Case · DiggingBeagle record

PowerHammer carried covert data through electrical power lines

PowerHammer research controlled CPU utilization to modulate system power consumption, creating conducted emissions that could be measured from electrical wiring and decoded as data.

Scope

Laboratory attack model requiring compromised code and attacker access to electrical wiring or a relevant measurement point.

Assessment profile

Assessment method
Setting
research demonstration
Exploitation
demonstrated
Evidence
Grade B
Basis and provenance

The CPU-current modulation and line/phase tap results describe a laboratory channel that presupposes both host compromise and receiver access. Assessment is a desk review of the retained canonical Claims and cited Source metadata at their recorded cutoff, not a new external verification. Impact remains ungraded.

Assessed Sep 24, 2026 using diggingbeagle.assessment/1.

At a glance

Mechanism and trust boundary

  1. 01

    Execute code on the isolated computer

    The transmitter is malware already running on the protected system.

    Boundary: air-gapped host / attacker-controlled code

  2. 02

    Encode data in CPU utilization

    The malware raises and lowers processor load according to the bitstream.

    Boundary: secret data / compute workload

  3. 03

    Convert workload into current fluctuations

    Changes in processor demand alter system power consumption and therefore current on the supply line.

    Boundary: software activity / electrical current

  4. 04

    Conduct the signal through power wiring

    The data-dependent current pattern propagates along the connected electrical infrastructure.

    Boundary: computer power input / building wiring

  5. 05

    Measure and decode at a tap point

    A prepared line-level or phase-level sensor records the fluctuations and reconstructs the transmitted data.

    Boundary: power infrastructure / attacker receiver

Timeline

  1. Apr 10, 2018
    Event type unspecified

    PowerHammer preprint released

    The paper described line-level and phase-level power-line exfiltration and reported their tested bit rates.

Claims & evidence

CLM-POWERHAMMER-RATESThe paper reports about 1000 bit/s for a line-level tap and about 10 bit/s for a phase-level measurement in the tested configurations.supported

Basis: reported finding

Link to claim
CLM-POWERHAMMER-MECHANISMPowerHammer demonstrated exfiltration by modulating CPU utilization so that data appeared as controlled fluctuations in current on connected power lines.supported

Implications

PowerHammer shows that an air gap leaves conductive infrastructure intact. Its operational risk is bounded by the attacker's ability to measure the relevant electrical path and by signal quality in the facility. For high-assurance design, power feeds and panels are therefore part of the containment boundary rather than neutral utilities outside it.

Controls and mitigations

  • Restrict physical access to line-level wiring, distribution panels and other useful current-measurement points.
  • Use power filtering or isolation appropriate to the protected environment and verify effectiveness against the frequencies of concern.
  • Monitor abnormal current signatures or patterned CPU-load behavior only where normal workload variability permits meaningful detection.

Unknowns and contradictions

  • The paper's reported rates come from specific line-level and phase-level measurement setups and should not be generalized to arbitrary buildings.
  • Electrical noise, power-supply design, wiring topology and measurement placement materially affect the channel.
  • The channel assumes prior host compromise and a prepared receiver on the power infrastructure.

Sources and citation

Material revision history

  1. Sep 19, 2026 · Published version · first publication · revision 15
  2. Sep 19, 2026 · Published version · material revision · revision 20Public fields changed: attack_path, implications, mitigations, narrative, timeline, unknowns
  3. Sep 25, 2026 · Published version · material revision · revision 36Public fields changed: assessment_profile