September 2, 2011
New NAND Flash Test Can Identify Counterfeit Devices
Researchers at the University of California at San Diego and Cornell University are developing software they say can test unique variations in flash behavior and identify counterfeit products.
Scholars are working on software to test the unique "fingerprints" of the chips, which many industry insiders believe will be a useful tool in the near future. By running the test in the factory and then again furthur down the supply chain, companies can compare results to verify the authenticity of the components. Counterfeit chips are currently a problem for vendors who often have to fill unexpected surges in demand by buying lots of the devices on the open market. There is an ever growing number of third parties that buy and sell these chips, and verifying parts is one of the most time consuming and expensive aspects of their business.
The researchers at UCSD believe that there are other possible uses for the technology, as well. They suggest that using their techniques, they can prevent the counterfeiting of devices that contain flash, such as cell phones and tablets. Because they would be testing the flash silicon instead of the entire device, no hardware changes are required, a major advantage over techniques avaiable today.
The system uses "physically unclonable functions" or PUFs to identify unique "flaws" created by variations in manufactoring. The researchers at UCSD explain, that data is written and erased from NAND flash by sending voltrage through, and thereby changing the state of, individual cells. Should the same cell be rewritten many times in a row, the voltage can bleed into adjacent cells, causing it to change it's state as well. In normal operation, the order in which cells are modified prevents this type of unintentional state change. In the test, however, the cells are disturbed intentionally and the number of repititions it takes to affect a neighboring cell is recorded. The number of repititions depends on many factors that vary from one chip to another, such as the thickness of the barriers between cells. Since the number of repititions needed is usually in the thousands, it can be used rather reliably to compare the devices, even at different times and locations along the supply line.
There is certainly value in this technique and it is likely to save NAND flash distributers much time and money that would normally be used to verify hardware. Industry insiders predict that, should demand for this test become widespread, it will likely be coming from those in the flash industry, rather than from the end-users of the flash devices.
Scholars are working on software to test the unique "fingerprints" of the chips, which many industry insiders believe will be a useful tool in the near future. By running the test in the factory and then again furthur down the supply chain, companies can compare results to verify the authenticity of the components. Counterfeit chips are currently a problem for vendors who often have to fill unexpected surges in demand by buying lots of the devices on the open market. There is an ever growing number of third parties that buy and sell these chips, and verifying parts is one of the most time consuming and expensive aspects of their business.
The researchers at UCSD believe that there are other possible uses for the technology, as well. They suggest that using their techniques, they can prevent the counterfeiting of devices that contain flash, such as cell phones and tablets. Because they would be testing the flash silicon instead of the entire device, no hardware changes are required, a major advantage over techniques avaiable today.
The system uses "physically unclonable functions" or PUFs to identify unique "flaws" created by variations in manufactoring. The researchers at UCSD explain, that data is written and erased from NAND flash by sending voltrage through, and thereby changing the state of, individual cells. Should the same cell be rewritten many times in a row, the voltage can bleed into adjacent cells, causing it to change it's state as well. In normal operation, the order in which cells are modified prevents this type of unintentional state change. In the test, however, the cells are disturbed intentionally and the number of repititions it takes to affect a neighboring cell is recorded. The number of repititions depends on many factors that vary from one chip to another, such as the thickness of the barriers between cells. Since the number of repititions needed is usually in the thousands, it can be used rather reliably to compare the devices, even at different times and locations along the supply line.
There is certainly value in this technique and it is likely to save NAND flash distributers much time and money that would normally be used to verify hardware. Industry insiders predict that, should demand for this test become widespread, it will likely be coming from those in the flash industry, rather than from the end-users of the flash devices.
