Who is Charging My Phone? Identifying Wireless Chargers via Fingerprinting

Jiayu Zhang, Zhiyun Wang, Xiaoyu Ji, Wenyuan Xu, Gang Qu, Minjian Zhao

TL;DR

This paper introduces WirelessID, a novel system that uses the unique electrical "fingerprint" of wireless chargers to detect malicious charging devices. By analyzing subtle non-linear distortions in the charging circuit's signal, WirelessID can accurately identify untrusted chargers, enhancing the security of wireless charging in public spaces.

What Problem Does This Paper Address?

The increasing popularity of wireless charging, especially in public IoT scenarios like airports and coffee shops, brings convenience but also introduces new security vulnerabilities. Unlike wired charging, wireless charging requires close proximity, making devices susceptible to short-range attacks. Attackers could potentially disguise malicious chargers to intercept data, damage batteries, or even exploit the charging process for other nefarious purposes. Existing security measures often rely on cryptography or hardware modifications that are not practical for widespread deployment or retrofitting into existing devices. This paper addresses the need for a practical, non-intrusive method to authenticate wireless chargers in public environments.

Methods

The core of WirelessID lies in hardware fingerprinting based on the non-linear distortion effects of inverters within wireless chargers. Here's a breakdown of their approach:

  • Fingerprint Generation: The system exploits the fact that even identical MOSFETs (the key components in inverters) exhibit unique non-linear distortion characteristics due to manufacturing variations. These subtle distortions in the electrical signal during the charging process create a unique "fingerprint" for each charger.
  • Signal Collection: Before a charging session begins, WirelessID collects signals from the receiver coil in the smartphone during the charger's standby stage. This stage is chosen to avoid risks associated with an active connection.
  • Signal Processing & Feature Extraction: The collected signals are pre-processed (filtered and normalized) to remove noise. Then, 40 scalar features are extracted from both the time and frequency domains. A feature ranking tool (FEAST) is used to identify the most discriminative features, forming the charger's fingerprint.
  • Verification: A trusted cloud server hosts the verification module. It uses a supervised learning approach, specifically the Extra Trees algorithm, to classify the extracted fingerprint. The system trains a binary classifier for each trusted charger. When a new fingerprint is received, it's compared against the database of known trusted chargers. If a match is found, the charger is authenticated; otherwise, it's flagged as untrusted.
  • Training: For training, a small number of traces (k) from a specific device are used as positive examples, while traces from all other devices are used as negative examples to train a binary classifier for that device.

Datasets

The authors created their own dataset for evaluation:

  • Wireless Charger Traces: They collected 8000 signal traces from 8 commercial wireless chargers over a 5-month period. Each trace lasted 50ms and was sampled at 500 kHz.
  • Training/Testing Split: 10% of the collected traces were used for training, and the remaining 90% were used for testing.

Key Insights and Ideas

  • Intrinsic Hardware Fingerprinting: The most significant insight is that subtle, inherent manufacturing variations in electronic components (like MOSFETs) can be reliably exploited to create unique, tamper-resistant fingerprints for devices. This bypasses the need for cryptographic keys or software modifications.
  • Standby Signal for Security: Utilizing the standby signal stage for fingerprint extraction is a clever way to gather security-relevant data before a potentially malicious charging session begins, minimizing risk.
  • Non-linear Distortion as a Signature: The paper highlights that non-linear distortions, often considered undesirable noise, can actually serve as valuable discriminative features for device identification.
  • Practicality for Public Spaces: WirelessID offers a promising solution for securing public charging infrastructure without requiring users to carry special hardware or modify their devices beyond the standard wireless charging receiver.

Key Results

The WirelessID system demonstrated strong performance in their experiments:

  • Overall Performance: Achieved 99.0% precision, 98.9% recall, and 98.9% F1-score on their test dataset.
  • Classifier Performance: The Extra Trees algorithm was identified as a top performer among 10 common machine learning algorithms.
  • Sampling Rate Impact: The system's performance remained high (above 98.6%) across various sampling rates (500 kHz to 8 MHz), leading them to select 500 kHz to minimize overhead.
  • Time and Environment Stability: Fingerprints were found to be stable over time (5 months) and robust across different environments (office, lab, conference room), with performance exceeding 98%.
  • Scalability: The system showed a low miss rate (0.9%) when identifying "alien" devices not present in the training set.

Limitations and Future Work

The authors acknowledge several limitations and propose future directions:

  • Smartphone Integration: The current prototype uses a USRP for signal collection. Integrating this functionality directly into smartphones is challenging because current smartphone wireless charging modules cannot easily transmit received charging signals to the OS. This would require adding dedicated AD sampling chips.
  • Multiple Coils: Wireless chargers with multiple coils (e.g., surface chargers) present a challenge, as different coils sharing the same inverter might lead to similar fingerprints. The authors suggest training for all coil+inverter combinations.
  • Software Tampering: WirelessID is primarily focused on detecting hardware tampering. It may not recognize attacks that involve solely tampering with a legitimate charger's software.
  • Future Work: The authors plan to expand the number and types of tested devices, optimize classification models, improve generalization, and research attacks based on software tampering and reverse wireless charging.

Summary updated

Abstract

With the increasing popularity of the Internet-of-Things (IoT) devices, the demand for fast and convenient battery charging services grows rapidly. Wireless charging is a promising technology for such a purpose and its usage has become ubiquitous. However, the close distance between the charger and the device being charged not only makes proximity-based and near-field communication attacks possible but also introduces a new type of vulnerabilities. In this article, we propose to create fingerprints for wireless chargers based on the intrinsic nonlinear distortion effects of the underlying charging circuit. Using such fingerprints, we design the WirelessID system to detect potential short-range malicious wireless charging attacks. WirelessID collects signals in the standby state of the charging process and sends them to a trusted server, which can extract the fingerprint and then identify the charger. We conduct experiments on eight commercial chargers over a period of five months and collect 8000 traces of signal. We use 10% of the traces as the training data set and the rest for testing. The results show that on the standard performance metrics, we have achieved 99.0% precision, 98.9% recall, and 98.9% <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$F1$ </tex-math></inline-formula> -score.
BibTeX
@article{zhang2020who,
  author = {Jiayu Zhang and Zhiyun Wang and Xiaoyu Ji and Wenyuan Xu and Gang Qu and Minjian Zhao},
  title = {{Who is Charging My Phone? Identifying Wireless Chargers via Fingerprinting}},
  journal = {IEEE Internet of Things Journal},
  year = {2020},
  url = {https://openreview.net/forum?id=rdDSBYsmVB},
}
Figure