP and CSO CAs to "operate an OCSP Responder". Other roots exist: Hubject publishes Certificate Policies for its own ISO 15118-2 and -20 V2G PKIs. A charger maker therefore ships trust anchors for more than one root, and PnC security depends on revocation actually being checked in the field.

How do ISO 15118 certificates travel through OCPP?

Through the charger's OCPP connection to its backend (CSMS), so the backend is part of the PnC trust chain. The Open Charge Alliance application note Using ISO 15118 Plug & Charge with OCPP 1.6 maps each step to the OCPP 2.0.1 use cases and wraps the messages in DataTransfer for 1.6:

  1. Charger certificate (A02/A03)The charger generates a key pair and sends a SignCertificate request with a CSR; the backend has a CA sign it and returns it in CertificateSigned. The charger must reject an invalid certificate.
  2. Contract certificate install or update (M01/M02)The EV sends CertificateInstallationReq; the charger forwards it as Get15118EVCertificate with action Install or Update, and relays the backend's response to the vehicle.
  3. Revocation (M06)The charger sends GetCertificateStatus and the backend returns the OCSP response.
  4. Inventory (M03–M05)The backend lists, deletes and installs root certificates on the charger.

For a charger maker, this means certificate handling sits in two codebases, the ISO 15118 stack and the OCPP client, and both need the same validation rules.

What does CISA's CVE-2025-12357 advisory say about the SLAC man-in-the-middle gap?

It says spoofed SLAC measurements let an attacker sit between the car and the charger. The advisory's CSAF record reads: "By manipulating the Signal Level Attenuation Characterization (SLAC) protocol with spoofed measurements, an attacker can stage a man-in-the-middle attack between an electric vehicle and chargers that comply with the ISO 15118-2 part. This vulnerability may be exploitable wirelessly, within close proximity, via electromagnetic induction."

Key facts from the record:

  • Weakness: CWE-923, improper restriction of communication channel to intended endpoints; reported by Mark I. Johnson of Southwest Research Institute.
  • Score: first published 30 October 2025 at CVSS 3.1 8.3; Update A on 17 March 2026 "Adjusted CVSS score based on feedback from ISO/IEC" to 6.3 (AV:A/AC:L/PR:N/UI:N/S:U/C:L/I:L/A:L).
  • Mitigation: "ISO recommends using TLS for all communications in accordance with ISO 15118-20. While the use of TLS is recommended in ISO 15118-2, it is required in the ISO 15118-20 revision. TLS should be implemented with certificate chaining."
  • Exposure: no known public exploitation, and "not exploitable remotely".

The SwRI announcement adds that the team confirmed the flaw over a direct connection and also demonstrated it "wirelessly, via electromagnetic induction, in a benchtop simulation". Why SLAC is so exposed: during matching the charger hands the vehicle the HomePlug network membership key, and the DCeption authors note that the "cleartext transmission of the NMK" leaves it open to eavesdroppers, while the HomePlug Green PHY "Secure SLAC" option "is explicitly not used by any existing version of CCS". Because the flaw comes from the standard's own requirements, there is no vendor patch to wait for: the control is TLS with certificate validation above SLAC.

What is the Brokenwire attack, and how is it different from the SLAC flaw?

Brokenwire is a disruption attack: it stops a session rather than reading or forging it. The NDSS 2023 paper shows that HomePlug Green PHY, the powerline carrier ISO 15118-3 mandates, can be disrupted wirelessly by exploiting its mandatory CSMA/CA behaviour:

  • Effective at ranges up to 47 m for a power budget of less than 1 W, exceeding broadband noise jamming "by three orders of magnitude".
  • Tested against eight vehicles and 20 chargers.
  • "The exploited behavior is a required part of the HomePlug Green PHY, DIN 70121 & ISO 15118 standards and all known implementations exhibit it."

The researchers' FAQ notes that most home chargers use AC charging under IEC 61851 and "won't be affected", and warns that interrupting critical vehicles "such as electric ambulances" can have life-threatening consequences. TLS does not help here: the attack works below it.

How does ISO 15118-20 change the TLS and authentication model versus ISO 15118-2?

ISO 15118-2 (2014) ISO 15118-20 (2022)
TLS Optional; needed for Plug & Charge Mandatory
TLS version TLS 1.2 TLS 1.3
Authentication Charger to vehicle only Mutual: vehicle certificate as TLS client
Version selection SupportedAppProtocol handshake Same handshake, so a session can still land on -2 or DIN 70121

The open-source EVerest charger stack documents the -20 requirement directly: its ISO 15118-20 module manifest says -20 "mandates TLS 1.3 mTLS", that "ISO 15118-20 is TLS-only ([V2G20-2677])", and that each protocol is offered "in the SupportedAppProtocol handshake".

The catch is fallback. Whether TLS is used at all is decided in SECC Discovery, a plain UDP exchange, and the protocol version in SupportedAppProtocol. A charger that still offers -2 without TLS, or DIN 70121, can be steered there by anyone who can inject traffic on the link. The DCeption paper (January 2026) demonstrated exactly this over the air: "full control over TLS usage and CCS protocol version negotiation, including TLS stripping attacks". -20 removes the weakness only for sessions that cannot fall back.

What has research measured in deployed chargers?

Field data shows the gap between the standard and deployments. Four published studies:

Study Finding
Szakály, Köhler, Martinovic 2024, 325 European DC chargers "only 12%" support TLS at all, 6% implement standards-compliant TLS 1.2, none implement mutual TLS; every charger accepts non-TLS sessions; most HomePlug modems run decade-old firmware
Same study, certificates Charger certificates chain to Hubject or Entrust roots; one manufacturer's devices shared two leaf certificates, one expired
DCeption 2026 Wireless MITM with TLS stripping; a vehicle displayed over 900 kW while receiving 40 kW, and a vehicle was overcharged at twice the requested current for 17 seconds
Löw et al. 2025, USENIX VehicleSec 2026 Relay attack on Plug & Charge: a fake charger relays the victim car's signature so the attacker charges on the victim's contract

Worked example: what should a charger maker check before a release?

Synthetic example, not a real product. A DC charger with a Linux controller and a HomePlug modem module is about to ship a firmware update:

Check Evidence to collect Result in this example
Protocols offered in SupportedAppProtocol Session log with a test vehicle or EV emulator DIN 70121, -2, -20
TLS offered in SDP for each Same log TLS for -20 only
Certificate chain and expiry SECC leaf and sub-CA certificates Leaf expires in 4 months
OCSP status check Logs of status requests Not configured
HomePlug modem firmware Module vendor's version string Five years old
Software components Firmware SBOM matched against CVE data OpenSSL and BusyBox findings to triage

Decisions: offer TLS for -2 sessions, enable OCSP, plan certificate renewal, and ask the modem vendor for a security update. The last row is where a firmware SBOM and CVE matching help; the protocol rows need a test bench.

What are the common pitfalls?

  • Treating "ISO 15118-20 capable" as "TLS enforced". Fallback to -2 or DIN 70121 removes it.
  • Trusting SLAC to identify the vehicle. It is a proximity measurement, not authentication.
  • Shipping with one hard-coded root when vehicles chain to more than one V2G PKI.
  • Letting SECC certificates expire in fielded chargers without a renewal path.
  • Forgetting the modem. HomePlug firmware is a separate component with its own update path.

What should EV charging hardware and backend vendors do?

  1. Measure what your firmware negotiatesin a real session (protocol, TLS, authentication direction); a "Plug & Charge ready" label does not answer it.
  2. Offer TLS on every protocol version you still support, and plan to retire non-TLS sessions; per CISA's record, this is the mitigation for CVE-2025-12357.
  3. Validate the full chain and revocation, using OCSP as the CharIN policy requires, and keep a renewal process for SECC certificates.
  4. Plan for RF disruptionon DC chargers serving critical fleets; Brokenwire is not fixed by TLS.
  5. Treat 15118-20 as a hardware decisionwhere controllers cannot run TLS 1.3 with larger certificates.
  6. Track ISO 15118 and OCPP advisories separately, since they sit at different trust boundaries, and feed both into the vehicle-side risk analysis that ISO/SAE 21434 and UN R155 expect.

SoftWSP covers the software side of that list: it extracts a CycloneDX SBOM from charger or controller firmware and matches the components against NVD, CISA KEV, FIRST EPSS and CISA ICS advisories, with an AI-drafted, cited report that an engineer reviews. It does not test TLS, SLAC or PKI behaviour. See KEV and EPSS prioritisation for ranking the findings, the NIST SP 800-82 guide for the OT context, or device & firmware security services to have an image reviewed.

This guide draws on CISA's CSAF record for ICSA-25-303-01, SwRI's announcement, the ISO 15118-3 and -20 previews, the EVerest ISO 15118-20 module, CharIN's and Hubject's PKI documents, the Open Charge Alliance's OCPP 2.0.1 paper, and peer-reviewed and preprint research (Brokenwire, Szakály et al., DCeption, Löw et al.), read on 2026-11-14. The ISO 15118 standards themselves are paywalled.

Put this guide to work

Frequently asked questions

What is ISO 15118, and what does it actually secure?

ISO 15118, "Road vehicles — Vehicle to grid communication interface," is the international standard (developed jointly by ISO and IEC starting in 2010, with the first parts published 2013-2014) that governs digital communication between an electric vehicle (EV) and the charging station it is plugged into — technically, between the Electric Vehicle Communication Controller (EVCC) in the car and the Supply Equipment Communication Controller (SECC) in the charger. It is distinct from OCPP (Open Charge Point Protocol), the separate backend protocol connecting the charging station itself to a network operator's management system; ISO 15118 covers only the vehicle-to-charger link. The standard defines the full session: low-level signaling per IEC 61851, powerline network pairing via the SLAC (Signal Level Attenuation Characterization) protocol, IPv6 addressing, the V2GTP transport protocol, and — where implemented — TLS encryption of the session plus the Plug & Charge (PnC) authentication flow that lets a vehicle authorize payment automatically via digital certificates instead of an RFID card or app. Security (TLS, certificate-based authentication) is present in the standard but, in the widely deployed ISO 15118-2 (2014) edition, it is optional rather than mandatory, which is the root of most of the documented exposure described below.

What is Plug & Charge, and how is it different from Autocharge or RFID/app authorization?

Plug & Charge is the ISO 15118 feature where a driver plugs in the cable and the vehicle and charging network authenticate and authorize payment automatically, with no card tap or app interaction, using a public-key-infrastructure (PKI) model: the EV holds a contract certificate (tied to a unique EMAID, the e-Mobility Account Identifier assigned by the driver's mobility service provider) that it presents to the charging station, which validates it against a V2G Root certificate chain. This is a materially different trust model from two older or simpler alternatives still in the field. External Identification Means (EIM) is the fallback non-PnC path — an RFID card, app, or QR code — where authorization happens outside the vehicle-to-charger link entirely. Autocharge, based on DIN SPEC 70121, is a lighter-weight, non-standardized predecessor that identifies the vehicle by its network MAC address with no certificate chain at all; because it has no cryptographic binding, some automakers (Volkswagen among them, per public reporting) now randomize vehicle MAC addresses for privacy, which breaks Autocharge compatibility as a side effect. Plug & Charge, by contrast, is the only one of the three built on signed, PKI-issued credentials rather than a static identifier or an out-of-band token.

How does the Plug & Charge PKI and contract-certificate trust chain actually work?

CharIN (Charging Interface Initiative e.V.), the industry body that also stewards the Combined Charging System (CCS), operates a V2G Root Certificate Authority open to any participating OEM, charge point operator (CPO), or e-mobility service provider (eMSP) that accepts its governance terms, documented in separate Certificate Policies for a first-generation PKI (compliant with ISO 15118-2) and a second-generation PKI (compliant with ISO 15118-20). In the contract-certificate flow, a Certificate Provisioning Service (CPS) signs the contract data — including the vehicle's EMAID — with a certificate chain that traces back to the CharIN V2G Root, and the signed contract certificate is placed in a Contract Certificate Pool (CCP) for installation into the vehicle, either at point of sale or over the air. During a charging session, the vehicle presents this contract certificate, and the charging station (or its backend) validates the signature chain up to the V2G Root already trusted by the ecosystem, without needing to contact the vehicle's home mobility provider in real time. CharIN's PKI participation rules require all participants to publish certificate status via OCSP or CRL so a revoked contract certificate can actually be checked, and to support cross-root interoperability so a contract issued by one PKI participant's CPS is still honored by a charging station operated by a different one.

What does CISA's CVE-2025-12357 advisory say about the SLAC pairing man-in-the-middle flaw?

CVE-2025-12357, disclosed by Mark I. Johnson of Southwest Research Institute (SwRI) and published via a CISA advisory, is an "improper restriction of communication channel to intended endpoints" weakness (CVSS 3.1 base score 6.3, vector AV:A/AC:L/PR:N/UI:N/S:U/C:L/I:L/A:L — network-adjacent, low complexity, no privileges or user interaction required) in the SLAC (Signal Level Attenuation Characterization) pairing step that ISO 15118-2 uses to determine which physical charging station a given vehicle is actually plugged into over the shared powerline network. CISA's advisory states the flaw lets an attacker manipulate or spoof SLAC signal-attenuation measurements to insert a man-in-the-middle between the EV and the charger, and notes it may be "exploitable wirelessly, within close proximity, via electromagnetic induction" — meaning a direct physical splice into the charging cable is not strictly required. Because SLAC pairing is a required step of the standard itself (specified in ISO 15118-3) rather than a bug in one vendor's implementation, CISA's advisory states the issue affects any EVSE or vehicle implementing ISO 15118-2 SLAC pairing, independent of firmware version. CISA's and ISO's own stated mitigation is to move to ISO 15118-20 with TLS and certificate chaining enforced for all sessions, alongside conventional network isolation and VPN guidance for remote access to charging infrastructure.

What is the Brokenwire attack, and how is it different from the SLAC MITM finding?

Brokenwire (presented at NDSS 2023, "Brokenwire: Wireless Disruption of CCS Electric Vehicle Charging") is a denial-of-charging attack, not an eavesdropping or impersonation attack like the SLAC MITM flaw — it does not read or forge data, it simply stops a legitimate session from continuing. ISO 15118-3 mandates HomePlug Green PHY (HPGP) as the physical-layer powerline carrier for the ISO 15118/DIN 70121 control channel, and HPGP's OFDM-based, CSMA/CA (carrier-sense multiple-access with collision avoidance) design — built to tolerate noisy electrical environments — behaves enough like an over-the-air RF protocol that it can be coupled into wirelessly, not just via a direct wire tap. The researchers showed that exploiting the mandatory CSMA/CA collision-avoidance behavior lets an attacker disrupt the link with a radio signal of under 1 watt, over three orders of magnitude more power-efficient than plain broadband jamming, at ranges up to 47 meters (151 feet), using off-the-shelf radio hardware, and validated it against 8 vehicles and 20 commercial chargers in real deployments. Because the ISO 15118 standard specifies that a charging session aborts automatically once short control messages are lost, the attack's effect — an interrupted, non-resumable charging session — follows directly from the standard's own fault-handling behavior. The researchers state plainly that "the exploited behavior is a required part of the HomePlug Green PHY, DIN 70121 & ISO 15118 standards and all known implementations exhibit it," making this a standards-level gap rather than an implementation defect in any single charger or vehicle.

How does ISO 15118-20 change the TLS and authentication model compared to ISO 15118-2?

ISO 15118-2 (2014), the widely deployed first generation, made TLS 1.2 optional and, where used, one-directional: the charging station (SECC) authenticates itself to the vehicle, but the standard provides no equivalent cryptographic mechanism for the vehicle to prove its identity to the charger beyond a challenge-response exchange carried inside that one-way channel. ISO 15118-20 (2022), the second generation, makes TLS 1.3 the baseline and requires mutual authentication: both the EVCC and SECC must present and validate a certificate chain before energy transfer begins, which requires issuing a vehicle-side TLS certificate in addition to the existing contract certificate used for Plug & Charge authorization. A session's actual protocol version is chosen in the SupportedAppProtocol handshake from the versions both sides offer, and whether TLS is used at all is decided before that in SECC Discovery — meaning a mixed fleet of 15118-2-only vehicles and 15118-20 chargers (or vice versa) will silently fall back to the weaker, optional, one-way TLS model. Multiple industry and research sources describe this fallback, plus the reality that TLS remains optional in practice on much of the installed 15118-2 base, as the reason mutual TLS is "not currently widely used" across real-world CCS deployments despite being specified since 2022; hardware is also a factor; several industry write-ups note that most ISO 15118-2 charger control boards cannot be firmware-upgraded to 15118-20's mandatory mutual TLS 1.3 and larger certificate payloads because the jump exceeds what their existing cryptographic hardware was built to handle.

Sources

  1. CISA Advisory ICSA-25-303-01 / CVE-2025-12357: ISO 15118-2 SLAC protocol man-in-the-middle
  2. Brokenwire: Wireless Disruption of CCS Electric Vehicle Charging (NDSS 2023 paper, arXiv)
  3. CharIN: Plug & Charge PKI overview (first- and second-generation Certificate Policies)
  4. ISO 15118-20:2022 standard record (scope/abstract; full text paywalled)
  5. ISO 15118 (Wikipedia overview, secondary source used to corroborate parts structure, Plug & Charge history, and Autocharge/EIM comparison)
  6. SwRI: SwRI identifies security vulnerability in EV charging protocol (researcher's own description of the SLAC/CVE-2025-12357 finding)
  7. CISA CSAF record ICSA-25-303-01 (cisagov/CSAF on GitHub): summary, mitigation, revision history
  8. ISO 15118-3:2015 preview (iTeh): SLAC and the matching process
  9. ISO 15118-20:2022 preview (iTeh): scope, normative references, contents
  10. EVerest: Evse15118D20 module manifest (ISO 15118-20 TLS 1.3 mutual TLS, SupportedAppProtocol)
  11. CharIN: Certificate Policy for the CharIN V2G second-generation PKI, v1.0 (2022-11-23)
  12. Hubject: PKI downloads (Certificate Policies for its ISO 15118-2 and ISO 15118-20 V2G PKIs)
  13. Open Charge Alliance: New in OCPP 2.0.1 (white paper)
  14. Open Charge Alliance: Using ISO 15118 Plug & Charge with OCPP 1.6 (application note)
  15. Brokenwire: project site and FAQ
  16. Szakály, Köhler, Martinovic: Current Affairs, a security measurement study of CCS EV charging deployments (arXiv 2404.06635)
  17. Szakály, Strohmeier, Martinovic, Köhler: DCeption, real-world wireless man-in-the-middle attacks against CCS EV charging (arXiv 2601.15515)
  18. Löw, Vasu, Hutzelmann, Hof: Charge It to My Neighbor, a relay attack on ISO 15118 Plug and Charge payment (arXiv 2512.15966, USENIX VehicleSec 2026)
Was this guide helpful?

In this cluster