Rambus-Arm security deal; EV sales; TPM vulnerabilities; NIST hardware security report; startup funding; AI replace engineers?; solid-state battery research; GM’s hand-free driving; smart skis; Winbond & ST partner.
Rambus will begin selling Arm‘s CryptoCell embedded security platform and CryptoIsland root-of-trust cores, setting the stage for a much broader push by Rambus into security for a wide range of connected devices, and ultimately into security as a service. Under the terms of the deal, Rambus’ customers will be able to license Arm IP directly from Rambus. For Arm’s existing customers, there will be no change.
Security has become a much bigger deal for the IC industry, in part because chips are being used in more mission-critical and safety-critical applications, and in part because the value of the data is growing and there is more data connected to other data. The challenge now is to be able to keep systems secure throughout their lifetimes, which in the case of automotive can be a decade or two. “You’re buying security, not to make a device foolproof, but to make it harder to hack — and to know when somebody has hacked into it,” said Neeraj Paliwal, vice president and general manager of Rambus’ Security IP Business. “We are starting to put structures in our IP that will help measurability. You can’t just put a signature on software or hardware so you know something has changed. And whenever you see a potential data leak, or where secrets are stored, how do you put some envelope around that?”
The electrification of vehicles is a key market for this. Passenger electric vehicle sales totaled more than 10.2 million units worldwide in 2022, up 65% year-over-year, according to a new report from Counterpoint. In Q422, BEV’s accounted for nearly 72% of all EV sales, while plug-in hybrids made up the rest. [Note: other hybrid EVs and fuel cell vehicles were not included in the study.] Seven of the top 10 EV models were Chinese brands in Q422. Full-year 2023 sales are projected to climb to almost 17 million units, mostly driven by U.S. tax credits.
The cost of borrowing is going up, but investors continued to pour money into the chip industry in February. Collectively, 132 companies raised more than $4.5 billion last month. Notable investments include battery-related technologies, automotive, AI hardware, bio-inspired AI chips, RISC-V processors, wireless and more. (See the summary table at the end of the report.)
Will AI ever really replace engineers or will it be a tool for them? Four industry experts dove into this topic at a recent DesignCon panel.
U.S. investigators suspect a 2014 Tesla Model S was in automated driving mode when it crashed into a parked firetruck last month, killing the driver. A special team is investigating the crash. There’s a larger investigation underway, as well, into multiple incidents of Tesla vehicles with “full self-driving” systems that been crashed into parked emergency vehicles.
There’s been a lot of activity around MIPI in automotive. Here are the driving forces behind it.
GM claims its Ultra Cruise ADAS technology will enable hands-free driving in 95% of all driving scenarios, featuring a 360° view, blend of short/long range radars, lidar behind the windshield, sensor fusion, and Qualcomm SoCs.
Solid-state batteries are being touted as a game-changer for the EV industry, but affordability and long-range powering on a single charge has been a challenge. A Berkeley-led team’s research zeroed in on designing a solid electrolyte with a mix of affordable and more available metals, observing the multi-metal mix displayed an “ionic conductivity several orders of magnitude faster than the single-metal materials.”
Infineon and UMC cemented a long-term agreement to increase capacity of the production of Infineon automotive microcontrollers with eNVM technology, with manufacturing at UMC’s Singapore fab. Infineon also is officially supporting Rust, a relatively new open-sourced programming language touted as being memory-safe, for its automotive MCUs.
Renesas deployed Cadence‘s new Versium AI-Driven verification platform on its R-Car automotive SoCs to improve debug productivity.
In a three-part series, industry experts from Flex Logix, Quadric, Perceive and Syntiant discussed AI’s impact in automobiles, where and why AI makes sense in cars, and security, safety and reliability of AI in autos. The discussion was held at the recent DesignCon.
Quarkslab researchers found two buffer overflow vulnerabilities in the Trusted Platform Module 2.0 (TPM) reference library specification. “An authenticated, local attacker could send maliciously crafted commands to a vulnerable TPM, allowing access to sensitive data. In some cases, the attacker also can overwrite protected data in the TPM firmware. This may lead to a crash or arbitrary code execution within the TPM. Because the attacker’s payload runs within the TPM, it may be undetectable by other components of the target device.” This update was issued to address these vulnerabilities.
Confidential computing is now a hot topic and an emerging industry initiative focused on securing data in use, without exposing it to the rest of the system. Case in point: NIST has published a draft report, titled “Hardware Enabled Security: Hardware-Based Confidential Computing,” which presents an approach for managing machine identities for protection against malware and other security vulnerabilities. Comments are due April 10, 2023. Also, LF AI & Data Foundation Technical Advisory Council accepted Open Federated Learning (OpenFL) as an incubation project to drive collaboration, standardization and interoperability. Intel developed OpenFL to “help data scientists address the challenge of maintaining data privacy while bringing together insights from many disparate, confidential or regulated data sets.”
Intrigued, yet nervous about ChatGPT? Researchers at CISPA Helmholtz Center for Information Security found that “adoption and deployment of LLMs [Large Language Models] may initiate a new attack surface targeting application integrated LLMs and their users. LLMs’ functionalities were shown to be easily modulated via natural prompts, opening the door for possible adversarial exploitation via prompt injection,” states the report.
The European Union Agency for Cybersecurity (ENISA) just published two reports addressing the telecom cybersecurity challenges, one for eSIMS and another for fog and edge computing in 5G.
Infineon released a new AIROC CYW43022 ultra-low power, dual-band Wi-Fi 5 and Bluetooth combo, featuring up to 65% reduction in power usage during deep sleep for battery-life extension, WiFi network offloads, and embedded Bluetooth stack. That, in turn, reduces power demands on host processors.
Winbond and STMicroelectronics are combining Winbond’s memory ICs (DDR3 dynamic RAM) with ST’s MCUs and MPUs for applications such as industrial gateways, data concentrators, smart meters, barcode readers, smart-home devices, and numerous applications that require both high performance and state-of-the-art security.
More functions, greater security risks, and increasingly complicated integration of IP and various components below 7nm is increasing the time and effort it takes to get a functioning chip out the door. In many of these devices, the network on chip is the glue between various components, but it can take up to 10% to 12% of the total area of the SoC. This tech talk dives into how to shrink the NoC area, improve security, and reduce time to market.
Smart skiing has advanced with new technology from a three-partner consortium (CEA, Rossignol, and Lumiplan). Piezoelectric devices “transform the weight of skis in bends into electric power to fuel ultra-low power electronics and estimate a skier’s weight on skis.” The new design fuses data produced from skis with smartphone data and assesses how powerfully a skier leans into turns, turn frequencies and speed, all optimizing the use of ultra-low energy before sending it to a skier’s phone via Bluetooth.
MIT researchers built an augmented reality headset that can view hidden objects by combining computer vision, wireless perception, and RF signals that can pass through everyday materials such as cardboard, plastic, and wood to find hidden objects that have been labeled with RFID tags.
Samsung launched its Bespoke refrigerator, featuring an energy mode that uses AI to track usage patterns and power consumption.
Find all the upcoming chip industry events here, including:
Read the latest automotive, security, and pervasive computing articles, or check out the latest newsletter.
Leave a Reply