Connect the Dots is a new, weekly blog that looks at innovations across the IoT technology chain and what they mean for solution providers. From processors, boards, software and design tools, to data analytics, application software and network and cloud services, we will keep you abreast of what it takes to create and implement your next best industry-specific solutions. Today, we discuss the “edge” application, in which an IoT system initially connects with the world through a sensor.
At next week’s Embedded World Conference, the industry will gather en masse to discuss the future of embedded: that is, the Internet of Things and all the ingredients and partnerships that implementation of such solutions entails.
Many of those ingredients have their own tracks and dedicated sessions. For example, wireless and wired connectivity, bus architectures, operating systems, software tools, security options, power and power management, and, of course, processors, from microcontrollers to multicore processors to FPGAs and systems-on-chip (SoCs). All those elements have been in existence long before the IoT, so what makes them different now? What sets them apart? And how do they fit in to the bigger IoT solution picture?
Let’s look at the “edge” application where an IoT system first interfaces with the real world through a sensing device. That sensing device can be a thermocouple or vibration sensor checking the health of a motor in on factory floor, or a tire pressure monitoring sensor (TPMS) tracking tire pressure on an 18-wheeler flying down the I-95 corridor in the rain.
Either way, the sensor data must be accurate and fed reliably to a processor via an analog-to-digital converter. Processors are the foundation of the IoT: Once in the digital domain, processors take over.
For remote applications, such as monitoring bridges, or earthquake research, some fundamental requirements of the IoT processor are that it:
- Consume as little power as possible as it may be battery powered
- Be inexpensive, as there may be hundreds or thousands of such processors deployed.
- Be highly integrated, to keep size and down while maximizing features and capabilities.
- Be easy to use with a solid ecosystem so developers and integrators can get their ideas and designs to market as soon as possible: IoT design cycles are shrinking monthly.
A good example of such a processor is the Intel Quark SE, which combines a microcontroller with an on-board sensor subsystem. The combination helps to minimize power consumption by managing wake cues.
In essence, the goal is reduce the amount of time the processor is in active mode, so the fewer times a sensor unnecessarily wakens it, the less power will be consumed. Having the sensors on board the chip helps ensure better control keeps them in lock step and reduces the processor “on” time, or “duty cycle.”
Figure 1 (right): The Quark SE was announced in November last year and is an extreme example of a low-power, highly integrated processor for IoT applications such as industrial control and monitoring, hence the wide temperature tolerance of -40 to 85°C.
The Quark SE is expected to be available to developers in the first half of this year (2016). In addition to its integrated sensor subsystem, Quark SE has pattern-matching technology that lets it learn through pattern recognition and differentiate appropriate response events.
Figure 2 (left): The Quark SE microcontroller is highly integrated, including features such as an internal sensor hub as well as on-board pattern-matching technology. That helps designers and developers as they can do more at the edge and be smarter about what data they transmit upstream.
The intelligent feature is important within the context of the whole IoT solution chain, as there is an on-going architectural debate as to how much intelligence should be at the edge, and how much at the core, or at the cloud. The simple answer is to have as much intelligence at the edge as you can afford in terms of power and cost as it helps to improve response times. It also can enhance security, as not all data needs to be sent upstream to the cloud, depending upon the edge-node programming and decision-making process.
Other features of the Quark SE include low-frequency operation (32 MHz) to minimize power consumption, as well as support for real-time operating system (RTOS) support, but not Linux. It is fully x86 instruction-set compatible.
With the choice of processor come other considerations such as software ecosystem, programming environment, OS, security, compatible components, and many other associated ingredients that go into making the board and the system.
Still, a solid, highly integrated hardware foundation provides the flexibility to scale and add new features over time, and at a higher level, to design a full end-to-end IoT system that distributes the data gathering and processing optimally.