Part of the  

Solid State Technology

  and   

The Confab

  Network

About  |  Contact

Headlines

Headlines

Hardware/Software Co-Development for IoT Applications: Q&A with Marie Semeria, CEO of CEA-Leti

By Ed Korczynski, Sr. Technical Editor

Marie Semeria, chief executive officer of CEA-Leti (http://www.leti.fr/en), sat down with SemiMD during SEMICON West to discuss how the French R&D and pilot manufacturing campus—located at the foot of the beautiful French alps near Grenoble—is expanding the scope of it’s activities to develop systems solutions for the Internet-of-Things (IoT).

Korczynski: You were talking about the need for new R&D into software in addition to the continuing R&D into hardware. Is this a little bit of a culture shift for Leti?

Semeria: The strategy of Leti is two-fold. One is to push scaling along to meet the challenge of HPC and the cloud infrastructure, so it’s a continued technological push along a roadmap. The other part is the IoT challenge where we need to handle everything from the sensors up to the service. In recent years Leti aggregated some skills in embedded software to develop IP around sensor devices, around MEMS, around NEMS, and around imaging-sensors. For IoT we need to have a global view of the usage case, and then come back to the specification of the technology. We need to collaborate with other institutes to have both hardware and software.

In data security we’ve started to embed the security into the hardware, and in the first layer of the embedded software (Figure). To have data security it makes no sense to only handle it with hardware, actually. If we want to have a secure system then we need to put the security at different levels:  firmware, hardware, software. We collaborate with CEA-List (http://www-list.cea.fr/en) to handle the upper layers of software, so that combined we cover everything from hardware up to system software.

Korczynski: For many years people have been talking about how the value-add in MEMS is in the algorithms not merely in the hardware. Also, with the vision for ubiquitous IoT and billions of sensors in the world, there’s not enough bandwidth to send all of the raw data around. So we have to be able to add some ‘intelligence’ into the devices so that we can transmit ‘information’ instead of mere data.

Semeria: Yes, but I believe that the differentiators will continue to be in the hardware, so we will not change our mission of providing differentiation through hardware. If we have to think about new sensors we need several years to be able to make them functional with very small form-factors and low power consumption. However, we do have to capitalize on the data, so just to have the loop of control on the hardware sensor itself we need to work with software.

Korczynski: Leti has historically had very strong focus on silicon wafers as a technology platform. A lot of people today discuss working with thin-film sensors which could be made with purely-additive technologies, so how much of Leti’s continuing R&D focus is on silicon?

Semeria: We will continue to use silicon to develop IoT applications, and to find new materials and devices which are compatible with CMOS. Doing so allow use of existing 200mm and 300mm silicon wafer fab facilities, representing a huge opportunity in the United States and Europe. Maintaining a core-competency on silicon is the best way to lower the cost for IoT, for automotive, for healthcare, for banking, for food and for other mass-markets. All of the solutions are coming from the co-integration of different functions on the silicon digital base. 3D-integration is key, bonding is key, and then you can start to imagine ways of developing new materials for sensors.

Korczynski: Certainly SOI provides an attractive foundation for co-integration of very different technologies.

Semeria: It’s a very attractive platform for IoT because it’s low-power. We are pushing FD-SOI down to 7nm, improving efficiencies using ‘boosters’ such as strain and silicon-germanium while keeping the advantages of lower energy consumption compared to finFETs. The idea is to use FD-SOI 28nm or 22nm as a generic digital platform, and to aggregate more and more functions compatible with silicon including MEMS, RF, and embedded-memory.

MEMS/NEMS solutions for IoT applications require the integration embedded software with the hardware sensors. (Source: CEA-Leti)

—E.K.



Tags:

Leave a Reply