Electronic Products & Technology

Smart nodes at the edge of the industrial Internet of Things

September 5, 2014  EP&T Magazine

We live in a connected world where the Internet brings us more information than ever before at ever increasing speeds through millions of connected devices. This Internet of Things (IoT) is poised to revolutionize our world, enhancing ease-of-use, performance, efficiency and access. And, industrial automation will lead the way with smart nodes at the edge. However, to support the performance and cost parameters that will spur adoption of IoT industrial devices and solutions, embedded designers will need to use the right mix of processors, memory, connectivity, sensors, security, power and operating systems to deliver economical solutions that can be broadly adopted and maintained.

Connecting electronics to the Internet in all manufacturing brings a huge opportunity for increased production. Fortunately, the system-level hardware components of this manufacturing template already exist. Just look at current and emerging atrobotics, automation, enterprise resource planning software, server processing capabilities and network fabrics.

Making industrial IoT a reality

Making industrial IoT a reality involves creating an infrastructure for highly optimized and intelligent manufacturing, the right system configuration and an increase in memory needs, connectivity requirements and a focus on reduced system latency.

Latency can be improved in many different ways including both hardware and software optimizations.

For example:

* Higher memory and processor bandwidth

* Upgrading connectivity to Ethernet or EtherCat increases network latency

* Software systems partition in ways that allow certain code to execute from memory directly

* ERP systems leverage faster memory such as Flash over HDD to improve latency

* Server virtualization increases utilization and optimizes scheduling

Reducing latency at the system level must be a key focus for industrial IoT. Benefits can be achieved by shifting the compute power of today’s central processing to a wide network of application-specific nodes, configurable for core functions within the overall system. Connected by a shared network, this fabric of ‘smart’ nodes can usher in greater efficiencies by automatically sensing environmental changes, reacting with quick parameterizations, storing valuable information, processing and executing tasks and streaming data to the cloud and other systems.

The market is exploring the need for this new smart compute fabric and how general purpose and application-specific processors will enable the transformation, as well as the necessity for more or faster memory throughout the system and the demands hardware components will need to face. One of the main demands is to increase the memory data throughput while at the same time decreasing the pin count. Here, a main bottle neck in a typical microcontroller-flash system is the interface between the flash memory and the microcontroller.

Developers moving towards low-pin-count interfaces

More and more developers move away from using the parallel interface to low-pin-count interfaces like serial peripheral interface (SPI) or Quad-SPI. However, there are already new bus interfaces that increase performance more than five times over traditional Quad-SPI flash. This kind of interface needs only 12 pins and can run up to 166MHz at double data rate and provides a max data throughput of 333MB/second. Furthermore the new bus interface overlays on the pin-out of a common 6mm x 8mm BGA package for a Quad-SPI memory and can be used for fastest boot or even execute-in-place (XiP) applications to reduce DRAM or to extend the embedded flash.

A smart node in this context is able to handle different communication protocols and front-ends to different applications, providing a holistic platform approach. These are highly reusable, scalable and quickly adoptable during development. For example, a designer can use the same core platform to build a wide range of devices such as a motion controller, a sensor phalanx, a gateway controller, a data concentrator, a monitor or an HMI console system. Additionally, designers can reconfigure the smart nodes in real-time to handle many different products. The possibility of self-determined control units becomes very real.

Connectivity is only 1 facet to IoT

Connectivity is just one facet of achieving industrial automation for Industrial IoT to become a reality. There are a number of key semiconductor device elements necessary to building this topography.

For example:

* Memory:

* Larger embedded density enables more complex execute in place software

* Higher densities also enable on-die data logging and storage

* Faster CPUs and integration of additional features such as communication-interfaces

* Connectivity:

* Microprocessor support for multi-protocol supports all environments

* Multiport interfaces provides broad connectivity

* Security:

* Boot block protection and sector erase are important embedded security features

* Microcontroller based encryption

* Safety:

* More periodic self-testing routines of hardware functions must be included in application software

* Redundancy, reliability, quality and endurance

* Combinations of dedicated hardware blocks and software routines can significantly increase functional safety

Enabling a system that decentralizes critical decision making functions requires many fail-safe features. Functional safety becomes an extremely important aspect of this feature set. In order to identify random failures during operation time of the system, more and more periodic self-testing routines of hardware functions must be included in application software. For example, CPU registers and program counters are checked against stuck-at errors and digital and analog I/Os of a controller are checked against fault conditions. Those tests are called BIST (build in self-test).

Combinations of dedicated hardware blocks and software routines can significantly increase functional safety. Growing requirements regarding risk reduction necessitate more and more dedicated hardware and software routines and therefore can also be seen as a driving factor toward more processing performance and a higher amount of embedded memories to host the additional functions.

With precise scalable solutions, microcontrollers designed for the industrial market with options from low power to performance that allows for differentiated features such as touch, connectivity and inverter drives, are essential for industrial automation. The combination of the MCU and software support will help customers simplify system design and accelerate time to market.

Conclusion

To make industrial IoT a reality, companies must address increased demands for better infrastructure, system configuration and reduced system latency. The right combination of memory, processing power, connectivity, power and security are also necessary to create intelligent nodes for machine-to-machine and human-machine interaction. Enabling a system that decentralizes critical decision making functions will revolutionize manufacturing. Industry will be better equipped to compete through this next wave of industrial revolution.


Print this page

Related Stories

Leave a Reply

Your email address will not be published. Required fields are marked *

*