Driving IoT innovation along a roadmap of hardware, software, and tools
April 12, 2018
Story
Intel's roadmap spans its processor portfolio with features that include increased performance scaling for the edge, graphics and media processing, improved determinism, better security, and more I/O.
Industry players are reviewing and enhancing their IoT capabilities to address emerging challenges and applications. Intel, for instance, has recently made announcements surrounding its capabilities roadmap that takes a wholistic approach to solving IoT development challenges.
Jonathan Luse, General Manager for IoT Planning at Intel, interacts with customers involved with IoT systems development, implementation, and operation.
What's driving Intel’s roadmap?
According to Luse, connectivity and the way in which traditional embedded systems are being integrated into larger real-time to cloud systems represent two of the key megatrends in IoT development.
“One key difference between embedded and IoT is the connectivity. We’re transitioning from isolated devices into a group of connected devices with awareness of their surroundings," Luse says. " Our roadmap changed significantly because processor and system requirements turned from device-centric design to device-within-a-system where information and workload is being shared. Devices for end customers are being used in a system, and that fundamentally changes how they think about processors."
One challenge for engineers developing systems like those mentioned above is the tremendous amount of data that is becoming available. Given the dynamic nature of this information, Luse cautions against over-engineering from the outset of designs.
"Data overload is a big trend right now,” he says. “There is a tremendous amount of data being generated by devices, as well as the sharing of this data, which generates more. Just because these devices can generate data doesn’t mean they should. You can end up in a situation with a huge signal-to-noise ratio (SNR), for example, which is a problem for businesses. To be successful, it’s important that business stakeholders and developers understand exactly what‘s signal and what’s noise. Developers are challenged to solve the data overload problem by identifying the important information and removing the noise.”
One example of such a system is a store surveillance camera that generates “bonus data.” The original application is simply security footage, but this information could also be used to analyze traffic patterns, what areas are frequently visited and why, areas where people spend the most time, and experimentation and pattern analysis to determine optimal product placement for maximizing revenue.
“If you have a good idea of how to use the data to achieve an objective, that’s valid data. Understanding the business questions you want to answer is the best way of approaching the technical design," Luse adds.
IoT technology investment and return
Intel launched its latest generation of client processors this week with many key features for IoT. The roadmap spans Intel Atom, Core, Xeon, and other processors, with a list of features that includes increased performance scaling for edge intelligence, graphics and media processing, improved determinism using Intel Time Coordinated Computing Technology (Intel TCC Technology), better silicon security through Intel's Trusted Execution Engine (Intel TEE), and more I/O for edge and cloud connectivity.
The technical feature list is supported by a long life and supply availability required for embedded, industrial, and IoT applications.
“Many of our markets have approvals or certifications associated with them,” Luse says. “It takes time to get these certifications. Once they have them, our customers want to have production supply in place for a long time – fifteen years in some cases – and our roadmap offers that.”
Software support is also a critical aspect of the roadmap, and with Intel’s history, there are many partners in the ecosystem.
“We have customers that are tied to multiple Microsoft engines, so we have a Windows OS suite as well as Linux and Android flavors. These tend to be the base operating systems that are universal on every roadmap. Xeon tends to be server version of Microsoft whereas Atom is a Windows IoT client, as well as Linux/Android.”
Of course, specific distributions are best-suited for specific projects. Examples include automotive, where you need features like fast boot (within 2 seconds) and industrial where real-time OS suites are needed. Virtualization capabilities are also of particular interest to customers wanting to do workload consolidation, Luse explains.
“There is a significant trend where separate systems are being consolidated on a single platform, but need to be logically isolated with hypervisors that provide resource and operational separation.”
However, it is the tools environment may be the most disrupted by IoT, Luse posits, given its mash-up of embedded processors and accelerators, network gateways, and cloud applications. He explains by referring back to the security system example.
“If you think about all the accelerators to do analytics – CPU, graphics, video accelerator, deep learning engine, FPGA – you're talking about 4-5 different programming environments,” he says. “Tools have become important to the developer experience. It’s not the same old tools environment. Tools must be done in a way that allows developers to move the workload and acceleration across all these accelerators in the cloud, gateway, and device as seamlessly as possible.”
“Extending the analytics system of a store using security camera footage on a DVR may require analytics at the store for traffic patterns and shopper behavior," Luse continues. "The larger corporation may pipe results into the cloud where they do cross-functional heat mapping for traffic patterns across all their stores. As the back end applications mature, they may be migrated the edge for real-time, instant feedback to get store helpers where they need to be when they need to be there. In order to do that you have to have tools that allow this transition. The device might have different accelerators than those available in the cloud. The challenge Intel is addressing is how to make this migration possible while not losing the ability to tune things for a specific accelerator in the IoT pipeline.
"Some want to hit the ‘easy’ button while others want the ‘powerful’ button," he concludes.
Connectivity options, the ability to migrate processing up and down the IoT technology stack, and the software and tools enable this creates unprecedented visibility and immense potential for developers and businesses.
