TL;DR
Crowdsourced Bluetooth trackers (like Tile or TrackR) help users find lost items, but their underlying architectures expose users to severe privacy and security risks, such as location spoofing and unauthorized tracking. This paper systematically analyzes five popular commercial trackers, finding critical design flaws in almost all of them, and proposes SECrow: a novel, secure tracking protocol that guarantees end-to-end encryption, physical ownership verification, and unspoofable location updates.
What Problem Does This Paper Address?
Bluetooth Low-Energy (BLE) trackers are incredibly popular for keeping tabs on keys, wallets, and luggage. Because these tiny devices lack GPS and internet connectivity, they rely on a crowdsourced location tracking system. If you lose your keys, any participating user's smartphone that walks past your tracker will detect its BLE beacon and silently upload its GPS coordinates to the vendor's cloud server.
While highly effective, this distributed model introduces massive security and privacy loopholes. Can an attacker hijack your tracker? Can they spoof its location to make you think your stolen bicycle is still in your garage? Can the vendor's server be breached, revealing the real-time locations of millions of valuables? The authors formalize the security properties required for a safe crowdsourced tracking system, evaluate existing commercial products against these properties, and design a system that actually meets them.
Methods
The researchers took a two-pronged approach: offensive analysis and defensive design.
- Security Analysis of Commercial Systems: The authors formalized eight Necessary Security Conditions (NSCs) across three entities: the Tracking Device (TD), the Communication Device/phone (CD), and the Tracking Service/server (TS). They then reverse-engineered and dynamically instrumented the Android apps of five popular trackers to intercept and manipulate API calls. This allowed them to test if they could register unowned devices, spoof locations, or trigger privileged commands (like making the tracker ring) without authorization.
- SECrow Protocol Design: To fix the broken ecosystem, the authors designed SECrow. It relies on public-key cryptography and hardware-backed security. Trackers are manufactured with burned-in private keys (using e-fuses). Smartphones use Trusted Execution Environments (TEEs, like ARM TrustZone) to cryptographically sign GPS coordinates, preventing OS-level location spoofing. Finally, location data is end-to-end encrypted so that the central server routes the data but cannot read it.
Datasets
Not discussed in the paper in the context of machine learning datasets. However, for their evaluation, the authors analyzed five commercial tracking ecosystems: TrackR, CUBE, Chipolo, Pebblebee, and Tile. They also built a functional prototype of SECrow using a Raspberry Pi 3B to simulate the BLE tracker and Python-based clients/servers.
Key Insights and Ideas
- MAC Address = Ownership: In several systems (TrackR, CUBE, Pebblebee), simply knowing a tracker's publicly broadcasted Bluetooth MAC address allows an attacker to register as its owner via the cloud API. No physical access to the tracker is required.
- Blind Trust in the Cloud: Most commercial servers blindly trust location updates from any phone. An attacker can easily write a script to spoof the location of a victim's tracker to the other side of the world, completely breaking the reliability of the service.
- Tile is Better, but Not Perfect: Among the tested devices, Tile was the only one that successfully enforced physical ownership (requiring a physical button press to register). However, it still failed to protect against location spoofing and lacked anonymous location updates.
- End-to-End Encryption is Practical: By leveraging TEEs on modern smartphones and basic crypto capabilities on BLE tags, it is entirely possible to build a system where the server acts blindly. The server can verify that a location update is authentic without actually knowing the coordinates.
Key Results
- Widespread Vulnerabilities: 4 out of 5 tested systems failed the "Physical Ownership" property. 0 out of 5 systems guaranteed "Anonymous and proximity-aware location updates" or a "Reliable location service."
- SECrow Performance: The authors' secure prototype proved highly practical. Cryptographic operations on the simulated tracker took only ~60 milliseconds.
- End-to-End Latency: A full, secure location update in SECrow took about 8.6 seconds (compared to 0.03 seconds for the insecure TrackR baseline). Because commercial apps typically only poll for location updates every 5 minutes, this overhead is negligible for real-world use.
Limitations and Future Work
The authors acknowledge a few limitations in their proposed SECrow system:
- Hardware Requirements: SECrow requires trackers to have secure storage (e-fuses) and cryptographic co-processors. While cheap, this is a step up from the absolute bare-minimum hardware used in some current trackers.
- Relay Attacks: Like many proximity-based systems, SECrow assumes BLE communication implies physical proximity. It remains vulnerable to relay attacks (where an attacker forwards BLE packets over the internet), which are notoriously difficult to prevent on low-power devices.
- Battery Drain DoS: A malicious smartphone could repeatedly spam a nearby tracker with cryptographic challenge requests, draining its small battery. The authors suggest implementing rate-limiting on the trackers as a future mitigation.
Summary updated