Same threats, different technology
Potential threats to radio frequency identification, or RFID, systems resemble the problems currently being encountered by IT network administrators, according to experts who presented at network security conferences held last month in Las Vegas.
“RFID has been called the Internet of things,” said Melanie Reiback, a researcher at Vrije Universiteit in Amsterdam. “The Internet of things has the same problems as the Internet of today. It’s susceptible to the same types of attacks.”
However, some independent analysts monitoring RFID security believe that such threats are overstated. “All security systems have their vulnerabilities. I don’t think this is anything specific to RFID,” said Sara Shah from ABI Research. “I didn’t find it be as earth-shattering as was portrayed in the media a few months ago.”
Threats and abuses of RFID technology include unauthorized tag reading, eavesdropping on transmitted data, tag duplication or “cloning,” denial-of-service attacks and the ability to insert viruses and other harmful software — or malware — into enterprise networks.
“RFID technology extends your network perimeter,” Reiback said. “People need to understand that.”
Reiback presented a research paper at an IEEE conference held last spring that outlined the potential for viruses to be embedded in RFID tags. She and her research team built a modular test setup that simulated a typical supply-chain application using an RFID reader and tags, middleware and four different databases.
A simple first-generation passive RFID tag can hold up to 128 characters of information, with newer tags able to store up to 256 characters. That capacity might not sound like much, but it’s enough to let a creative programmer store non-standard characters designed to generate system faults when processed and then follow that with concise programming commands in languages such as SQL or XML.
Reiback tested a method to use corrupted tags on an improperly secured system to inject commands into a back-end SQL database — including the ability to query a database for a piece of information and then write it back to the tag.
“Admittedly, this requires a little bit of insider knowledge,” she said. “But you can write out data to the tag, and you don’t even have to be there.” Her favorite discovery is the ability to conduct a denial-of-service attack by using a corrupted tag to issue a shutdown command to the database. “It’s not fatal; but if it goes from reader to reader, it can be a mess.”
While her paper caused quite a stir, some are less impressed with Reiback’s work, feeling it lacks real-world practicality. “You’re working with [RFID chip] storage, and she’s correct there, but it’s nothing new,” said Lukas Grunwald, a computer security specialist with more than 20 years of experience. “We did it two years ago with RFDump [software]. She built a vulnerable system, not demonstrated how this could be done on a real [production] system.”
He said Reiback’s test didn’t incorporate any standardized security practices or state-of-the-art database security solutions that would be more representative of a real-world system.
Grunwald practices what he preaches, sometimes with disastrous results. He created an RFID tag that crashed the access control system at his office, leaving all the doors unlocked without triggering any alarms. Then neither he nor the service company that installed the system could revive it. His malicious tag had inadvertently erased the underlying microprocessor control software.
At the BlackHat USA 2006 security conference in Las Vegas, Grunwald quickly demonstrated how easy it is to read the information contained on his German passport’s embedded RFID chip using a laptop, then modifying it and writing it out to another chip, a process that took less than 2 minutes. While it took two weeks of research time — “billable at $15,000” — to figure out the process, the open source software can be downloaded free of charge from the Internet. A blank chip costs $20 in small quantities, and a USB-compatible RFID-tag writer costs less than $200.
International standards for incorporating RFID technology into passports have been established by the International Civil Aviation Organization (IACO). According to ABI Research’s Shah, the IACO standards are basic, but the threat of a duplicated e-passport is not significant.
“Let’s say you clone my passport,” Shah said. “You take all of my information in my chip. All that’s going to have is my information on the front page of my passport, my pictures, some biometrics. What are you going to do with that? When you go to an airport, you’re not me, it’s not your picture and you don’t have the same fingerprints that I do.”
Furthermore, both Shah and Grunwald agreed that should the information in a cloned passport be changed, the new information wouldn’t match the unique digital “hash” signature generated when the original information is written out. That would raise a red flag to border control agents inspecting the e-passport.
The smart chip on an e-passport can store up to 72 kilobytes of information, offering ample capacity for storing non-standard information, such as malicious code. One possibility Grunwald suggested concerned enabling an e-passport to also work with an RFID access control system, which could allow someone to enter a secured area without the proper authorization.
Both Grunwald and Reiback are working on new tools to protect RFID technology. Grunwald is developing a laptop-based simulation capable of emulating either an RFID tag or reader. It would be used to test back-end databases and middleware, allowing someone to cycle through many different types of bad tags to see what would happen.
At DEFCON, the annual informal hacker’s conference that followed BlackHat, Reiback introduced a mobile hardware device designed for “personal RFID privacy management.” It is a two-way RFID communicator with a broadcast range of about 1 meter.
Wardrivers blast off
Discovering open Wi-Fi hotspots has been a hobby of the hacker community for years. Members use car-mounted laptop computers for “wardriving” around neighborhoods to scan for open access points. Intrepid souls quickly moved their setups to bikes, boats, planes and helicopters as they sought new ways to rapidly cover territory.
Amateur rocketeer Rick Hill has managed to trump all other wardrivers to date. Hill, a senior scientist at a computer security firm, combined his hobby with his day job by installing an 802.11b Wi-Fi scanning payload into a rocket.
“I first got the idea through trying to set up my own wireless Internet service provider business,” Hill said. “There are lots of trees and a forest canopy where I live.” A rocket provides a virtual tower in the sky above trees and terrain clutter to look for Wi-Fi emissions.
His first setup used a one-third-scale model of a 1950’s-era NASA sounding rocket capable of reaching a height exceeding one mile. With a thrust of 95 pounds and a cost of $200 per solid motor, the rocket is a far cry from the small backyard-style hobby models with which most people are familiar. It is so powerful that launching one requires FAA clearance and a remote launch site so it doesn’t inadvertently hit an airplane on the way up.
The Wi-Fi scanning payload weighs less than 3 pounds and includes an HP iPAQ PDA, a stripped-down Deliberant 2300 access point, a cluster of 9 V batteries to power the access point and a circularly polarized antenna designed to fit in the 12×5.5-inch payload bay. The PDA automatically records Wi-Fi data and broadcasts in real-time from the access point to a ground station for redundancy.
Hill noted that payloads encounter forces up to 10 times that of gravity during launch, and amateur rocketry can be fraught with mishaps. “If it can fail on a rocket, it will,” he said.
But once the rocket is launched and reaches its peak altitude, a pair of explosive charges fire to separate the nose cone and deploy the parachute. Total flight time is about 6 minutes, with about 5 minutes of useful scanning time from parachute deployment until the rocket descends to under 500 feet.
Hill has conducted a total of three launches in rural areas of Maryland and Virginia, two with the Nike Smoke rocket and one with a smaller rocket that rises to 2000 feet and doesn’t require FAA launch approval. Two of the launches successfully picked up wireless access points in a 50-square-mile area around the launch site that weren’t detectable by ground-based wardriving gear. The third launch, conducted near the University of Virginia in Charlottesville, rose to about 200 feet, then nosed over and flew parallel to the ground for about 1 mile before landing. It picked up seven Wi-Fi access points, but Hill had hoped to map the UVA campus and surrounding area if his flight had gone straight up, as planned.
— Doug Mohney