Advancing 60 GHz Contactless Communications Using HPC Technologies
Complete the form below to unlock access to ALL audio articles.
Whether it’s charging a phone on a wireless docking station or playing a mobile game with a friend, it’s easy to appreciate the speed and convenience of close-range wireless communications.
Low-power transceivers make it possible to deliver fast data exchange over point-to-point wireless connections. Industrial designers can forego Ethernet connections that need physical cables and connectors with low-power millimeter wave devices that transfer data much faster.
But these new-generation wireless devices can be complex to create and validate.
That’s why semiconductor manufacturer STMicroelectronics (ST) is using advanced HPC technologies to help its customers and partners explore the world of next-generation wireless.
The company’s low-power ST60 RF transceiver can transfer data over short distances at up to 6.25 Gb/second using the 60 GHz V-band.
Since the antenna operates at such a high frequency, however, any variation in the manufacturing process can dramatically affect its performance
The main challenge is fitting together the system’s multiple components, says Olivier Bayet, ST’s leader in silicon packaging co-design, signal integrity and power integrity.
“The very high frequency of 60GHz means that both environment and variation in the manufacturing process can significantly affect the performance of the antenna,” he says. “We need to ensure the robustness of the antenna design and make sure the package and system elements work together correctly to ensure fast, reliable data transmission.”
The proximity of radio frequency (RF) and digital processing on a single silicon device can also be an issue, Bayet says. “We need more computer power to solve the challenges of integrating more components in such a small area of the package.”
More HPC capacity would also help decrease the critical time-to-market interval. ST simulates and analyzes thousands of cases to confirm the antenna’s robustness in relation to manufacturing tolerances.
“We want to bring products to market within a shorter development cycle,” Bayet says. “We need HPC capacity to run all the analysis and simulation within our desired timeframe.”
That’s why ST expanded its on-premises HPC infrastructure using HPE ProLiant servers based on Intel Xeon Gold processors. These servers offer robust four-socket performance and built-in workload acceleration, with advanced security technologies for cloud and networked clusters.
The solution also uses Ansys High-Frequency Simulation Software (HFSS), a leading package for designing and simulating high-frequency electronic products, including antennas.
Engineering teams across ST uses Ansys engineering packages for analysis and simulation. The company’s software uses Intel AVX-512 and oneMKL to optimize efficiency and performance.
Wim Slagter, strategic partnerships director at Ansys, says this can save weeks of development time while increasing design robustness. “Running simulation applications faster means more time to iterate and improve designs, as well as reduced time to market.”
When the load on its-on-premises compute farm gets too high, the ST solution can also access dedicated resources in the Microsoft Azure cloud, where machines are configured with the same Intel technologies and Ansys HFSS software as the on-premises systems.
The Azure cloud lets ST take advantage of more systems running in parallel. “We have the capacity we need for our critical analysis and simulations, and we don’t have to worry about disrupting other ST workloads,” Bayet says. Most of the servers used have 512 GB of RAM, but some runs require up to 800 GB of RAM per analysis.
By upgrading its Intel Xeon Scalable processors and adding Azure’s elastic capacity, ST has been able to dramatically speed up analysis and simulation workflow.
Shaving days off complex simulations leaves more time to analyze difficult aspects of the design and helps reduce validation costs, Bayet says. Analyses that would have taken five days can now be done in two; one analysis of 6,561 cases that took 11 days can now be finished in under a week.
Bayet’s team has been able to streamline the development process by redesigning its workflow to perform more analysis and simulation during the design phase.
As a result, there are fewer cases to analyze during the product validation phase, which means fewer physical samples are required, which in turn means lower costs.
It’s easier to identify areas that may need more focus once samples are in hand, and helps head off manufacturing tolerance errors. The result is a more reliable product that maximizes data transfer rates while minimizing power consumption.
There were multiple changes from the beginning to the end of this work that made this performance possible, Bayet says, including changing tool versions, hardware setup and tool setup.
Together, Bayet explains, Ansys software and Intel technology-based infrastructure has helped ST produce innovative 60 GHz V-band products within schedule requirements, saving both time and money. Having similar setups in the cloud and on-premises helped ease the switch between the two environments.
“With our Intel-based HPC and the performances increases of the last few years, we have been able to perform analyses that would not have been feasible before,” Bayet says.
About the author:
Julian Smith is an award-winning green tech, conservation and travel writer based in Portland Oregon whose work has appeared in Wired, Smithsonian, New Scientist and the Washington Post among others.
# # #
This article was produced as part of Intel’s editorial program, with the goal of highlighting cutting-edge science, research and innovation driven by the HPC and AI communities through advanced technology. The publisher of the content has final editing rights and determines what articles are published.