28-06-2018 | | By Rob Coppinger
How long it takes for a memory microchip to delete data is to be used to detect fake or failing flash memory with a smartphone app expected next year.
The electronics industry faces two challenges with components, the infiltration of counterfeit parts into the supply chain, which will render a device useless, and recycled parts that are later found to be too damaged to work effectively.
Researchers at the University in Alabama in Huntsville have found that measuring the time it takes for flash memory to delete data is an effective way of determining that the component is genuine and in good order. “We found that this delete time is a strong function of age of the memory. If the memory is heavily used then the erase time becomes much longer, but the read and write [time] they are not much affected,” said assistant professor Biswajit Ray, the research team’s leader in the University’s electrical and computer engineering department.
The smartphone app that Ray expects to be released next year will allow people to check if their phone’s memory is new or has been recycled. Industry could also use a version of the app for checking its component supply. Ray said that while they expect to release the app next year, “we need to do a lot of testing because we can’t say that good memory is bad.”
The amount of time it takes to delete data can be used because the flash memory manufacturers produce datasheets and that erase duration is one of the characteristics they publish. “We are using the datasheet for a particular chip and then we are measuring the timing of that part of the chip and from that we conclude whether it is recyclable or not,” Ray added.
Measuring how long it takes to delete data is also a way of evaluating other types of memory, not just flash memory. Ray expects that the timing method can be applied to other types of memory chip. A challenge for Ray’s team is being able to detect problematic flash memory when it has not been used very often. “If you have used your memory 100 times, writing and deleting, it is easier to detect, but if you have used the memory ten times, for example, then it is not easy.”
Another approach to detecting flawed or fraudulent chips is what Ray describes as, “hammering”. This is the repeated writing and deleting of data. “Another strategy is the hammering technique. You write some data and you hammer it so there are some failures. Bad memory will show failures much quicker so in that way we can distinguish between used memory and fresh memory,” Ray explained.
Other methods have been proposed for monitoring the quality of components, but they can be costly and time consuming, for example, databases. Databases recording the characteristics of each and everyone of the millions of microchips the industry produces each year would be used to compare the performance of recycled chips while detecting fraudulent components. Ray said: “That means a huge database, hardly any company does that.”