Resistive RAM (ReRAM) technology has been on its way to its second place in solar for several years now. Despite its placement on the roadmaps of many chip manufacturers and as part of novel architectures such as HPE’s closed memristor-based “The Machine” challenge, it is by no means a revolutionary reminiscence technology. I did not appear. The idea makes wonderful sense.
Manufacturability, price and other factors worked, but for firms like Crossbar, which bet a full-fledged venture on Reram’s growing work, the gamble has paid off. The same firm immediately introduced a remarkable new life to ReRAM technology—in security.
Crossbar, Inc. states that ReRAM technology can be used in the creation of Physically Unclonable Operate (PUF) cryptographic keys because each secret is unique to the individual IC.
The concept has some support, with Dr. Bertrand Cambou of North Arizona College’s Nanotechnology and Cyber Security Group. He says that after evaluating various PUF approaches, there are benefits to ReRAM. “Due to its distinctive stochastic and electrical characteristics, the ReRAM PUF of the crossbar allows for safer programs than existing PUF applied science.” Compared to ReRAM, Kambo discusses these different approaches in detail. This is a useful description of how PUFs are properly matched to ReRAM.
“The method of generating PUFs on ReRAM arrays is promising mainly based on statistical evaluation of resistive gadgets,” says Kambou. He provides, “Resistive RAM is a sexy reminiscent technology for designing secure functions, PUF and RNG. It is low energy, faster, compact, and less vulnerable to channel attack than flash memory.”
Frankly speaking, the role and safety of ReRAM technology in PUF research is not a new idea. One of the idea’s many first appearances came out of Panasonic’s semiconductor group in 2016 using a 40nm ReRAM test chip to test the idea (with reasonable success). Despite the analysis from 2016 to now, it remains a bit fringe in terms of usage examples we could find and, like ReRAM, it has all the appropriate parts, but doesn’t have the appropriate “this problem”. broad market.
It must be annoying as hell for the crossbar. The company finds itself in another tough market after RERAM’s memory innovation does not materialize as a major factor in the following. There is already a lot of applied science on the PUF matter (all of which use SRAM, at best, which is reasonable). Compared to SRAM-based PUF approaches, their technology allows for the next diploma of randomness, reduced bit error charges, and all this “without the need for fuzzy extractors, auxiliary information, or bulky error correction codes”. Davis, the crossbar president.
When it comes to ReRAM, it’s worth taking a more in-depth look at (we’re chasing one) why ReRAM never got off the bottom. It will probably be less than 10nm, has excellent flash-outspacing learning latency and can write significantly faster than NAND. It doesn’t require any tremendous fancy manufacturing magic. So why did the market surpass Reram? Dive deeper on that coming.
Crossbar says that now for its business, it’s becoming available in multiple CMO foundry courses of nodes to achieve greater density and more tightly manufactured devices. Corporate survives, at best, by licensing SoC and memory firms each off-the-shelf and customized IP cores.
It first hit the market in 2010, early in the next generation recall and while the angle of security functions would likely help them keep up with the ReRAM struggle, it’s hard to tell what the way forward for the technology generally looks like. Like in the coming years.