Defeating the smartcard code-crackers

Some two million new smartcards are rolled out every month, so the encrypted personal data that people store on these cards must be safe. Yet the security threat is growing, as the electronic devices capable of breaking the card codes become cheaper and more powerful.

“It takes little more than an oscilloscope and a standard PC to mount a digital attack on an unprotected smartcard,” says Klaus-Michael Koch. He is coordinator of the IST project SCARD, which aims to increase the security of chips on smart cards.

With equipment like this and some know-how, attackers can expose the content that a smart card is supposed to protect. Using techniques such as side-channel analysis (SCA), they can reveal part of a secret key, notably by examining a chip’s power leakage as it performs computations or by scrutinising its thermal or electromagnetic radiation. If the card’s owner is the attacker, he or she could upload money to an electronic purse, access a satellite TV system for free or claim to be someone else.

Under SCARD, the partners put together a ‘design flow’ that allows semi-automatic implementation of countermeasures. The design flow is the digital design of a chip – the specifications, modelling of performance, algorithms and functionality up until the stage when the chip developer can start the synthesizer and compiler. Typically, this design process is costly and may take several years.

In-chip countermeasures must be included during the design period. They cannot be simulated, so developers must experiment with the shielding of a card’s chip to limit temperature and voltage variations, or they must laboriously place transistors on it by hand.

For the hardware security issue, the partners developed prototypes of a design flow and carried out chip testing. They also paved the way for an automatic chip design process which would allow other companies to develop new and more secure chips.

“We succeeded in making the hardware more secure against side-channel analysis (SCA),” says Koch. “The chip we built was used to deduce the measurability limits, enabling us to assess the sort of countermeasures necessary against differential power attacks.”

To tackle leaky circuits, the SCARD partners developed two main countermeasures. The first introduces circuits with constant power consumption, irrespective of the tasks being performed. Says Koch, “Each clock cycle has the same energy. But these circuits must be perfectly executed, since even a three or four percent difference in energy can be seen.” The second involves adding random values to the chip, masking the circuit’s real values. Noise could also be added, though this is not currently feasible in smartcards due to energy-loss restrictions.

They have also developed an eight-bit test chip, featuring both unprotected and protected versions of the same circuit. The chip includes a microcontroller, is fully programmable and has reduced leakage. It is also capable of resisting over 500,000 attempted measurements, as opposed to the 15,000-measurement threshold for an unprotected chip. As a result, researchers can for the first time directly compare the effect of certain countermeasures on unprotected or protected versions of the same circuit.

“Our new chip is not one hundred percent secure,” acknowledges Koch. “However, it is far more difficult to crack than existing unprotected versions and represents a quantum leap forward in security.”

The new chip was produced using the project’s own design flow, taking just one year from specification to production. “We demonstrated that our chip design flow – our set of tools and methods – really works,” he notes.

Two partners, Institut für Angewandte Informationsverarbeitung und Kommunikationstechnologie (IAIK, Austria) and Infineon, have applied for international patents stemming from their project work. These include countermeasures with new secure logic styles that cover innovative transistor circuits. Some of the countermeasure technology developed is also being used in IAIK’s security crypto-modules.

The project results are now being disseminated through teaching – since some of the project partners are universities or technical small and medium-sized enterprises.

Source: Based on information from SCARD