Designing For The IoT

Each day of the Design Automation Conference (DAC) starts with a keynote. What was particularly noteworthy this year was that they all focused on some aspect of the (IoT), and the value of hardware within the IoT.

Semiconductor Engineering attended all of the keynotes. What follows are highlights of those talks, along with some analysis. (The fourth day’s keynote can be found here.)

_______________

Joe Costello, CEO of Enlighted, which focuses on smart building technology. Well-known within the EDA community, Costello became Cadence’s CEO in 1987, turning it into the largest EDA company during his tenure.

The Keynote
“The Internet boom was tremendous and was probably the biggest revolution that we had had in technology, but the IoT is going to be bigger,” Costello said. “The Internet was somewhat of a negative for the design industry because it caused a tremendous amount of consolidation of the industry. In the beginning, it was about the desktop computer and a little bit of laptops and connecting those to servers and over time it became just laptops and servers and then the mobile phone revolution came along and then smart phones and then there was a consolidation of vendors. The number of components and vendors that are successful has gotten smaller and smaller. So, instead of diversification there has been tremendous consolidation. The IoT will be the opposite. It will be a great time for design and design automation. There will be a tremendous proliferation caused by the IoT. Why? Because there is a tremendous number of things and they will not see the same consolidation. Everything has different requirements.”

He noted that different technologies are required to bring the IoT experience because the size, shape, communications requirements and sensor requirements are all different for all classes of things. So rather than the consolidation caused by the Internet, the IoT will drive tremendous differentiation throughout the next decade.

Enlightened is a case in point. The company develops sensor systems for commercial real estate (CRE). Seven sensors are clustered into a module that fits into each light fixture, which guarantees (in some cases, by law) that they are well distributed throughout a building. Each has about the same processing power of the original iPhone and collects data about light, heat and temperature, among other things. But the real key is the DSP built on top of the sensors to process data and extract relevant information.

That information can be used for such things as space planning and asset tracking. The value of a lighting application is a few cents per square foot. Space, in comparison, costs about $3 per square foot, making it about 100X more valuable than energy. Asset tracking can include such things as wheelchairs in a hospital, which can delay patient departures.

“The data is the critical thing,” Costello said. “And you have to protect it like crazy because that is where the real value is. The flip side is it is not enough to just have data. You have to do something with it. It doesn’t matter what category you pick – it is a sensor system platform for that type of equipment. In the end, there will not be 50 or 100 platforms. There will be a handful that ends up surviving. The biggest challenge is to be one of the last five standing. So scale is important.”

There is a divide between operational technology (OT) and information technology (IT). OT is the lighting, HVAC facilities, infrastructure. It is 100 years old. It is very slow moving. But that is where the sensory system has to go. There is a huge gap between these two groups of people. That means you have to have a good enough understanding of the OT side. But just building a bridge is not scalable. You have to do more to scale. Over time, you have to give up most of what you built during the first 5 years of your company. You have to turn the verticals into a set of horizontals to be able to attack the world and that is hard. That is how you scale.

Analysis
Costello believes the money in IoT will be made by systems that analyze data, and which can take actions within a physical space. While sensors are necessary to gather data, there is not necessarily long-term value in the sensor market. The real value is in the data, and the ability to analyze and then utilize that data is what is scalable. That makes it imperative that the data is protected and secure.
_______________

Chuck Grindstaff, executive chairman of Siemens PLM Software, a business unit of the Siemens Digital Factory Division, and the unit that recently purchased Mentor Graphics.

The Keynote
The starting point for his thesis is that IC designs are at the heart of innovation. The underpinnings of the digital world is compute power, storage, and the ability to solve problems both centrally and on the edge of the network that were not possible in the past.

“Many companies are faced with a challenge in today’s world of digitizing their operations,” Grindstaff said. “While individual pieces of an organization may be digitized, those systems are not fully integrated. Across the supply chain there is information that is re-entered, and across disciplines productivity is lost. The digitalization of a company, from design to manufacturing as well as manufacturing into operational phase, will make the difference between success and failure of product companies in the future.”

Siemens makes many things, including wind generators, which have control systems to optimize the operation of the turbines. What is important is that the business model associated with a wind turbine has evolved over the years. Instead of just selling one, their goal is to sell capacity, capability, and performance of an entire wind farm. If the devices are connected and instrumented, then it becomes possible to optimize performance, predict failure and coordinate activity.

That requires leveraging intelligence in the operational phase that was designed in from the beginning. The business model transitions from selling a product to selling a capability over the lifetime of the product in the marketplace.

Grindstaff provided an example. “An automotive company’s business model is based on the idea that you will generate cars, sell them and make money. But they will be parked for 95% of the time, which means that business model is at risk if a competitor figures out how to bring autonomous driving and instant scheduling together to deliver mobility. If competitors are able to monetize the operation phase of the vehicle – not just at point of sale, it changes what a car company could be.”

The ability to ubiquitously put sensors and feedback into products, provides the basis for looking at the operation of the product in the field, predicting failures and optimizing operation. “A lot of stuff is getting connected together and it is getting connected because of a change of what is possible,” Grindstaff said. “Many companies are becoming specialty chip designers, integrating controllers with MEMS, RF – everything necessary to compute on the edge. What is critical is that each innovation across the spectrum is integrated into, and feed into, business model changes.”

The V diagram is at the core of many engineering foundations. Progress has been made in EDA to move to higher levels of abstraction, and this is happening in other areas as well. He believes we are progressing on what can be synthesized from requirements into control system behavior and the integrations of those control systems with the final product. That leads to a fully multi-disciplinary design process.

“We can look at the signals coming from the sensors and identify one that is different,” Grindstaff said. “It used to be in this range and now there is an anomaly. It could be that there is an ambient temperature impulse that caused the change in behavior. The functional model is the core of the next generation of analytics. We call that data-driven analytics. We combine data with model analytics and integrate them together. It allows us, in the cloud, to bring together simulated data, raw data, and the models from design systems and meld them together.”

Analysis
By owning the devices into which sensors are being placed, it can be transformed from a product that is sold to a product that produces revenue throughout its lifetime. Not only is the data not available to others, but the functional model on which analysis is based is proprietary. They have overcome the OT/IT divide because they own both sides of the issue and they are isolated from competition in the IoT space. They continue to compete on overall product capabilities and lifetime costs.

_______________

Tyson Tuttle, the CEO of Silicon Labs, a provider of solutions for the IoT.

The Keynote
By 2025, the IoT is expected to be comprised of 70 billion devices. Today there are about 20 billion. The economic value per year for the IoT, for connecting all this together, is $11 trillion. The global GDP for the entire world economy this year is about $78 trillion. In 2025, global GDP will be about $100 trillion.

“With improvements in productivity and growth with inflation, about 10% of that is being driven by automating the world economy and delivering all of this value to help the economy grow and raise people out of poverty,” Tuttle said. “Inform the world and make things more efficient.”

The semiconductor industry generates about $350 billion in sales each year. The IoT value is in the hardware, the applications, and the data. Tuttle points out that it is not just one application or customer or company. It is tens of thousands of customers, thousands of applications, and it opens up an opportunity for a wide variety of technology.

Tuttle believes the infrastructure and a critical mass of technologies exist today to be able to support this growth, but it is not yet fully deployed. “I view this as going to take multiple decades to play out,” explained Tuttle. “You have the ubiquitous networks, you have the Internet, you have mobile networks, you have the Cloud, you have Big Data, AI – all of the algorithms running on these computers. Certainly we are making them faster and more capable over time.”

IoT SoCs look a little different than those that have been deployed in smart phones. First, the economics are different. A product that is not hooked to anything is static. When it gets connected, it adds inherent value, which is the inflection point. The network effect, which includes ecosystems and multiple devices working together, creates a multiplier on the value.

From there, things have to evolve. Once a product is connected it lives into the future, and you have to manage it, update it, evolve the functionality, and that creates new business opportunities as well as challenges.

“Consider a power drill,” Tuttle said. “They added connectivity – Bluetooth — to add a better user interface. Every contractor has a smart phone. That was cool, but who is going to pay you more because you have an app on the smart phone? They started to add multiple features. They made it so that if you don’t have your smart phone, the tool doesn’t work. This is a big problem on job sites. That prevents theft. Every time the smart phone sees the tool, it records the location, so you know where it was. Asset tracking is important.”

It turns out that for large-scale solar installations in the desert, over-torquing the screws can cause problems about 180 thermal cycles later. As a result, the utility programmed a special torque curve into the drill to avoid that problem.

“They could automate the regulatory compliance paperwork,” added Tuttle. “It eliminates a load of manual operations. They could take this simple idea, adding Bluetooth, and migrate that into a new business model and charge for services, enhancing their value proposition. There are thousands of examples like this. There is a long tail of applications, and that is what makes this the most exciting area to work in.”

The challenge for Silicon Labs was how to scale a company so that you don’t get integrated out and you increase your value proposition. They needed to find a way to address tends of thousands of customers.

An IoT device has embedded flash, embedded RF, an ARM processor, energy management, a bunch of digital logic. “You need a portfolio of these devices to support the different applications,” Tuttle said. “You need different memory configurations, different sensor interfaces. Do you try and build a platform that can serve the entire market, or specific optimized devices for each application?. We are working on the platform approach so that you can do one hardware platform and get multiple products out of it and address as many applications as possible.”

The other challenge is scaling support and making complex software and hardware easy to use so that industrial customers can design this in and deliver value in their product. One way is to take chips and put them into modules that are pre-certified. The lower the cost, the more it will get deployed. That means it has to be integrated onto a single chip, and it needs security that is optimized for these applications. And it needs to be able to evolve and adapt over time.

“Since World War II the globalization of supply chains has brought tremendous improvements in productivity—being able to make things cheaper and get them to markets across the world,” Tuttle said. “But it also had huge upheavals in terms of politics and policy. Lots of people lose their jobs, but at the same time it creates opportunities and value. We are going through another one of those times now. We have to keep an eye on how the technology impacts the world, impacts society.”

Analysis
Someone has to make the devices that connect things together. Not everyone will want a custom chip or will be able to afford to develop all of the software necessary to get it integrated into a product. The long tail will create opportunities for those companies that can fill the gap economically with platform-based solutions.

Conclusions
Three presentations, each from a hardware perspective, examine different ways to capitalize on the growing IoT market. Siemens owns the entire product and lifecycle of the product and uses IoT to add value to that product. The company has overcome the OT/IT divide because it controls all of the pieces themselves.

Enlighted is focused on understanding and penetrating the OT side to build the IT piece of the solution. The company invested five years building sensor units in order to be able to show the true value of a connected office. It sees most of the value turning data into actions.

Enlighted is willing to let go of the low-level hardware piece to companies like Silicon Labs. While this is the most likely piece of the puzzle to become commoditized, Silicon Labs believes enough value can be encapsulated into the IoT devices, and that a catalog of devices will make them the MCU company of the IoT age.

What is clear, though, is that the real value is about the data—collecting, distilling, and utilizing it. That trend has not been lost on anyone in the hardware space. The only question is how that will play out as a business—whether it will be centralized as it is today with big companies such as Google, Facebook and Amazon, or even Siemens and Enlighted—or whether it will be distributed across smaller companies that can turn that into a specialized business.