Jan. 20, 2009 –Vivek Khandelwal, the Director of Marketing at Verayo, explains how they create chips for use in RFID tags that protect against the cloning which the popular press keeps warning us about.

Vario has introduced chips that can make RFID tags unclonable. Would you tell us what about those chips makes that so?

KHANDELWAL: Vario has introduced a new silicone technology that’s called Physical Unclonable Function. It has been well known in semiconductor circles that all chips are different. No two chips are exactly alike, even though they may have the same design and mask; the chips are slightly different.

These are physical characteristics, correct?

KHANDELWAL: Yes; even though you have the same design physically the chips are slightly different. These differences are interjected because of the variations in the manufacturing process. Physical Unclonable Function takes these differences and digitizes them into unique characteristics so you can use them to authenticate a chip.

What we’ve introduced are RFID chips that are using this particular phenomena as a way to authenticate, for various applications like ID cards, pharma product or other kinds of products.

These would be chips that you would then supply to a tag manufacturer?

KHANDELWAL: Exactly.

Does it matter whether it’s HF or UHF?

KHANDELWAL: The first product that we’ve introduced into the market is HF. This is a silicone technology, so it can go into any other kind of RF chip also.

When you supply the chips, have you pre-measured the PUF factor?

KHANDELWAL: Let me walk you through it. I mentioned there are variations and these variations are different from chip to chip. It’s very difficult to clone variations from one chip to the next. That’s what makes our chips unclonable.

The way you can operate – the way the interface for a programmer or solution developer – is there is something called "challenge and response." A challenge is the input to a PUF circuit and a response is what comes out of the PUF. These challenge/response pairs are unique from chip to chip. So you give the same challenge to millions of other chips and you will get different responses from each. That’s what you use for authentication.

Now coming to your specific question, do we extract these challenge/response pairs? We sure can. We can extract challenge/response pairs from the chip at the wafer itself and provide them along with the chip. Think of it as chips coming with a database that provides authentication information about those chips.

Otherwise, if you don’t do that, do you create the chips in such a manner that it’s easy for your buyer to do it, or do you instruct on how to do it?

KHANDELWAL: The challenge/response mechanism is provided as an interface for anyone to collect challenge/response pairs.

Even if somebody is not interested in extracting challenge/response pairs from the chip, there are API demands that the chip will be able to handle as providing the responses. So you can even do it in the field on your own.

Is there some silicone or some types of chips that are less conducive to being used in this manner than others?

KHANDELWAL: No, there isn’t anything like that. Any kind of semiconductor chip can have this circuitry.

Then for the challenges and responses, that’s how up and down the line of authentication that people are able to tell that it’s the actual product they were hoping it was?

KHANDELWAL: Right. So these challenges are basically any value number that you can pick and you give them as an input to the PUF circuit.

In the case of RFID, you transmit that number, that challenge input to the PUF using RF and that’s where we are using HF today. The chip creates a response and again uses RF to get that number back to us.

That’s the response we use for subsequent authentication. So, you connect a bunch of these challenge/response pairs from every chip and use them for authentication.

Now, are they stored in the master database; is that how you compare?

KHANDELWAL: Yes; you put them in a back-end database of some sort.

How is the challenge in the field issued, let’s say at a retail store or something like that?

KHANDELWAL: The way we are suggesting a solution should be created here is that you have collected a bunch of challenge/response pairs from a chip at an initial location, wherever this product was manufactured. Then at the retail point of sale you scan this product and it will say, I am RFID ID number so and so.

You compare that to a back-end database serving you a challenge that it has collected from this chip before. The chip generates a response right there and then. You send that response and the back-end will say yes or no; whether it is the intended chip or not.

How does the security level differ from a CRC, a cyclic redundancy count, that I don't know if all chips are using, but some of the tags are using now?

KHANDELWAL: This is a very different approach to security. What those CRC and all those numbers do is eventually tell you whether data is changed. Here what we are saying is, is it the same chip or not?

Think of it as you can have two chips with the same data and same CRC, but they can be clones. If they are on two drug bottles, you don’t now which bottle is the real one. What we are doing is being able to test that, "This chip was the one I put on in my manufacturing facility and the one I recognize, and the other one: I don't know what it is."