System level electromagnetic (EM) simulation is becoming an increasingly important technology needed to support the design of next-generation smart cities and transportation as well as global communications systems. I recently spoke with ESD Alliance member company CEMWorks’ President, Jonatan Aronsson, and Cielo Gerrie, Vice President of Business Development, about the challenges and opportunities that are driving the need for automating EM simulation. CEMWorks is on the cutting edge of EM in-house simulation research.
Smith: Which market segments require automated tools to model systems and optimize them for electromagnetic effects?
Jonatan Aronsson: Many market segments including systems based on semiconductors and telecommunications require the use of automated tools for EM simulation. With the strong push toward self-driving cars, the automotive industry is an example of multiple telecommunications applications that require EM analysis. This includes internal communications required for critical vehicle functions such as steering, motor control and braking as well as vehicle-to-vehicle and vehicle to outside communications services.
Other examples include the defense industry’s efforts to design radar-evading stealth aircraft that requires simulation and analysis of communication links and electronic warfare systems. The aerospace industry requires simulation to determine placement of antennas on passenger aircraft. It’s quite complicated because of the many antennas and potential interference between them and the fact that antenna behavior and performance changes when mounted on a large metallic body.
Smith: I can see how challenging that could be and a problem with an antenna could be dangerous. How do engineers get their arms around that?
Aronsson: EM simulation plays an important role in determining antenna placement. The simulation must consider all the antennas, how they interact with each other, and how they interact with their surroundings including potential interference from other on-board or nearby communications systems. Solving a problem of this scale of complexity requires automated solutions.
Another growing area is embedded devices. Sensors can be embedded into someone’s body or on the surface. Safety aspects need to be considered and engineers using EM simulation need to predict how much energy might get absorbed.
Smith: You mentioned self-driving cars and communications. How complex is the modeling of the EM effects?
Aronsson: There are so many more systems coming online that need to communicate. It used to be that a car would have a global positioning system (GPS). Now it’s GPS coupled with radars and communications between vehicles. Some new vehicles are likely to have vehicle-to-vehicle communication where vehicles can exchange sensor information. One vehicle may notice an obstacle and transmit that information for other vehicles to pick up. Or vehicle-to-external (V2X) where vehicles will communicate directly with traffic lights.
Smith: The traffic light will take over my car and apply the brakes?
Aronsson: Yes, quite possibly. If all vehicles have this communication, then traffic lights potentially wouldn’t be needed because they could synchronize. If that happens, the reliability of the communication link becomes critical particularly when there’s no line of sight. Think of an intersection with skyscrapers at each corner and no clear line of communication. The vehicle will have to rely on communication signals that bounce off the buildings. Analyzing how these signals propagate and how to control them is an increasingly complex challenge that needs to be solved as these technologies are rolled out.
5G is another area where the complexity is increasing with the push toward using millimeter wave frequencies. These electromagnetic waves don’t travel as far and are easily obstructed by buildings, trees, vehicles and even pedestrians. At lower frequencies, they can penetrate through buildings and bounce around corners. For millimeter wave applications, this means antenna placement for base stations becomes more critical. This is also the reason why 5G networks typically rely on small cells to achieve adequate coverage, that can be mounted on existing structures like light posts and traffic signals.
Another interesting technology that is needed to improve millimeter wave communications is what we refer to as metamaterials or reconfigurable intelligence surfaces. These materials can be affixed to the side of a building, for example, to do beam forming that allows the wave to be reflected in different directions in real time. This is one way to overcome the line-of-sight challenges that exist in dense urban areas.
Smith: Is there actual mechanical action going on within these metamaterials? Are MEMS involved?
Aronsson: No and no for the applications that we are working on. It is more efficient to use an array of diodes that can be switched on and off. Switching the diodes on and off changes the resonant frequencies or radiation pattern of the surface. It is typically based on using passive electronics though some approaches rely on control by active circuitry. Metamaterial surfaces can be manufactured cheaply and they can be printed, even on flexible film and then affixed to building materials.
The challenge on one end is the base station, usually an antenna array. Beam forming might be used to track a pedestrian. With traditional antennas, energy goes in all directions. For millimeter wave 5G networks, it needs to be concentrated and it is challenging to figure out where to direct the beam. On top of this, adding reconfigurable intelligence surfaces adds one more layer of complexity because the operator of a reconfigurable surface must decide where to direct the beam and provide real time instructions on when to change the configuration. The configuration can be changed using very low power because it is simply turning diodes on and off.
Smith: Are these materials new and deployed yet?
Aronsson: Metamaterials have been actively researched for over 20 years. The idea of using them to improve the millimeter wave or 5G coverage is intriguing and telecom companies are working on testing them out.
Gerrie: During Mobile World Congress in Barcelona, Greenerwave and Anritsu demonstrated a prototype for telecommunications use. It’s a focused area right now with publications and international collaborations to optimize these frequency-selective surfaces, not only the surface itself but also the application of the whole technology to improve the network.
One interesting case study was the development of metamaterials to put on the side of buildings instead of big clunky antennas or hardware –– more like a film that would adapt to the surface of a building, billboard or window. It’s something that doesn’t require power, enhances signal integrity and does not interfere with the environment.
The challenges with metamaterials include ensuring optimal performance, safety and efficacy. This requires expertise in electromagnetics and material science, coupled with numerical methods and computational techniques.
We are involved in a number of projects with partners in the European Union. One was called the AI-enabled Massive MIMO (AIMM) project. It touches on the ability of these surfaces to adapt with artificial intelligence (AI) and machine learning (ML). In addition, we are the project leaders of InnoStar which aims to develop a new design-flow for designing hardware that will improve network integrity.
Smith: What was the AIMM project focus? Has it concluded?
Aronsson: We were investigating the application of AI to learn how to optimize the beamforming capabilities of these surfaces. Over time, AI would learn the environment and then be able to optimize the beamforming. The project which included British Telecom and some other companies in the U.K. and Germany successfully concluded.
Smith: You mentioned a number of different challenging vertical markets. What’s the biggest and thorniest?
Aronsson: In terms of EM simulation challenges, there are many. 6G is challenging because it’s moving to 28 gigahertz and 100 gigahertz, and possibly up to even 300 gigahertz. Companies are trying to build semiconductor systems that can handle these high frequencies. Engineers are looking at new ways to implement antenna arrays, like embedding them in the package or chip.
Given the high frequencies involved, the traditional approach where each component or block is simulated separately and then tied together with a SPICE model does not supply the accuracy needed. The system must be simulated as a whole including all of the interactions. This becomes a complex computational problem. A solution to overcome that kind of challenge is something we’ve been looking at.
Robert (Bob) Smith is executive director of the ESD Alliance, a SEMI Technology Community.