The industry revolves around two types of AI, namely research AI versus production AI. I wasn’t sure what these terms meant and their differences, so I asked an expert to join me on this week’s Embedded Executives podcast. John Weil is the Vice president and General Manager of Synaptics’ IoT Business Unit. After setting me straight, John talks about the last 20% of a design that the customer needs to figure out for themselves, amongst other things.

What is the real definition of a software-defined vehicle (SDV)? It depends, mostly on who you are asking that question of. I asked John Heinlein, the Chief Marketing Officer at Sonatus and he had a great response on this week’s Embedded Executives podcast. We discussed the differences between various tiers of vehicles, how they are upgraded, and what we should be looking out for as we move forward. Note that John has his podcast and was gracious enough to have me on as a guest recently. Hear that one here.

Is documentation an issue for your embedded project? Wouldn’t it be great if you didn’t have to create that documentation yourself, yet have it be as good (or better) than you could do yourself? That’s what the tool from Driver does. In full disclosure, I didn’t understand how the company’s technology worked, so I had a great discussion with Driver’s CEO and Co-Founder, Adam Tilton on this this week’s Embedded Executives podcast.

The amount of NAND memory that exists today is staggering. There are many different kinds of NAND, and the needed type is based on the intended application. Sometimes you need high-speed memory, sometimes very dense memory, and sometimes you need the memory to go into an environment that’s less than favorable. But the memory itself is just one piece of the system puzzle. You also need a controller to go with that memory. You might think like I did, that the controller is fairly simple to design and would be something the memory suppliers can handle without breaking a sweat. As I learned in this week’s Embedded Executives podcast with Robert Fan, a Senior Vice President with Silicon Motion, it’s harder than you might think. And for that reason, the controllers from vendors like Silicon Motion are highly in demand. 

With all the focus on AI and machine learning these days, we sometimes take other facets of design for granted, some of which are vital, like the wireless component. Synaptics’ Veros architecture fits the bill for state-of-the-art wireless communications with high throughput, low power consumption, and top-notch security. If you’re designing an AI-based platform using the company’s latest AI platform, Veros has built-in hooks to attach the two subsystems seamlessly. To understand what that means and how it works, I spoke with Venkat Kodavati, Synaptics’ SVP and GM for the company’s Wireless Products Division, in this week’s Embedded Executives podcast.

We are being bombarded with data; it’s coming at us from so many avenues, particularly as we move into the world of AI and the analytics that it brings. The key to a proper edge-to-cloud strategy is being able to make sense of that plethora of data. That statement sounds obvious and simple, but it’s quite the opposite. If you don’t have the proper strategy in place, your design could be doomed before it even gets started. To understand where to begin, and what path to take with the data strategy, I spoke to Johannes Biermann, the President and COO of aicas, who gives some guidelines for developers, on this this week’s Embedded Executives podcast.

When designing systems with high integrity, say for automotive applications, what is the programming language of choice? I believe that’s referred to as a loaded question, because there are so many variables involved, and it’s a question that’s almost impossible to answer. Unfortunately, many developers make a decision for the wrong reasons. To get to the root of the matter, I invited Quentin Ochem, the Chief Product and Revenue Officer at AdaCore, to be my guest on this week’s Embedded Executives podcast.

What is rad-hard memory, and when and where should it be used? And what makes it different from conventional memory? The answers may seem obvious, but that’s not the case. That’s exactly why I spoke to Helmut Puchner, Vice President and Fellow in the Aerospace and Defense division of Infineon Technologies in this week’s Embedded Executives podcast. Obviously, such memories cost more, so you want to be sure you understand when they need to be deployed and when they don’t.

The automobile's in-cabin user experience (UX) is undergoing some serious changes. We’re nearing the point where the automobile can be viewed as an office or entertainment center that also transports you from point A to point B. With all those changes, it’s important to ensure that these non-mission-critical functions communicate with the mission-critical tasks at some point. To make sense of this, I spoke to Chet Babla, a Senior Vice President at indie Semiconductor on his week’s Embedded Executives podcast.

The Zero Trust Architecture (ZTA) provides a new approach to cybersecurity, and outlines a great model for developers to follow. Like the name implies, you should trust no one, at least initially, and assume that all networks and traffic are potential threats. We know that no network is ever 100% secure, but beginning with this assumption gives you the safest starting point and should reduce the risk of data breaches. ZTA is based on the principle of “least privilege,” which means that users and devices are only granted the permissions they need to perform their specific tasks. To help explain what this means, I spoke to Steve Hanna, a Distinguished Engineer from Infineon Technologies on this week’s Embedded Executives podcast.