MARCH/APRIL 2017 VOLUME 15 | 2 EMBEDDED-COMPUTING.COM
TRACKING TRENDS
High-res audio for IoT apps PG 5
IOT INSIDER Defining (artificial) intelligence at the edge for IoT systems PG 6
2017 embedded wireless landscape: LPWA standards seek Industrial IoT connections
Development Kit Selector
www.embedded-computing.com/ designs/iot_dev_kits
PG 10
IoT platforms:
The secret sauce for connected vehicle applications PG 24
The
OpenFog
Reference Architecture and the cloud-to-things continuum PG 26
AD LIST PAGE ADVERTISER 17
ACCES I/O Products, Inc. – Saving Space, the Final Frontier
7
American Portwell Technology – Empowering the Connected World
29
COMMELL Systems Corporation – PICMG 1.3 Fanless Barebone Based on 6th Gen Skylake-H
1 Digi-Key – Development Kit Selector 21
27
Kensington Electronics, Inc. – Do You Want Crucial Data Transferred to Be Resistant to It or Immune to It?
EMBEDDED COMPUTING BRAND DIRECTOR Rich Nass rnass@opensystemsmedia.com EMBEDDED COMPUTING EDITORIAL DIRECTOR Curt Schwaderer cschwaderer@opensystemsmedia.com TECHNOLOGY EDITOR Brandon Lewis blewis@opensystemsmedia.com
CONTENT ASSISTANT Jamie Leland jleland@opensystemsmedia.com
AUTOMOTIVE CONTRIBUTOR Majeed Ahmed
AUTOMOTIVE CONTRIBUTING EDITOR Jeremy S. Cook
DIRECTOR OF E-CAST LEAD GENERATION AND AUDIENCE ENGAGEMENT Joy Gilmore jgilmore@opensystemsmedia.com
CREATIVE DIRECTOR Steph Sweet ssweet@opensystemsmedia.com
SENIOR WEB DEVELOPER Konrad Witte kwitte@opensystemsmedia.com
Micro Digital, Inc. – SMX RTOS: Ideal for Your Project.
DIGITAL MEDIA MANAGER Rachel Wallace rwallace@opensystemsmedia.com
32 Supermicro – Connecting the Intelligent World from Devices to the Cloud
CONTRIBUTING DESIGNER Joann Toth jtoth@opensystemsmedia.com
31 u-blox – Turning the Industrial IoT Toolbox Blue(tooth) 3
WinSystems, Inc. – Power to Perform
SOCIAL
Facebook.com/Embedded.Computing.Design
@Embedded_comp
LinkedIn.com/in/EmbeddedComputing
WEB DEVELOPER Paul Nelson pnelson@opensystemsmedia.com
SALES/MARKETING SALES MANAGER Tom Varcie tvarcie@opensystemsmedia.com (586) 415-6500
MARKETING MANAGER Eric Henry ehenry@opensystemsmedia.com (541) 760-5361 STRATEGIC ACCOUNT MANAGER Rebecca Barker rbarker@opensystemsmedia.com (281) 724-8021 STRATEGIC ACCOUNT MANAGER Bill Barron bbarron@opensystemsmedia.com (516) 376-9838 STRATEGIC ACCOUNT MANAGER Kathleen Wackowski kwackowski@opensystemsmedia.com (978) 888-7367 SOUTHERN CAL REGIONAL SALES MANAGER Len Pettek lpettek@opensystemsmedia.com (805) 231-9582 SOUTHWEST REGIONAL SALES MANAGER Barbara Quinlan bquinlan@opensystemsmedia.com (480) 236-8818 NORTHERN CAL STRATEGIC ACCOUNT MANAGER Sean Raman sraman@opensystemsmedia.com (510) 378-8288
ASIA-PACIFIC SALES ACCOUNT MANAGER Elvi Lee elvi@aceforum.com.tw
BUSINESS DEVELOPMENT EUROPE Rory Dear rdear@opensystemsmedia.com +44 (0)7921337498
Pinterest.com/Embedded_Design/
Instagram.com/Embedded Computing
youtube.com/user/VideoOpenSystems
WWW.OPENSYSTEMSMEDIA.COM PRESIDENT Patrick Hopper phopper@opensystemsmedia.com
EXECUTIVE VICE PRESIDENT John McHale jmchale@opensystemsmedia.com
EXECUTIVE VICE PRESIDENT Rich Nass rnass@opensystemsmedia.com
CHIEF FINANCIAL OFFICER Rosemary Kristoff rkristoff@opensystemsmedia.com
CHIEF TECHNICAL OFFICER Wayne Kristoff GROUP EDITORIAL DIRECTOR John McHale jmchale@opensystemsmedia.com
VITA EDITORIAL DIRECTOR Jerry Gipper jgipper@opensystemsmedia.com
MANAGING EDITOR Jennifer Hesse jhesse@opensystemsmedia.com
EMBEDDED COMPUTING DESIGN ADVISORY BOARD Ian Ferguson, ARM Jack Ganssle, Ganssle Group Bill Gatliff, Independent Consultant Andrew Girson, Barr Group David Kleidermacher, BlackBerry Jean LaBrosse, Silicon Labs Scot Morrison, Mentor Graphics Rob Oshana, NXP Jim Ready, Independent Consultant Kamran Shah, Silicon Labs
2
ASSISTANT MANAGING EDITOR Lisa Daigle ldaigle@opensystemsmedia.com SENIOR EDITOR Sally Cole scole@opensystemsmedia.com
ASSOCIATE EDITOR Mariana Iriarte miriarte@opensystemsmedia.com
CREATIVE PROJECTS Chris Rassiccia crassiccia@opensystemsmedia.com
FINANCIAL ASSISTANT Emily Verhoeks everhoeks@opensystemsmedia.com SUBSCRIPTION MANAGER subscriptions@opensystemsmedia.com
CORPORATE OFFICE 16626 E. Avenue of the Fountains, Ste. 201 • Fountain Hills, AZ 85268 • Tel: (480) 967-5581
REPRINTS WRIGHT’S MEDIA REPRINT COORDINATOR Wyndell Hamilton whamilton@wrightsmedia.com (281) 419-5725
Embedded Computing Design | March/April 2017
www.embedded-computing.com
POWER TO PERFORM
Rugged, reliable and resilient embedded computing solutions WinSystems’ embedded single board computers are designed to support a broad range of industry applications in challenging operational environments. From energy and transportation management, to industrial IoT and automation—our systems enable the collection, processing and transmission of real-time data requirements at the heart of your overall system. From standard components to full custom solutions, WinSystems delivers world-class engineering, quality and unrivaled technical support. Our full line of embedded computers, I/O cards, and accessories help you design smarter projects offering faster time to market, improved reliability, durability and longer product life cycles. Embed success in every application with The Embedded Systems Authority!
SBC35-C398Q Quad-Core Freescale i.MX 6Q Cortex A9 Industrial ARM® SBC PX1-C415 PC/104 Form Factor SBC with PCIe/104™ OneBank™ expansion and latest generation Intel® Atom™ E3900 Series processor
SCADA
ENERGY
IOT
AUTOMATION
TRANSPORTATION
Single Board Computers | COM Express Solutions | Power Supplies | I/O Modules | Panel PCs
SCADA
SCADA SCADA SCADA ENERGY
ENERGYENERGYENERGYIOT
SCADA
IOT
IOT
IO60-M410 Data acquisition module for embedded systems with IO60 expansion and 24 GPIO
817-274-7553 | www.winsystems.com
IOT TRANSPORTATION TRANSPORTATION TRANSPORTATION TRANSPORTATION AUTOMATION IOTAUTOMATION AUTOMATION AUTOMATION TRANSPORTATION ENERGY
AUTOMATION
ASK ABOUT OUR PRODUCT EVALUATION! 715 Stadium Drive, Arlington, Texas 76011
CONTENTS
March/April 2017 | Volume 15 | Number 2
FEATURES
opsy.st/ECDLinkedIn
10 2017 embedded wireless landscape: LPWA standards seek Industrial IoT connections
@embedded_comp
COVER
By Brandon Lewis, Technology Editor
14
Low-power, wide area (LPWA) networking standards are primed to connect a sizeable chunk of Industrial Internet of Things (IIoT) systems. As such, several technologies have emerged as viable options, including LTE Category NB-1, LTE Category M-1, LoRa, Sigfox, and others. The Silicon section of Embedded Computing Design March/April addresses the advantages and disadvantages of each.
WEB EXTRAS 12 Low-power, wide-area network standards, advantages, and use cases
18
By Dave Kjendal, Senet
14 The landscape of wireless things: Business dynamics, vertical markets,
Who’s behind the wheel of autonomous vehicle development By Shawn Andreassi, SAE International
By Kailash Narayanan, Keysight Technologies, Inc.
http://bit.ly/SAEautonomousdrive-2
18 RF testing: The basis for automotive V2X
How to get a handle on TrustZone for ARMv8-M software development
By Dr. Thomas Brüggen, Rohde & Schwarz
24 IoT platforms: The secret sauce for connected vehicle applications
By Rich Nass, Executive Vice President and Brandon Lewis, Technology Editor http://bit.ly/embeddedworld2017recap
LPWA technology, and industry challenges
embedded world 2017: Embedded Insiders show recap
By Diya Soubra, ARM http://bit.ly/HandlingARMTrustZonev8M
By Ron Felice, IBM
26 The OpenFog Reference Architecture: A baseline for interoperability in the IIoT cloud-to-things continuum
Roundtable interview with members of the OpenFog Consortium
5
High-res audio for IoT apps
TRACKING TRENDS
9
By Curt Schwaderer, Editorial Director
6
IOT INSIDER
By Brandon Lewis, Technology Editor
30 EDITOR’S CHOICE By Jamie Leland, Content Assistant
8
4
MUSINGS OF A MAKERPRO
Download the Embedded Computing Design app iTunes: itun.es/iS67MQ Magzter: opsy.st/ecd-magzter
COLUMNS Defining (artificial) intelligence at the edge for IoT systems
DOWNLOAD THE APP
AUTOMOTIVE ANALYSIS
Connected cars: eCall systems kickstart V2I agenda
Published by:
By Majeed Ahmad, Automotive Contributor
A MakerPro’s first foray into robotics By Jeremy S. Cook, Contributing Editor Embedded Computing Design | March/April 2017
2017 OpenSystems Media® © 2017 Embedded Computing Design All registered brands and trademarks within Embedded Computing Design magazine are the property of their respective owners. iPad is a trademark of Apple Inc., registered in the U.S. and other countries. App Store is a service mark of Apple Inc. ISSN: Print 1542-6408 Online: 1542-6459 enviroink.indd 1
10/1/08 10:44:38 AM
www.embedded-computing.com
TRACKING TRENDS
cschwaderer@opensystemsmedia.com
High-res audio for IoT apps By Curt Schwaderer More than ever, we’re using smartphones and Internet of Things (IoT) entertainment systems to consume audio. As the capabilities of these devices expands, so has demand for high-resolution audio for music streaming subscriptions. Companies are now exploring how to bring highdefinition audio to these resource-constrained mobile and IoT devices. The smartphone and associated music download sites have become one of the primary ways we consume music. MP3 compressed audio has been the common format due to the balance of audio quality and compression. However, as smartphones advance in memory and sophistication, there has been a shift toward high-resolution audio in these environments. Likewise, for in-home IoT applications, audio playback is a fundamental capability and lossless audio streaming is particularly important. Nigel Burgess, Principal Product Line Manager with Cirrus Logic, talked with me about the opportunities for highresolution audio in the mobile and IoT markets. Small device, lots of interference “Our objective at Cirrus Logic has always been to provide hardware and software solutions across the entire audio and voice processing chain and create a transparent audio path. One of the primary issues with highresolution audio involves clock jitter affecting the audio signal, especially within the smartphone and embedded environments,” said Burgess. “The Cirrus Logic 512 single-bit digital-to-analog converter (DAC) architecture is designed to reduce jitter that can cause audio performance degradation while filtering out unwanted noise to deliver clear audio. Digital audio passed through each of the pipelined 512 single-cell DACs provides sequential filtering out of unwanted audio. “Increased storage capacity and higher delivery bandwidths can now support high-resolution audio. It’s our job to make sure the hardware isn’t the limiting factor,” he added. In response to the high-resolution audio demand, Cirrus Logic has introduced two new product streams: ›› CS43130 SmartHIFI – Combines DSP processing and high performance without excessive battery drain (Figure 1) ›› CS4399 MasterHIF – Provides high-quality standard audio converters that do not require DSP programming www.embedded-computing.com
A number if interesting technical features of the CS43130 that bear mentioning: ›› Non-oversampling mode recreates the “old school” sound from non-oversampling amplifiers. There is a resurgence in the popularity of vinyl records and valve amps that produce a unique sound some audiophiles prefer. The original CD players used non-oversampling DACs, and over time that sound has been lost due to modern digital oversampling techniques. ›› Playback up to 32-bit and 384 kHz sample rates provides high-quality audio playback for stored and streaming content. ›› 130 dB dynamic range allows for quietest to loudest audio without reduction in quality. ›› Dedicated processing for DSD audio up to DSD256 Advanced Hi-Fi filters help system designs achieve the “right sound” by avoiding things like “pre-ringing” – for example, the waveform sound of a ringing cymbal may start before the cymbal is struck in the audio mastering. These filters can cancel out these issues for truer audio. ›› Up to 4x lower power. This is perhaps the biggest factor enabling high-resolution audio within the smartphone environment. The parts consume up to four times less power than competing parts.
FIGURE 1
Cirrus Logic’s CS43130 combines DSP processing with minimal battery drain.
Movement back to over-the-ear headphones Another interesting phenomenon is the movement toward larger over-the-ear headphones. It’s important that smartphones and embedded IoT devices drive these headphones properly. Software development kit Cirrus Logic has a development kit called WISCE that provides tools and utilities for a range of Cirrus Logic devices. The kit allows users to change register settings easily and make a smooth migration from prototype boards to production. There are also Linux drivers that plug into the Android solution for faster development and integration. Embedded Computing Design | March/April 2017
5
IOT INSIDER
blewis@opensystemsmedia.com
Defining (artificial) intelligence at the edge for IoT systems By Brandon Lewis, Technology Editor Merriam-Webster defines intelligence as “the ability to learn or understand or to deal with new or trying situations.” By that definition, “intelligent systems” on the Internet of Things (IoT) require much more than just an Internet connection. As more IoT devices come online, machine intelligence will increasingly be classified as a system’s ability to ingest information from its ambient environment via sensor inputs before taking immediate, autonomous action – even in the event that the readings associated with those inputs have not been encountered before or exist outside of the parameters defined by programmers during system development. To date, this type of intelligence has largely been the domain of computer vision (CV) applications such as object recognition, where a growing number of machine learning development libraries like Caffe, TensorFlow, and Torch have enabled the creation of models used in the training and inferencing of deep and convolutional neural networks (DNNs/CNNs). While artificial intelligence (AI) technology is thus far still in its infancy, its benefits for advanced driver assistance systems (ADAS), collaborative robots (cobots), sense-and-avoid drones, and a host of other embedded applications are obvious. However tantalizing, machine learning does come with drawbacks for embedded systems. First, the computational resources required to run a CNN/DNN have principally only been available in the data center. While arguments have been made that embedded devices could simply transmit findings to the cloud where large-footprint machine learning algorithm processing can occur, the latency requirements and transmission costs of doing so are essentially non-starters for many of the safety- critical applications mentioned previously.
6
with its Jetson TK1 development kit, and since then the company has continued to drive down the power consumption and development hurdles of super computing at the edge. Recently the Jetson ecosystem was extended again with the release of the Jetson TX2 module, an embedded board based on NVIDIA’s Pascal architecture that adds a dual-core Denver heterogeneous multi processor (HMP) along with double the memory and storage performance and capacity. The real achievement of the Jetson TX2 is in its energy efficiency, however, as Deepu Talla, Vice President and General Manager of NVIDIA’s Tegra product line explained in a March 6th press briefing that the TX2 provides a 2x improvement in that area over the TX1 (or less than 7.5 watts of total power consumption). An update of the JetPack software development kit (SDK) to version 3.0 also enables faster development of neural network-based systems through support for the Linux Kernel 4.4 and Multimedia API 27.1, which form the basis for exploiting TensorRT and cuDNN (deep learning); VisionWorks and OpenCV (computer vision); Vulkan and OpenGL (graphics); and libargus (media) libraries. But going even deeper into edge of AI is a processor vendor well known to the embedded space, as Xilinx officially unveiled its reVISION software stack on March 13th. reVISION addresses machine learning application, algorithm, and platform development for engineers working with Xilinx Zynq architectures, but does so in an extensible way that allows software developers to leverage C, C++, or OpenCL languages in an Eclipse-based environment when programming with frameworks such as Caffe or OpenCV. The stack also supports neural networks like AlexNet, FCN, GoogLeNet, SqueezeNet, and SSD.
Likewise, most, if not all, machine learning frameworks were developed to run on data center infrastructure. As a result, the software and tools required to create CNNs/DNNs for embedded targets have been lacking. In the embedded machine learning sense, this has meant that intricate knowledge of both embedded processing platforms and neural network creation has been a prerequisite for bringing AI to the embedded edge – a luxury most organizations do not have.
Taken cumulatively, this approach not only reduces required hardware expertise, but allows, for example, leveraging Caffe-generated .prototxt files to configure an ARMbased software scheduler that drives CNN inference accelerators pre-optimized to run on Xilinx programmable logic. According to Steve Glaser, Senior Vice President of Corporate Strategy at Xilinx, combining the reVISION software stack with FPGA technology enables low latency sensor inferencing and control down to 8-bit resolutions in a power envelope as low as the mid-3 watt range.
Embedded processor vendors enable AI at the edge 2014 saw NVIDIA take what is widely regarded as one of the first forays into embedded machine learning
These approach deeply embedded power benchmarks. Be on the lookout: True intelligence at the edge is upon us.
Embedded Computing Design | March/April 2017
www.embedded-computing.com
MUSINGS OF A MAKERPRO
www.youtube.com/c/jeremyscook
A MakerPro’s first foray into robotics By Jeremy S. Cook, Engineering Consultant In 2000, after getting pretty good grades in my first semester of engineering school, I was allowed to briefly* join the honors college. This meant that I got to take a much more interesting version of Engineering 102, where we were tasked with building robots to find their way around an obstacle course; You know, the kind of thing kids seem to do in middle or high school now. I had a great time, and it was my first experience programming with a development board, which was the Basic Stamp II from Parallax. After getting my own board months later, I got to work building my own robot, which was based on one I’d seen a vague description and pictures of online. It’s weird to think about now with the current abundance of YouTube videos and STEM programs, but this was the first time I’d seen this kind of robot – a simple hexapod with three servos – online. In this case, one servo caused the robot’s body to tilt left and right, while the other two made the front and back legs go forward and backwards in pairs. It was a simple but effective walking motion.
“THERE IS SOMETHING TO BE SAID FOR BEING ABLE TO ACTUALLY BUILD, OR AT LEAST UNDERSTAND HOW TO BUILD, MOST OF THE STUFF YOU DESIGN.” After a week or two of work programming and cutting legs out of wood and screws, I had something that worked, if just barely. I improved upon it over the next year or so, making it into something that actually worked pretty well, with a slow, interesting gait. Practical lessons learned Soon after building this contraption, I started on a co-op program (or extended internship) where I would be given access to true machine tools. Instead of thinking tolerances of 1/16" were quite good, I could now hold things to within a few thousandths of an inch. This made a huge difference, and after remaking some parts out of polycarbonate, the look and function of everything was much better. Another great lesson I learned, both from that job and searching the Internet, is that there is a whole world of fasteners and small components out there that can make
8
Embedded Computing Design | March/April 2017
the details of your project much better. Instead of bent paper clips for control rods, I started using threaded rods with eyelets made for model airplanes and the like. Instead of Phillips or flathead screws, I started using hex machine screws. There is a lot of stuff out there if you know where to look. There were many other things I leaned through this build process, but getting practical experience in industry helped with this build and throughout my career. There is something to be said for being able to actually build, or at least understand how to build, most of the stuff you design. Robotics today Today, things have changed quite a bit. I’ve built and set up many robots and robotic systems, both in industrial applications and for my own amusement. Personally, I’m not sure why I find this kind of thing so fascinating. Perhaps it’s the ability to create something that imitates in some small capacity the abilities of a human. Or maybe it’s the blinking lights and fun noises, or even the smell of new electronics. Who knows? On a broader level, robotics isn’t just in the realm of advanced university courses and industry, but is taught at a much younger age with programs like FIRST. Although I don’t necessarily think individual kids should be pushed into STEM fields simply for the sake of keeping up with whatever standard has been set, I think it’s great that people are given the opportunity to discover these interests at a young age. With outlets like YouTube, exposure in schools, and the availability of much cheaper development boards, opportunities for learning interesting technology are everywhere. After all, an Arduino clone development board can be had for a few dollars, and searching Google for “how to build a hexapod robot” yields 67,400 results as of this writing. It’s an exciting time to grow up, and it will be interesting to see what kind of robots the next generation of hackers and MakerPros come up with. * After getting fairly marginal grades my second semester, I graduated without completing the honors program. I didn’t realize it was my meticulous studying during the previous semester that had paid off – not just my inherent brilliance. Discovering Starcraft didn’t help either. So, final lesson, keep studying even if things look good! The process of learning isn’t always fun, but if you want to work with technology at a high level, you generally have to pass Calculus 2. www.embedded-computing.com
AUTOMOTIVE ANALYSIS
Connected cars: eCall systems kickstart V2I agenda By Majeed Ahmad, Automotive Contributor Vehicle-to-infrastructure (V2I) communications and its counterpart vehicleto-vehicle (V2V) initiative are lagging behind other advanced driver assistance system (ADAS) technologies. V2I technology employs car and infrastructure communications systems to warn drivers about road hazards, traffic jams, etc., and while industry observers acknowledge the critical significance of this technology for intelligent transportation systems, they also add that it’s not quite there yet. The billion-dollar question that has been haunting the V2I premise is, “Who will pay for the legions of sensors and cameras to be sprinkled all over highways and streets?” Governments or the private sector? And what about the return-on-investment? The upfront cost of furnishing roads with these technologies is mindboggling. Currently the electronics industry is looking for ways around these technology and business conundrums, with Audi, for instance offering an advanced traffic management system in several U.S. cities that will operate with select 2017 models. The Audi Traffic Light
Information service will monitor traffic lights via an onboard 4G LTE data connection and inform drivers about the remaining time until signals change (Figure 1). However, so far, the most prominent V2I use case has been emergency call (eCall) systems. In the event of an accident, eCall systems allow carmakers to automatically broadcast vehicle location via GPS and contact the nearest 24-hour emergency call center for help. Therefore, highly reliable connectivity and battery subsystems are imperative to ensuring that nearby emergency services are contacted when sensors initiate a call – even if a vehicle is involved in an accident just minutes after being parked for several months. In pursuit of automotive-grade reliability, Intersil recently unveiled a battery charger IC that extends the life of Lithium Iron Phosphate (LiFePO4) batteries commonly used in vehicular eCall systems. The ISL78693 is 3.6V single-cell battery charger that prevents rapid discharging of a car’s backup battery during an accident or when a vehicle has been parked for an extended period of time. One of the key requirements of a battery charger IC serving eCall systems is low leakage current, which allows the eCall backup battery to remain charged for a longer period of time. Here, Intersil’s chip boasts battery temperature monitoring and 3.6V low output voltage to safeguard and extend the life of popular LiFePO4 batteries. The battery charger IC also offers a charge current thermal foldback feature, which prevents overheating by automatically reducing battery charging current. Now that Europe has mandated the eCall systems for all new cars by 2018, expect more silicon solutions that further push the V2I agenda later this year.
FIGURE 1 Audi’s V2I service is designed to maximize the number of green lights while optimizing vehicle speed and route.
www.embedded-computing.com
Embedded Computing Design | March/April 2017
9
SILICON: WIRELESS LANDSCAPE AND MODULE REPORT
2017 embedded wireless landscape: LPWA standards seek Industrial IoT connections By Brandon Lewis, Technology Editor
Shortly after acquiring semiconductor IP vendor ARM last fall, Softbank CEO Masayoshi Son predicted a future with 1 trillion connected devices. Although it’s unclear exactly how Son reached that number, what is undeniable is that, however many devices are eventually connected, the majority will belong to the Industrial IoT (IIoT).
M
eanwhile, telecom companies in the U.S. and elsewhere have begun planned shutdowns of legacy 2G cellular networks that have served industrial machine-to-machine (M2M) applications for more than a decade, leaving a void in the IIoT connectivity landscape. This void has created an opportunity for low-power, wide-area network (LPWAN) technologies that – while not suited for the latency, determinism, or quality of service (QoS) requirements of high-value or safetycritical systems – do offer the low cost and simplicity necessary to connect scores of devices and applications that could not afford to be networked previously. While not mandatory for inclusion, many of today’s LPWAN solutions operate in unlicensed spectrum at sub-GHz frequencies where they target low data rate (bits per second throughput), non-real-time systems in markets such as agriculture, utilities, and mining that require ten or more years of battery life. Given the greenfield potential of never-before-connected devices in these sectors, Lee Ratliff, Principal Analyst for Connectivity and IoT at IHS Technology says that standards such as LTE Category-M1, LTE Category-NB1 (formerly NB-IoT), LoRa, Sigfox, and other proprietary implementations have emerged as competitors in a race to the bottom of IIoT networking. “Up until now, the only solutions the IIoT has had essentially are 2G and 3G cellular M2M, which are fairly expensive solutions, and not just in terms of the initial cost,” Ratliff says. “Today 2G has gotten down to $5-$7 in terms of module cost, but there’s provisioning, SIM cards, usually a technician truck roll to get these things going, and then there are ongoing service costs. The LPWAN standards promise to reduce all of that by orders of magnitude.” Of the major technologies currently occupying the LPWAN space, Ratliff identifies LoRa and Sigfox as market leaders, with LTE standards lagging due to late moves
10
Embedded Computing Design | March/April 2017
towards standardization. Of the alternatives, each looks to address one or more of three technology use cases, as Will Yapp, Vice President of Business Development at LoRa network service provider Senet Inc. explains. “On the technology side, we break the total addressable market for LPWAN into three different technology use cases: uni-directional communication, which is just the device communicating up to the network; bi-directional, which is some command and control over the endpoint as well as having it communicate up to the network; and then devices in motion or mobility, which are things like asset tracking and fleet tracking,” Yapp says. “LoRa was the only one of Sigfox, Ingenu, and LoRa that covered all three of those,” he continues. “Sigfox is predominately a uni-directional solution. Ingenu is both uni-directional and bi-directional but doesn’t address devices in motion, and the 2.4 GHz range that it uses is limited www.embedded-computing.com
SILICON: WIRELESS LANDSCAPE AND MODULE REPORT
in terms of range and overall interference issues, especially when you get into cities and smart buildings. LoRa gave us the biggest, broadest market and the most access to technology going forward. “LoRa also gave us the best chance of limited interference on the 900 MHz ISM band, as long as you don’t have two 915 MHz base stations sitting right next to each other,” Yapp says. “There’s a very low likelihood that that spectrum will ever be reallocated. As a matter of fact, governments have been using 900 MHz since the creation of radio access technology.” Service providers like Senet represent an interesting development in the communications market, as the lack of timely 3GPP standards enabled them to build out their own infrastructure and adopt similar revenue models around monthly data usage. Senet, for example, deploys 8- to 64-channel LoRa base stations capable of scaling from localized urban deployments to supporting millions of devices over geographies of up to 1,000 square miles each. “We’re the only network service provider that has implemented a network control system that allows us to implement devices that can be handled by any of the gateways on any of the channels and do some adaptive data management across those channels,” Yapp adds. “It gives us a great sense of scale.” Narrowband LTE: The elephant in the room One force working against technologies like LoRa is the affinity of major carrier for 3GPP standards, which Ratliff speculates could drive them to throw their weight behind LTE Category-M1 and -NB1 technologies once they are both fully off the ground (LTE Category-M1 is being rolled out, Category-NB1 is scheduled for initial deployment at the end of this year). In this regard, a significant advantage of both narrowband LTE technologies is that a worldwide network infrastructure www.embedded-computing.com
capable of supporting them is already in place, which Craig Miller, Vice President of Worldwide Marketing at Sequans Communications says enables “just add water” software updates to get the new LTE standards up and running. “It’s an existence issue, and LTE exists,” Miller says. “Couple that with telco qualities, the five-nines reliability, security that’s built into the network systems, the authentication, the SIM authentication regimes. These are proven and robust technologies. You’ve also got the longevity assurance of these networks. LTE players have spent billions of dollars on spectrum and network deployments to cover their geographies. The long-term availability of those networks is not in question. “Now along comes LTE Category-M1 that can use that same network,” he continues. “You don’t have to build any new towers, no new backhaul. I just have to do a software upgrade to my base stations to enable those narrowband channels and schedulers. It’s a pretty low cost, low investment to get this new stuff onto the existing network. Everywhere Verizon radiates LTE today, they will radiate Category-M1 by the end of March. That’s how fast they can deploy. Both Category-M1 and -NB1 devices are capable of higher throughput than competing LPWAN technologies, peaking in the 1 Mbps and 250 Kbps ranges, respectively. Sequans Communications’ Monarch SX system on chip (SoC) integrates either Category-M or -NB1 baseband and RF transceivers alongside an ARM Cortex-M4 core, graphics processing unit (GPU), media processing unit, sensor hub, display controller, and embedded RAM and power management. It might all come down to cost One way that the 3GPP’s narrowband IoT LTE standards have been able to compete with the likes of LoRa, Sigfox, and others is by reducing the number of antennas from two to one, which saves complexity and thus cost. Still, in such a competitive market, “you’re looking at the very best Category-M1 module pricing being $10-$12 and the very best Sigfox pricing being $2,” Ratliff says. “Any way you slice it, you’ve got a factor of five there.” Miller correctly points out that each option serves certain use cases, with Category-NB1 and Category-M1 solutions each moving further down the predictability, reliability, and determinism continuum of risk from other LPWAN technologies. As so often is the case, this leaves device manufacturers with the task of selecting the most costeffective solution for their application needs. Regardless of technology choice, however, the emergence of LPWAN technology is set to unlock value in the industrial space that has yet to be seen. For everyone in the IIoT industry, Ratliff says, that’s a win. “To enable 1 trillion devices you have to have connectivity solutions that cost pennies. Pennies,” Ratliff explains. “Because cellular M2M is expensive, industry has only been deploying it in applications where the data is valuable. But there’s a whole host – the other 90 percent of industrial applications – where the data is just not that valuable. A $1 or $2 module solution available on the market could open those up. There are tens of millions or a hundred million of cellular M2M connections today, but technologies like LPWAN that take the cost down by another order of magnitude could jump that up from 100 million units to … a billion? Two billion? They’re going to enable all of these applications.” Embedded Computing Design | March/April 2017
11
SILICON: WIRELESS LANDSCAPE AND MODULE REPORT
Low-power, wide-area network standards, advantages, and use cases By Dave Kjendal
W
ith technology advancements and market opportunities of such significance, one of the primary challenges introduced by the IoT is how to best collect, transport, filter, and consume the vast amount of raw data that will be generated and to apply it to the differing needs of commercial, industrial, and civic IoT programs. Faced with this critical, foundational task, decision makers should start by asking themselves one fundamental question: Is it worth paying to collect and transmit data that is more detailed than I need? If the answer is “no” then you should consider using a low-power, wide-area network (LPWAN) for your IoT initiatives.
12
The Internet of Things (IoT) may represent more innovation and change than any other technological development in our lifetime. By instrumenting the world with low-cost sensors, consumers, businesses, cities, and entire countries have the opportunity to change the way value is created by providing ground truth data to transform outcomes, markets, and economies. Connecting the Internet of Things The emergence of LPWANs has fundamentally changed the IoT landscape. LPWANs are designed for sensors and applications that need to send and receive small amounts of data over long distances a few times per hour, or maybe only once a day. By collecting and transmitting only the data that is needed to optimize specific applications or operations, LPWANs offer value that cannot be achieved with other network technologies, including: ›› Substantially lower cost in both capital and operational expense ›› The ability to deploy sensor solutions with an extended lifespan of more than 10 years
Embedded Computing Design | March/April 2017
›› Long range wireless coverage with a small number of gateways ›› Accelerated time-to-market for end solution deployments The market opportunity for LPWA-based solutions is rapidly emerging and is quite significant. Because of the unique characteristics of LPWA technologies, of the 50 billion devices estimated to be connected to the IoT by the end of 2021, it is expected that more than 60 percent of these devices will be connected with LPWANs[1]. LPWA communications: Standards, differentiation, and deployment models The gap left by high-cost, high-function cellular and low-cost, localized Wi-Fi and www.embedded-computing.com
Senet
www.senetco.com
TWITTER @senetco
www.linkedin.com/company/senet
YOU TUBE
www.youtube.com/user/senetco
SILICON: WIRELESS LANDSCAPE AND MODULE REPORT
Bluetooth connectivity will be dominated by public LPWANs, and not surprisingly, LoRaWAN has emerged as a leading LPWA communications technology for IoT by virtue of its open ecosystem and technical superiority. LoRa networks are built using open standards, which provides a broad vendor community to support applications – an essential aspect of driving the adoption of any successful network technology. This open ecosystem is instrumental in overcoming markets’ natural resistance to new technologies, while technical advantages allow LoRaWAN to address more use cases than legacy networks and competing LPWA solutions. Cellular networks, for example, are built for the needs of smartphone users, delivering faster connections to support more data. While the idea of piggybacking off of an existing cellular infrastructure may seem appealing, the price to keep up with the evolution of cellular technologies doesn’t make sense for most industrial and commercial IoT applications. Wireless technologies such as Wi-Fi, ZigBee, and Bluetooth are also available for IoT applications, but these legacy technologies are characterized by short-range, low-power communication capabilities, thus restricting their usage to limited coverage areas. As compared to proprietary LPWA networks, networks supporting the LoRaWAN protocol foster unparalleled business continuity and deliver extreme flexibility. Selecting LoRaWAN not only provides application portability and network provider choice, but its operation can be conducted on a public network, in a semi-private style, or completely private depending on the market and application requirements. In addition to the benefits of LoRaWAN being an open standard, there are critical security, reliability, and scalability benefits provided by the architecture. The LoRaWAN protocol was designed with end-to-end security as a fundamental element of the architecture. Communications on the network between end nodes and the application server are www.embedded-computing.com
secured with AES-128 encryption. This end node ‘VPN-like’ service ensures data integrity and security for sensitive application data. Network reliability is supported by the ability to deploy redundant gateways in a very cost-effective manner, minimizing communication disruption in the event of a localized outage. In addition, end nodes are capable of sending readings more often than required and can resend messages to increase reception outcomes. Messages can also be sent requiring acknowledgements, assuring the desired outcome is achieved. Scalability is key to any commercial and Industrial IoT (IIoT) solution deployment. The coverage provided via LoRaWAN gateways can easily be right sized for the desired application. Commercially available Macro Cell Gateways provide 10-20 miles of wireless coverage and connectivity for tens of thousands of devices per gateway, while Micro Cell Gateways provide 1-2 miles of coverage and hundreds of devices per gateway. Through flexible and scalable deployment options, LoRaWANs deliver the coverage and scalability necessary for a diverse set of monitoring and control applications across agriculture, supply chain, asset tracking, smart city, and other demanding markets and use cases. Smaller Pico Cell Gateways are designed to support residential or commercial applications and provide coverage for tens of devices per gateway. This option is ideal for a range of intelligent building applications and even consumer solutions such as home automation and assisted living. Applications and use cases LPWANs are playing an important role in connecting a range of devices that require features such as low power consumption, low cost, and long battery life. As noted above, these networks are ideal for a diverse range of industries, such as agriculture, utilities, oil & gas, and smart buildings/smart cities, and cover a range of applications and deployment scenarios that cellular and short-range wireless network technologies are not suited for.
Another interesting aspect of LPWAN solutions is the ability to cost-effectively support the efficient functioning of components and systems that lead to opportunities that will massively impact resource conservation, waste reduction, safety, and security. Waste management, for example, is growing in importance for ecological sustainability in many countries. LPWAN solutions can help with streamlining the collection process (i.e., collect bins only when full), route optimization, and resource planning. LPWAN smart metering and resource management solutions represent a paradigm shift in understanding energy consumption and conservation at all the levels of upstream and downstream operations. Also, supporting the efficient functioning of all components and systems of a building, such as lighting, safety and security, emergency systems, HVAC, and other systems, can enhance public safety and deliver energy savings with marginal human involvement and cost. The strategy of partnerships, alliances, and collaboration is the approach adopted by industry leaders to achieve growth in the LPWAN market. In March 2015, the LoRa Alliance was incorporated as a non-profit to maintain and advance the LoRaWAN specification. The LoRa Alliance has grown since that time from 20 founding members to more than 400 today. The LoRaWAN specification provides an open, secure protocol definition for the operation of end devices and the networks that support them. The openness of both the specification and the LoRa Alliance has nurtured an ever-growing ecosystem of technology and solution providers that is bringing a building momentum to IoT deployments around the globe. Dave Kjendal is Chief Technology Officer at Senet. References: 1. MarketsandMarkets. “Low Power, Wide Area Network (LPWAN) Market – Global Forecast To 2021.”
Embedded Computing Design | March/April 2017
13
SILICON: WIRELESS LANDSCAPE AND MODULE REPORT
The landscape of wireless things: Business dynamics, vertical markets, LPWA technology, and industry challenges By Kailash Narayanan
T
The phenomenon of ubiquitous connected sensors providing key data and further processing of that data in the cloud to deliver business insights presents a huge opportunity for many players in the electronics and software industry.
he Internet of Things (IoT) is less about wireless radios and technologies. It is more about the combination of key data transmitted by sensors and the post-processing of that data to deliver key business insights, resulting in efficiency and savings. There is now enough compute horsepower in a tiny ARM processor to make things more powerful than desktop PCs in the early 1990s, especially with unlimited, almost free cloud-based computing resources. This phenomenon, and advances in software, are enabling breakthroughs in machine learning and artificial intelligence (AI) to perform that post-processing. Machine learning and AI are at the root of almost every money-making value proposition in IoT, as a thriving IoT relies on four crucial items: 1. A viable business model 2. A robust connectivity topology, 3. Reliable sensors 4. AI techniques to extract and deliver the right insights Business dynamics of wireless data The path from 3G to 4G to 5G data services for an operator is a path to becoming a converged faceless utility: the average revenue per user (ARPU) declines yearover-year; data consumed has risen exponentially; the revenue per MB has declined
14
Embedded Computing Design | March/April 2017
exponentially. Therefore operators are faced with making huge investments in infrastructure to support declining revenue streams, and over-the-top (OTT) providers, aspirations of big c ompanies, and merger and acquisition (M&A) activity will further challenge operators as each has the potential to introduce services that will find users the cheapest connection on a minute-by-minute basis. Hence the motivation for low-power, widearea (LPWA) IoT technologies. There are hundreds of applications for which operators can charge from pennies to dollars per month. The data volume associated with these applications is tiny so the load on existing infrastructure is small. In the future it can be expected that operators will earn most or all of their profit from these new services based on LPWA technologies. www.embedded-computing.com
Keysight Technologies, Inc. www.keysight.com
TWITTER @Keysight
www.linkedin.com/company/ keysight-technologies
www.facebook.com/ Keysight
GOOGLE PLUS
www.plus.google.com/+keysight
YOU TUBE
www.youtube.com/user/ keysight
SILICON: WIRELESS LANDSCAPE AND MODULE REPORT
IoT technologies and applications LPWA networks are designed to support machine-to-machine (M2M) applications with low data rates that are able to sustain long battery lives and operate unattended for years. There are many LPWA technologies, including NB-IoT, LoRa, and Sigfox. LTE Category-NB1 (formerly NB-IoT) is governed by 3GPP standardization and has widespread operator backing and infrastructure support. NB-IoT will provide an alternative technology for utility meters, parking bay sensors, fire detection, security, etc. (Figure 1). All of these industries have existing networking technologies so NB-IoT won’t be transformative, but rather an enabler to make them more ubiquitous. LPWA technologies change the business model for companies that want to stay connected to products in the field. Today, larger capital assets have cellular modems in them for usage tracking and predictive maintenance, however the usage of lower cost products is not easy to cover. NB-IoT in particular offers the potential for every product to be shipped with a built-in tag that will allow manufacturers to track usage, status, and gain insight for continuous
“THE NUMBER OF APPLICATIONS THAT WILL BENEFIT FROM LPWA TECHNOLOGY ARE COUNTLESS, AND INCLUDE STAKEHOLDERS FROM AN INDUSTRIAL BATTERY VENDOR WHO WANTS TO TRACK USAGE OF THEIR BATTERIES WITHOUT THE COLLABORATION OF THE VEHICLE MAKER (FORKLIFT) TO THE END USER (WAREHOUSE).” product improvement and marketing, and do so in a way that doesn’t require network collaboration with an end user. As there is currently no way to find and track products that are switched off without the collaboration of customers or installing a suppliers’ wireless network at the customer, NB-IoT tags are positioned to solve a significant challenge. The number of applications that will benefit from LPWA technology are countless, and include stakeholders from an industrial battery vendor who wants to track usage of their batteries without the collaboration of the vehicle maker (forklift) to the end user (warehouse). Influence of IoT on vertical industries The IoT phenomenon has had a multi-dimensional influence on vertical industries in terms of quality of life and commercial value. From a quality of life perspective, the promise of IoT is to take a big step in closing the gap between providing comfort and the efficiency, which required significant human NB - IoT
FIGURE 1
$ High Density
Low Cost
•Devices/Cell: 10,000+
•Modules < $5
•Data Rates: 10s of kbps •Connection Frequency: Low
•Basic Features •Reliable, Stable
www.embedded-computing.com
Superior Battery Life •Up to 10+ Years •Enhanced Sleep Modes
Extreme Coverage
Software Upgrade
•+20 dB Compared to GPRS
•To Existing RAN Infrastructure
•Remote Areas
•Global Technology Standard (3GPP)
LTE Category-NB1 (formerly NB-IoT) defines an alternative networking technology for low-bandwidth Industrial IoT (IIoT) applications that will enable ubiquitous, low-cost connectivity.
Embedded Computing Design | March/April 2017
15
SILICON: WIRELESS LANDSCAPE AND MODULE REPORT
resource in the past. By knowing ever more about us, companies will be able to surround us with helpful intelligence, orchestrating and managing every detail of our lives – banking, entertainment, commuting, purchases, healthcare, education, travel, and childcare. As a result, we are likely to see fewer advertisements for things we do not care about and spend less time searching for the things we do. In commercial value proposition dimension, IoT enables energy and resource utilization efficiency, and there are hundreds of value-add opportunities in many vertical industries (Figure 2): ›› ›› ›› ››
Truck driver feedback to save fuel Finding an empty parking space Predictive rather than prescriptive maintenance Self-learning production lines to optimize energy, minimize scrap, and improve yield ›› Optimizing wind turbines and other energy assets ›› Supply chain management (know where every item is in your supply chain at all times)
The whole point of IoT is to take people out of process loops. The key is to look at end-to-end processes and remove people from having to go through process steps through the use of wireless sensors enabled by data analytics and machine learning.
Many of the propositions mentioned can be delivered as services, enabling a whole new set of service companies. IoT enables the transition from capital expenditures (CAPEX) spending to operational expenditure (OPEX) spends, more aptly aligning with the sentiment, “I don’t want to own a music player, I just want to listen to music now.” As the world moves to an outcome-based frame of mind, vertical industries will need to rethink their business models. We will see nearly all capital supply industries transform to service industries – e.g., self-driving cars, renting heavy-duty goods, etc. The outcome economy also opens a third budget – not CAPEX or OPEX, but savings. For example, IoT companies targeting the mining industry might be able to reduce fuel costs and accidents resulting from operator fatigue, allowing the IoT company’s business model to be based on a royalty percentage of the savings achieved in mining operations. That is a win-win.
Wearable Devices
-Smart Watch -Wearable Glasses -Smart Bands -E-textiles -Hearing Aids -Health Monitors
FIGURE 2 16
Smart Home
-Controller -Appliances -Sensor & Security -Climate Control -Lighting -HVAC -Audio/Video
Connected Car
• V2V/V2X/V2I • eCall • Infotainment • GNSS • Autonomous Driving
There is an argument that all commerce will move to outcome-based transactions, as no customer will want, or need, to take a risk that a product won’t deliver on its promised return. In the future, customers will purchase a result and the supplier will take risks based on whether they can deliver the outcome. This will open up a new type of financial market where suppliers will insure against the risk of not being able to meet promised outcomes.
Industry challenges Industry challenges center around coming up with a viable business model, time to market, speed to revenue, and deploying reliable sensor networks. Sensor design challenges involve tracking evolving standards, device miniaturization, and signal integrity issues, as well as requirements around long battery life (7-10 years without battery change) and certification. From a technical perspective, superior battery life in the range of several years
Smart Grid, Smart Energy
Smart City, Automation
-Smart Meter -Smart Appliances -DC Distribution
-Sensor Network -Smart Machine -Surveillance Camera -Asset Tracking -Wi-Fi -Backhaul
The Internet of Things (IoT) value proposition varies based on vertical market.
Embedded Computing Design | March/April 2017
www.embedded-computing.com
is one of the design targets for all IoT device makers. They need to characterize device battery life and current drain under realistic operating conditions. Sensor makers also need to address signal integrity issues such as crosstalk as tighter interconnects become more prevalent with miniaturization.
Volume
Technology
Application
Good
Wi-Fi
High data rates, limited coverage
LTE
High data rates, national coverage
Bluetooth and Bluetooth Low Energy (BLE)
Low data rates, low power, narrow coverage
LPWA, NB-IoT, and Category-M
Low data rates, ultra-low power, and national coverage
From a business perspective, challenges vary depending on the IoT application. While cost is a major factor in consumer IoT applications such as wearables and home automation, Industrial IoT (IIoT) applications such as smart grids, connected cars, and transportation require unfailing reliability, longevity, security, and the ability to operate devices with little or no human intervention.
Low
zigbee, Thread
Limited coverage, home automation
Wi-SUN
Low data rates, wide coverage, meter readings
TABLE 1
Several wireless formats are available to Internet of Things (IoT) device manufacturers, each with fundamental advantages and disadvantages.
Today there is a plethora of wireless formats available for IoT designs, which is not uncommon during periods of rapid innovation in a new space. Table 1 outlines high-level attributes for the industry to consider when selecting a wireless format, but is by no means an exhaustive coverage of all IoT radio formats. Market outlook IoT is likely to be a significant enabler of many disruptive business models and market efficiency. Although we haven’t seen operators talk seriously about service models yet, a future seems likely in which operators could offer a box of tags to put on our bikes, cars, pets, kids, lawnmowers, and car keys powered by a ‘Find my X’ application for an extra $1 to $5 per month on our smartphone plans. This would equate to a roughly 25 percent increase in operator ARPU, and may ultimately be where really big IoT device volumes start being generated. As the world moves to an outcomebased economy, it is inevitable that IoT powered by machine learning and AI will become a key enabler in connecting multiple vertical industries on a real-time basis, providing efficiency, savings, and flexibility that benefits both suppliers and consumers. Kailash Narayanan is Vice President and General Manager of the Wireless Devices Segment at Keysight Technologies, Inc. www.embedded-computing.com
Embedded Computing Design | March/April 2017
17
SILICON: WIRELESS LANDSCAPE AND MODULE REPORT
RF testing: The basis for automotive V2X By Dr. Thomas Brüggen
Automated vehicles can safely navigate the road only if they have precise knowledge of the environment and traffic situation. A wide variety of sensors and cameras already provide some of this information. Additional information is supplied by using wireless technologies to connect vehicles. To ensure that safety-related data messages are received even under poor transmit conditions, the transmitter and receiver must adhere to minimum standards. Adherence can be verified using RF tests.
R
oad safety has improved steadily as a result of many inventions. Mechanical systems such as safety belts and airbags as well as electronic safety equipment such as anti-lock brake systems (ABSs) or electronic stability programs (ESPs) have decreased the number of accidents and their consequences over the decades. Over the last few years, however, there has been no further significant reduction in the number of persons involved in accidents. According to the German Federal Statistical Office, this number remains at around 390,000 annually. In order to further significantly reduce the number of accidents, new technologies
18
are needed. Critical traffic situations can be detected before they occur through the wireless exchange of information between vehicles (vehicle-to-vehicle (V2V) communications), as well as with the traffic infrastructure and all traffic participants (vehicleto-everything (V2X) communications). For example, all vehicles that drive through a crossing can exchange information about speed and direction. This makes it possible to detect potential collisions, issue appropriate warnings, and autonomously initiate early countermeasures. However, this scenario can only become a reality if there is a reliable wireless exchange of information between the vehicles, even under poor transmit conditions and without line of sight. If a single piece of information is missed, then one or more of the vehicles will gain an inaccurate view of the actual situation, conceivably with deadly consequences. Possible interference Wireless communications systems can be affected by different types of interference. Many of these are known collectively as fading. This includes shadowing and interference caused by physical effects such as scattering, diffraction, refraction, and reflection, which cause multipath propagation of the signal (Figure 1). In other words, multiple
Embedded Computing Design | March/April 2017
www.embedded-computing.com
Rohde & Schwarz
www.rohde-schwarz.com/us/
@RohdeSchwarz
www.linkedin.com/company/ rohde-&-schwarz
www.facebook.com/ RohdeAndSchwarz
GOOGLE PLUS
www.plus.google.com/ 108183321348713191769
YOU TUBE
www.youtube.com/user/ RohdeundSchwarz
SILICON: WIRELESS LANDSCAPE AND MODULE REPORT
versions of the same signal are received at the antenna at different times and with different signal levels and interference. Superpositioning of the individual signals at the receive antenna can distort, attenuate, or even cancel out the received signal. Traffic participants are also continuously moving, which adds a time variant to the fading scenario. All receivers inside vehicles are confronted with continually changing conditions and signal quality. Safety-critical applications must function reliably under these conditions. If a receiver cannot handle fading with the time variant, then it will not be able to detect and process the received signal at the antenna. Strong coding and specialized protocols cannot compensate for the loss of the receive signal. Safety-relevant information is lost. This can represent a considerable safety risk, especially when drivers have come to rely on warnings from a V2X system and do not expect it to fail. Testing physical transmission To minimize the safety risk arising from poor transmit conditions, the RF transmitters and receivers found in on-board units (OBUs) and road-side units (RSUs) of communications systems must exhibit certain characteristics. Developers and users wanting to integrate V2X components into their system can use RF tests to verify these requirements. The two lowest layers of the Open Systems Interconnection (OSI) model factor into these tests because they are responsible for the physical transmission of the data message (Figure 2): ›› The physical layer (Layer 1) is responsible for the physical transport of the data via a transmission medium. In the case of V2X communications, data is wirelessly transmitted via the air interface. This layer uses specific modulation modes, carrier frequencies, and bit rates. Often the quality of the channel over which the data is transmitted is also taken into consideration. ›› The data link layer (Layer 2) is made up of two components: the medium access control (MAC) layer and the logical link control (LLC) layer. The MAC layer regulates access to the transmission medium for multiple subscribers (this is relevant for RF measurements). The LLC layer handles tasks such as error detection and correction at the protocol level. In contrast, tests at the protocol level (i.e., from the LLC layer up to the application layer (Layer 7)) are not suitable for verifying RF characteristics. These tests check that the bit stream, which is generated in the LLC layer from the received signal, is processed correctly. The RF signal at the receive antenna is ignored. As a result, all tests at the protocol level rely on one important prerequisite: that the signal with the message has been received at the vehicle antenna and converted into a bit stream. It is acceptable for the bit stream to contain bit errors, but only as many as can be corrected using channel coding. For error-free processing of the bit stream, the message www.embedded-computing.com
FIGURE 1
Example of fading due to multipath propagation without a line of sight path.
Layer
Name
7
Application Layer
6
Presentation Layer
5
Session Layer
4
Transport Layer
3
Network Layer
2
Datalink Layer
1
Physical Layer
FIGURE 2
Consists of
Logical Link Control (LLC) Medium Access Control (MAC)
The physical layer and logical link control layer of the Open Systems Interconnection (OSI) model are critical to RF testing of V2X components because they are responsible for the physical transmission of data messages[1].
must appear in the application layer precisely as it was received at the OBU antenna. This triggers additional actions, such as a warning message on the display or automatic braking. The RF module (i.e., the MAC layer and the physical layer) in the OBU must therefore meet certain minimum requirements with respect to power and frequency accuracy and packet error rate (PER)[2]. In addition, the transmitted signal must not interfere with any of the transmission technologies in the adjacent frequencies[3]. These characteristics can only be verified with RF tests and not protocol tests because any interference in the transmitted signal is conducted to the receiver via the OBU’s RF module. But how can RF module requirements be tested? And how can it be ensured that a transmitted message is actually received? A look at the mobile communications industry shows that three different types of RF tests are used to validate and certify smartphones: ›› Regulatory tests check the interference that the transmitsignal causes in other frequencies to ensure that it stays within a specified limit. Typically, country regulatory agencies specify these limits, and adherence is legally binding. These types of specifications are already available for V2X units. Embedded Computing Design | March/April 2017
19
SILICON: WIRELESS LANDSCAPE AND MODULE REPORT
›› Conformity tests ascertain whether a smartphone meets the RF specifications of the technology it uses to transmit. For example, smartphones must not exceed a specified maximum PER or maximum transmit power. A separate test specification often describes how to perform and evaluate these tests. ›› Some mobile communications providers mandate stricter or additional tests beyond those described here to differentiate themselves from the competition by providing better transmission quality and better network reliability. Mobile devices must meet these special requirements in order to be allowed into the network. Radio over cable The automotive industry tests automotive components and electronic control units (ECUs) in the lab, on testing grounds, and even on public roads. For wireless communications this is the equivalent of field tests, offering a realistic environment for RF tests. However, other influences such as the weather can unpredictably change the RF characteristics of the radio link. The test setup and test sequence depend on the vehicles involved and the antenna locations, and often can only be changed with considerable effort. This is not practical for testing a new device that is still in the development stage. What is needed instead are alternatives that permit realistic testing in the lab. In wireless communications, conducted tests represent an alternative to field tests. A test system simulates the transmission channel while a cable replaces the actual radio link. These types of RF tests can be performed for each prototype and each time the software or hardware is changed. This has many advantages that make development faster and less expensive and reduce errors: ›› The tests can be performed at any time and at low cost. ›› The test conditions are clearly defined at all times and can be changed at any time irrespective of outside influences. ›› Clearly defined test sequences, when performed under the same conditions, lead to comparable results. ›› Reproducible and comparable results facilitate debugging. ›› Parameters can be easily modified. This is in contrast to the great deal of effort required to modify the fading profile in a field test, for example. ›› Multiple tests can be combined into test sequences and then performed automatically. This makes it possible to run long-term tests to study the reliability of a prototype. ›› Certain RF tests such as error vector magnitude (EVM) or receiver (RX) sensitivity tests only make sense as conducted tests. In a field test, uncontrolled noise and interference from external sources falsify the measurement results. Depending on the selected scenario, channel simulation exactly simulates the physical attributes of the radio link. Today’s signal generators can also simulate the special V2X fading profiles in real time.
20
Embedded Computing Design | March/April 2017
Examples of RF tests that verify the transmit characteristics (TX) of an OBU or RSU Frequency Accuracy Modulation Accuracy Out-of-Band Emissions Transmission Power Level Spectrum Emission Mask Spurious Emissions
FIGURE 3
RF transmit test specifications include in-band and out-of-band testing.
Examples of RF tests that verify the receive characteristics (RX) of an OBU or RSU Adjacent Channel Rejection Non-Adjacent Channel Rejection Decentralized Congestion Control Out-of-Band Emissions When Transmitter is Off Performance with Fading (Packet Error Rate) Sensitivity
FIGURE 4
RF receive test specifications include in-band and out-of-band testing.
Field tests still make sense, especially for antenna measurements (e.g., for determining antenna characteristics or for beamforming tests). Although conducted tests cannot completely replace field tests, channel simulation can simplify testing, easily and cost-effectively supporting development in the lab. Detecting RF problems To be able to compare the test results of the various hardware and software versions of a V2X unit, all test procedures must be clearly defined. Some countries have therefore defined test specifications for V2X systems that include test cases in four categories (Figures 3 and 4): ›› RX in-band: The test cases in this group test the transmitter (TX) characteristics, for example maximum and minimum transmit power, frequency accuracy, and modulation accuracy. ›› TX out-of-band: The unwanted transmit power outside of the allowed frequency band must not disrupt other technologies. TX out-of-band test cases therefore measure this transmit power and compare it against the allowed limit. ›› RX in-band: This category tests the receiver (RX), for example by measuring the lowest receive power at which the received signal can still be decoded or by measuring performance with fading. Figure 5 shows a screenshot of an R&S SMW200A vector signal generator with a configured V2X fading profile. ›› RX out-of-band: Specialized test cases measure whether the OBU or RSU unintentionally emits transmit power in other frequency bands when the transmitter is switched off. www.embedded-computing.com
FIGURE 5
The Rohde & Schwarz SMW signal generator showing a fading profile for V2X at 5.9 GHz.
Various plug tests for V2X have shown that the TX out-of-band and fading tests are especially problematic for many devices under test (Figure 6)[4]. It is possible, however, to detect these RF problems during the development phase by using appropriate test instruments. The RX tests can be performed with a signal generator that is capable of generating a V2X signal. A signal analyzer covers the TX test cases. Depending on the dynamic range of the analyzer, a filter for the V2X signal is needed to cover the broad frequency range of the out-of-band measurement.
FIGURE 6
TX out-of-band test: The transmit power (blue line) of an 802.11p unit exceeds the allowable limit (red line) at multiple points. The frequency range between 5855 MHz and 5925 MHz is reserved for V2X in Europe and in the U.S.
At present, various wireless technologies are under discussion for implementing V2X communications, in particular WLAN 802.11p, LTE, and 5G, which will be available several years from now. Regardless of which technology is used, test and measurement equipment manufacturers such as Rohde & Schwarz already offer the test solutions needed for V2X. For example, solutions based on widely distributed LTE technology can be
Shock, Vibration, Fretting Corrosion Do you want crucial data transferred to be Resistant to it or Immune to it? Smiths Interconnect Nebula, KVPX, and KA Series: Proven to Perform in ultra rugged conditions.
•
• •
SATA 6.0 Gbps, meets footprint of std COTS connectors VPX compatible to VITA 46 & 48 standards KA series MILDTL-55302 Compliant Connectors
www.keiconn.com 800-325-0534 sales@keiconn.com
Visit us at Space Tech Expo in May at booths 3014 and 8030
Smiths KEI 1_2 pg ad Apr-May 2017.indd 1
www.embedded-computing.com
3/10/2017 1:05:50 PM
Embedded Computing Design | March/April 2017
21
SILICON: WIRELESS LANDSCAPE AND MODULE REPORT
tested using the TS8980 RF test system family. The available tests are continually being updated based on each LTE development, making it also suitable for V2X.
receiver. Only RF tests can ensure that the minimum physical requirements are met by OBUs and RSUs so that lives can be saved.
For 802.11p, the TS-ITS100 RF test system, for example, contains the complete package of global 802.11p test cases for:
Dr. Thomas Brüggen is Project Manager of RF Test Systems for Intelligent Transport Systems (ITS) at Rohde & Schwarz in Munich. He studied electrical engineering at RWTH Aachen and received a Ph.D. in communications engineering.
›› Europe at 5.9 GHz (ETSI EN 302 571) ›› United States at 5.9 GHz (IEEE 802.11-2012) ›› Japan at 760 MHz (TELEC T257 and ARIB STD-T109) For out-of-band tests, the test system permits measurements up to 18 GHz and can accept a variety of filters as needed for various regions. The system hardware is already set up to handle diversity and multiple input, multiple output (MIMO). 802.11p tests pose a special challenge in that there is no defined, uniform interface to 802.11p units. In order to configure the unit for a test case, the test software must address each unit with individual commands. The test system already contains plugins for many units in order to permit fully automated testing. Summary Development of wireless, automated communications between traffic participants will continue in order to improve road safety. Safety-critical messages must be transmitted reliably and quickly in every environment and every traffic situation. Protocol tests alone are not sufficient because they do not test the circumstances under which the transmit signal arrives at the
Notes: 1. OSI model: The Open Systems Interconnection model (Figure 2) was standardized as a reference model in order to permit data message transmission across a wide variety of technical systems. It combines the various tasks and protocols for a network into layers. The two lowest layers are the physical layer for the physical transmission of data messages (e.g., via a copper cable, a fiber-optic cable, or radio frequency) and the data link layer consisting of the media access control (MAC) layer and the logical link control (LLC) layer. 2. Packet error rate (PER): The number of faulty packets at the receiver that cannot be corrected, referenced to the total number of transmitted packets. 3. RF module: The RF module is the part of a data message transmitting device that converts a bit stream into an RF electromagnetic wave and then transmits this wave over an antenna. In the reverse direction, it receives a wave and converts it to a bit stream. 4. Plug test: Event during which manufacturers of test equipment and on-board/roadside units interconnect their devices in order to test how well they work together.
OpenSystems Media works with industry leaders to develop and publish content that educates our readers. How Bluetooth Low Energy technology powers the IoT By Silicon Laboratories The Internet of Things (IoT) is making our lives easier in some fascinating ways. Learn how low power consumption, pre-certified and affordable components, and Bluetooth Smart technology is making the construction industry safer, protecting bee colonies, and helping neonatal infants.
http://embedded-computing.com/white-papers/white-low-energy-technology-powers-iot/
22
Embedded Computing Design | March/April 2017
Check out our white papers at www.embedded-computing.com/ white-papers www.embedded-computing.com
2017 Submissions Now Open for the Annual Embedded Computing Design Innovation Awards
Top Embedded Top Innovative Innovators Products â&#x2122;Ś
Submit here: www.bit.ly/2017embedded
www.embedded-computing.com
SOFTWARE: CLOUD PLATFORM UPDATE
IoT platforms: The secret sauce for connected vehicle applications By Ron Felice
A
The times they are a changin’.
ccording to the IBM Institute for Business Value study “Automotive 2025: Industry without borders,” the automotive industry is making a drastic change[1]. These changes are due to a number of factors, including pervasive disruption and shifting consumer engagement and expectations. Digitization and connectivity are the precipitative elements driving the shift in consumer engagement and expectations. There is strong demand for a seamless connected life, and among some groups a constant social engagement. These drastic changes are resulting in the need to engage a broader ecosystem of stakeholders and partners, and to become more agile. This requires a departure from the closed and siloed methods and solutions of the past. According to a report by ABI Research, by 2020 telematics will power more than 73 million commercial vehicles[2]; another report estimates that 342 million automotive infotainment systems will ship between the years of 2015 and
24
2020[3]. If these were the only connected vehicles on the road, this would number about 415 million. This is an incredible figure, and one that illustrates a large potential for captive touchpoints. Many owners of connected vehicles today are realizing some of the benefits of that connectivity. These benefits include things like safety and security features (for example, roadside assistance and stolen vehicle location assistance); remote access features (like remote door lock/unlock and remote start); and, for some, the ability to incorporate Internet-based infotainment services like Yelp. I’ve looked at clouds from both sides now Minimally, an IoT platform is a technology that enables connection and management of devices – this, however, is nothing more than an entry point today. While connectivity is a critical component of the overall architecture, an IoT platform, to be truly useful, must support the management of data in addition to the management of connected devices. And, should an IoT platform wish to be considered comprehensive, it must provide the capability to develop applications and analyze the data it ingests, both in real-time and in batch. In a recent report on the topic, BI Intelligence broke platforms out into “three buckets”[4]: ›› Open, building block platforms – Mainly for developers working on apps at their companies ›› Closed, high-end platforms – Primarily for business or government ›› Product management platforms – For connecting and managing products provided by a company
Embedded Computing Design | March/April 2017
www.embedded-computing.com
IBM
www-935.ibm.com/industries/automotive/
@IBMAutomotive
SOFTWARE: CLOUD PLATFORM UPDATE
While I agree that there are platforms that exist within these three silos, more capable platforms exist that transcend these boundaries. These cross-categorical platforms provide a more well-rounded set of capabilities that enable the expansive ecosystems we see developing in the connected vehicle space. For example, a “closed” product management platform in which personally identifiable information (PII) can be kept secure with an open platform that allows for third-party application development to deliver valueadded content to vehicle occupants to improve the user experience. Dreaming about the things that we could be In today’s expanding connected vehicle environment, it is critical that our platforms support millions of simultaneously connected vehicles. The gateway provided by the platform must be scalable to support a range of applications. The platforms need to support the ingestion of contextual data, which is the data that exists in the area(s) vehicles operate. Incorporating real-time traffic information, point-of-interest information, as well as real-time weather data can significantly increase the depth and breadth of services that can be provided to the customer. With a platform-based approach, in addition to ingesting the data, it is possible to analyze the data in real-time as it is streaming onto the platform. Imagine a road hauler with a heavy load winding its way through the mountains. What had been originally predicted as a mild rainfall quickly begins to show signs of turning into a significant weather event with dangerously high winds. Realtime analysis of the storm trajectory coupled with the knowledge of the truck’s route and current position allow the service provider to make a real-time route update and send it directly to the in-cab navigation system, guiding the driver through a much safer route. A platform-based approach also provides the ability to more easily aggregate streams of data. For example, imagine www.embedded-computing.com
the traditional approach of delivering vehicles to market with adaptive cruise control. Information about whether the adaptive cruise feature is active could be captured and stored, and occasionally may even be looked at. If it is looked at, observers may notice that it is not being activated as frequently as expected. While this may be an interesting data point, to what could it be correlated? In a different part of the company, the marketing organization sees blips come across the corporate feed, “Don’t trust ACC!”; “Been driving for years, I can control my car better than a computer #AccOff.” Once again, interesting data points, but marketing organizations are more concerned about addressing real complaints about vehicle functions and not the “occasional” gripe about a feature. Now, imagine a platform-based approach where aggregation of not just data but disparate data streams is possible, and on that platform there exist services that can handle the unstructured data that is social media and classify comments while correlating them to actual relevant key performance indicators (KPIs) of interest to the product teams (for example, the enabling and disabling of the new adaptive cruise control feature). Very quickly, the platform dashboard would light up with indicators showing that there was a general lack of “trust” in the new feature. The ability of the team to respond and take corrective action is improved. Further analysis could also be done on ways to increase trust. Modifications to software could be implemented, built, and deployed from this same platform. As these various use cases illustrate, development on a platform provides many advantages over traditional development environments. In addition, the presence of a platform in the operational stage of vehicles enables richer engagement with the customers. Whether it’s a new lower latency or more secure protocol, a database with faster access time and transfer rates, new cognitive capabilities, or third-party applications, as new technologies and services become available a platform makes adoption of those technologies easier. There are other use cases that could be discussed, some well understood and adopted (like usage-based insurance) and others just beginning to emerge. Instead, however, I will leave you to dream of what a connected vehicle could become. Share those dreams with me on Twitter @r_felice. Ron Felice is a Solution Architect for the IBM Analytics and IoT division. He has 15 years of automotive engineering experience and is certified for automotive functional safety by the TÜV NORD. References: 1. “Automotive 2025.” IBM Automotive 2025 - United States. February 15, 2017. Accessed March 01, 2017. www-935.ibm.com/services/us/gbs/thoughtleadership/auto2025/. 2. “Telematics to Power More than 73 Million Commercial Vehicles by 2020 Amidst Cut-throat Competition.” ABI Research. Accessed March 01, 2017. www.abiresearch.com/press/ telematics-power-more-73-million-commercial-vehicl/. 3. “Over 342 Million Connected Automotive Infotainment Systems to Ship between 2015 and 2020.” ABI Research. Accessed March 01, 2017. www.abiresearch.com/press/over-342million-connected-automotive-infotainment/. 4. Business Insider. Accessed March 01, 2017. www.businessinsider.com/intelligence/ research-store/?IR=T&utm_source=businessinsider&utm_medium=report_teaser&utm_ term=report_teaser_store_text_link_the-iot-platforms-report-how-software-is-helping-theinternet-of-things-evolve-2017-1&utm_content=report_store_report_teaser_text_link&utm_ campaign=report_teaser_store_link&vertical=IoT#!/The-IoT-Platforms-Report/p/79018048. Embedded Computing Design | March/April 2017
25
STRATEGIES: FOG COMPUTING
The OpenFog Reference Architecture: A baseline for interoperability in the IIoT cloud-to-things continuum By Brandon Lewis, Technology Editor
Fog computing concepts have been floating in the ether for some time now, but it seems that industry has been challenged to put the theoretical models behind the architecture to use in the real world. Recently, however, the OpenFog Consortium released the OpenFog Reference Architecture (RA), a foundational document that will enable interoperable semiconductors, systems, and software for Industrial Internet of Things (IIoT) stakeholders, industry-wide. In this roundtable interview, Dr. Maria Gorlatova, Associate Research Scholar at Princeton University and Co-Char of the OpenFog Consortium Communications Working Group; Brett Murphy, Director of Business Development for IIoT at Real-Time Innovations (RTI) and Co-Chair of the OpenFog Consortium Software Infrastructure Group; and Rob Swanson, Principal Engineer at Intel and Technical Chair of the OpenFog Consortium, explain how the OpenFog RA’s eight technical pillars of Security, Scalability, Openness, Autonomy, Remote Access Services (RASs), Agility, Hierarchy, and Programmability provide a roadmap for developing fog solutions for the cloud-to-things continuum.
26
Embedded Computing Design | March/April 2017
Can you briefly define a fog computing architecture as it would exist in a moderate-scale IIoT deployment? MURPHY: For the OpenFog Consortium, the IIoT covers more than manufacturing or process automation; it also includes transportation, healthcare, energy, city infrastructure, and more. Across these industries, as IIoT is deployed, we see challenges that make fog computing a necessity. With terabytes of data being generated near the “edge” of these systems, it’s not practical to stream data to the cloud and back, all day, every day. There are latency challenges, network bandwidth costs to consider, the availability of those networks on top of this, and security concerns. Fog computing addresses these challenges in a way to process, protect, and act on this data much closer to where it’s created. For example, think of all the data in an airport that is constantly using video surveillance to monitor the activities of tens of thousands of passengers and employees every day. There would be huge amounts of video data created that is best processed and analyzed by fog compute nodes near the parking garage entrance, airport entrance, security line, and airport gate where the video cameras sit. License plates, luggage, and other pertinent data are connected to unique people identified as they enter the airport and proceed through it. People are checked against a no-fly list in an airport data center and suspicious behavior is flagged to airport security. Data is passing across layers of local fog compute nodes and analytics and other applications are running across many different layers in the system. When an aircraft leaves a gate, the destination airport’s system is notified through the cloud so it can track the same people as they leave the airport. SWANSON: This same “mesh” like architecture of networked fog compute nodes will be used across other industries and scenarios, with data being passed peer-to-peer within layers of fog compute www.embedded-computing.com
OpenFog Consortium
www.openfogconsortium.org
@openfogconsortium.org
www.linkedin.com/company/openfog-consortium
YOU TUBE
www.youtube.com/channel/UC5qlopog0HuzeKm5dLq_D-A
SOFTWARE: FOG COMPUTING
nodes and between layers, all the way up to the cloud or data center. If you think of a pump analytics system, you will need to use machine learning with inferencing based upon trained models for various pump scenarios (such audio and vibration), etc. The training of these machine learning models will take place in large servers in a data center or the cloud, while the models/algorithms themselves will be deployed to fog compute nodes on or near the pumps. In addition, since pumps are physically connected to each other through the pipes and fluid that is running through them in sequence, the behavior of one pump affects all others downstream. So the analytics are connected as well. Through fog computing and peer-topeer data communications, the analytics can be more reactive to adjacent pump behavior.
“THE NEXT STEP IN SYSTEM CAPABILITY COMES WITH DEPLOYING ANALYTICS AND PROCESSING TO THE EDGE DEVICES OR GATEWAYS, TURNING THEM INTO FOG COMPUTE NODES. THIS PROVIDES THE IIOT SYSTEM THE ABILITY TO HANDLE MUCH MORE DATA, TO INCREASE THE DEPTH OF THE ANALYTICS IN THE SYSTEM, AND TO REDUCE THE NETWORK BANDWIDTH REQUIRED BACK TO THE CLOUD.” much more data, to increase the depth of the analytics in the system, and to reduce the network bandwidth required back to the cloud. GORLATOVA: In addition, low latency and “intelligence at the edge” of a fog network enable fundamentally new capabilities in IoT devices, such as context awareness and adaptive behavior at a level that is impossible with current networking approaches. What is the OpenFog RA, and how will it help stakeholders overcome the obstacles of fog deployment in the IIoT? MURPHY: Some open architecture systems in a few industries that use fog computing concepts have just recently begun to get deployed, mostly in pilot projects. But those
What are the proposed benefits of fog networking over traditional network architectures for IoT? SWANSON: Fog computing really works around three main vectors: minimization of data backhaul to a cloud (on-premises or off-premises); reduced latency; and reliability of operation. As IIoT systems grow in complexity and capability, more capability is provided beyond monitoring to extend into optimization and eventual autonomy of system processes. With processing close to the edge, quick response time to events is better assured, and with processing deployed across layers from the edge to the cloud, resilience and security are increased. MURPHY: The simplest IIoT architectures use IoT gateways to gather data from edge devices and move it to the cloud for analysis and processing. These are typically monitoring use cases. The next step in system capability comes with deploying analytics and processing to the edge devices or gateways, turning them into fog compute nodes. This provides the IIoT system the ability to handle www.embedded-computing.com
SMX RTOS ®
Ideal for Your Project. • Integrated platforms for 150+ boards • Progressive MPU security • TCP/IPv4/6, mDNS, SNMPv3, SNTP, smxrtos.com/mpu SSH, TLS/SSL, HTTPS, SMTPS • Advanced RTOS kernel smxrtos.com/special WiFi 802.11n, P2P, SoftAP, WPA2 • • USB host and device • Broad ARM & Cortex support smxrtos.com/processors Flash file systems • • Drivers, BSPs, Bootloader, • Full source code. No royalty. GUI, IEEE754 Floating Point • Custom source eval and free trial Y O U R
R T O S
P A R T N E R
www.smxrtos.com
Embedded Computing Design | March/April 2017
27
SOFTWARE: FOG COMPUTING
are separate efforts in a few industries with little cross-pollination and duplication of effort. Traditional systems in most industries are still developed with proprietary vendor platforms, with some including community or ecosystem partners. This walledgarden platform model with vendor lock-in is one inhibitor to adoption. In addition, there is the cost of deployment of fog computing (more compute capability across the system with pervasive networking) over deploying simple IIoT gateways and doing cloud computing. There has to be a business, technical, and/or regulatory reason to move to fog computing. This is why we believe certain scenarios will see movement to fog computing prior to others. SWANSON: The OpenFog RA is a framework and roadmap to help software developers and system architects create the first generation of open fog computing systems (Figure 1). It creates a common language for fog computing, representing a unified framework for providing computing, networking, and storage in the cloud-to-things continuum. We’ve
FIGURE 1 28
“THE OPENFOG RA IS A FRAMEWORK AND ROADMAP TO HELP SOFTWARE DEVELOPERS AND SYSTEM ARCHITECTS CREATE THE FIRST GENERATION OF OPEN FOG COMPUTING SYSTEMS. IT CREATES A COMMON LANGUAGE FOR FOG COMPUTING, REPRESENTING A UNIFIED FRAMEWORK FOR PROVIDING COMPUTING, NETWORKING, AND STORAGE IN THE CLOUD-TO-THINGS CONTINUUM.” been working on this since OpenFog was formed in November 2015, and the OpenFog RA document has just been released. The OpenFog RA intends to align requirements for all of the suppliers – from silicon manufacturing to system manufacturing to software – in the fog computing environment. This means that silicon manufactures need to provide a baseline of technology and system manufacturers need to utilize them. This is required to set an established baseline for software running on these systems. Most setbacks are a result of incomplete requirements or just flatly not following system design requirements. Our goal with the OpenFog RA, and the subsequent work of OpenFog, is to address those requirements for areas of importance in fog computing. How will advanced technologies such as 5G and artificial intelligence (AI) factor into fog computing, and will the OpenFog RA address integration of these technologies? MURPHY: The future of IIoT will be driven by the deployment of pervasive networking, pervasive computing, and AI. Fog computing is the architecture that brings
The OpenFog Reference Architecture (RA) relies on eight technical pillars (Security, Scalability, Openness, Autonomy, Remote Access Services (RASs), Agility, Hierarchy, and Programmability) to enable fog networking, compute, storage, control, and acceleration for Industrial Internet of Things (IIoT) deployments. Embedded Computing Design | March/April 2017
www.embedded-computing.com
those three elements together. AI enables computer systems to think and operate independently on data. This helps IIoT systems better deliver on advanced use cases around optimization and autonomy. It is critical that as AI emerges we have a stable, baseline architecture for computing. 5G promises to connect and allocate bandwidth more efficiently than 4G, but that promise needs to be realized. GORLATOVA: Fog will bring AI close to the endpoint devices. In developing the OpenFog RA, we made multiple decisions that will allow this to happen. The IoT has been described as an ecosystem with more standards bodies than actual standards. How will the OpenFog RA integrate with other leading technology standards, if at all? SWANSON: The OpenFog Consortium is not a standards organization, but much of our work focuses on creating and testing the requirements for the eventual standards that will be created to enable component-level interoperability. We are partnering with standards organizations such as the IEEE in this work. We are identifying many of the main standards development organizations (SDOs) and consortia that we want to work with so that we can leverage their work and not re-invent the wheel as it relates to IoT requirements. The OpenFog RA is an important first step in establishing interoperability. But we know that even with the most stringent standards, interoperability is challenged, so we will also include fog fests to help address these issues. We are first addressing the various interfaces between our architectural layers. This will help with overall system composability. One such example is the liaison arrangement we have with the Open Connectivity Foundation (OCF), where we are working to integrate their efforts into ours. GORLATOVA: We also recently signed a partnership agreement with ETSI-MEC and are actively collaborating with them. Where and when should we expect to see fog truly take hold? GORLATOVA: Fog computing will accelerate dramatically over the next 2-5 years. On the academic side, we already see a dramatic increase in interest in fog-related challenges, with many exciting projects currently underway at universities worldwide. We are highly likely to see the academic community solve many important challenges in the different aspects of fog deployment. SWANSON: Weâ&#x20AC;&#x2122;re starting to see adoption take place, but we expect this to really accelerate as the specifications for standards start to emerge. There will be early market adopters in certain industries where fog is essential and necessary to the use cases. For example, visual analytics will likely be an early market where fog computing takes off. This is really addressing the backhaul minimization aspect where you cannot afford the network costs to process everything in the backend cloud. We also think that transportation, energy, and smart cities will be early adopters. The City of Barcelona, for example, is already using fog computing for waste management, traffic management, and smart lighting, so city administrators can have real-time information to make decisions in a single platform. MURPHY: We believe fog computing is the fundamental enabler of the more advanced IIoT use cases coming in the future. It brings together the pervasive networking and computing needed for IIoT. Our goal at OpenFog is to define an open architecture for fog computing that ensures a vibrant ecosystem of providers and interoperable solutions that will accelerate the IIoT. Editorâ&#x20AC;&#x2122;s Note: The OpenFog Reference Architecture can be downloaded free of charge from www.openfogconsortium.org/ra. www.embedded-computing.com
Embedded Computing Design | March/April 2017
29
Editor’s Choice
EDITOR’S CHOICE
www.embedded-computing.com/editors-choice
Xilinx enables vision-guided machine learning applications The new Xilinx reVISION stack allows software and system engineers with little or no hardware design expertise to develop intelligent visionguided systems, enabling vision-focused applications in markets such as high-end consumer, automotive, industrial, medical, and aerospace & defense. Developers can use a C/C++/OpenCL development flow with industry-standard frameworks and libraries like Caffe and OpenCV to develop embedded vision applications on a single Zynq system on chip (SoC) or multiprocessor SoC (MPSoC). The Xilinx reVISION stack includes support for popular neural networks including AlexNet, GoogLeNet, SqueezeNet, SSD, and FCN; library elements include pre-defined and optimized implementations for deep and convolutional neural network (DNN/CNN) layers; and a set of acceleration-ready OpenCV functions for computer vision processing. The reVISION stack also includes development platforms from Xilinx and third parties that include various types of sensors.
Xilinx
www.xilinx.com/revision | www.embedded-computing.com/p374099
Development kit for evaluation and prototyping of mid-air haptics in touchless gesture controls Ultrahaptics’ UHDK5 TOUCH Development Kit allows companies to evaluate gesture controls enhanced by ultrasound, enabling users to “feel” virtual buttons, switches, dials, and other objects in mid-air. Ultrahaptics’ self-contained plug-and-play UHDK5 TOUCH Development Kit works out-of-the-box and contains demonstrations that require no technical knowledge from the user. For developers, the embeddable architecture (utilizing an ARM core plus FPGA) eases integration and provides a production-ready design. Similarly, APIs coded using the C++ programming language allow software engineers to adapt their application interfaces with customized sensations.
Ultrahaptics
www.ultrahaptics.com | www.embedded-computing.com/p374100
Low-cost oscilloscope series caters to students, sub-economy market The new low-cost InfiniiVision 1000 X-Series oscilloscopes from Keysight Technologies, Inc. include 50 to 100 MHz models; come with a standard educator’s resource kit containing built-in training signals, a comprehensive lab guide, and fundamentals slide set for professors and lab assistants; and start at $449 USD. In addition, the 1000 X-Series oscilloscopes are also serial protocol analyzers, digital voltmeters, and frequency counters. The scope features 24 typical oscilloscope measurements to analyze signals and determine signal parameters. Additional signal analysis is provided by the gated fast fourier transform (FFT) function, which allows users to correlate time and frequency domain phenomena on a single screen. Mask limit testing is also available to help users detect signal errors. The 1000 X-Series supports decoding and analysis for a range of popular embedded and automotive serial bus applications, including I2C, SPI, UART/RS232, CAN, and LIN.
Keysight Technologies, Inc.
www.keysight.com | www.embedded-computing.com/p374101
30
Embedded Computing Design | March/April 2017
www.embedded-computing.com
EXECUTIVE SPEAKOUT
Turning the Industrial IoT toolbox Blue(tooth) Interview with Mats Andersson, Senior Director of Technology, Short Range Radio Product Center, u-blox Given recent specification updates, what is Bluetooth’s outlook in the Industrial Internet of Things (IIoT) sector? ANDERSSON: Bluetooth 5 introduces several enhancements on top of those enabled by Bluetooth Low Energy (BLE) technology that further allow Bluetooth to move from a cable replacement technology to a network-oriented technology, and thus better able to support IoT applications in all vertical segments. This evolution began with the Bluetooth Special Interest Group (SIG) adopting BLE technology (previously known as Bluetooth Smart) into the mainline of the Bluetooth Core Specification in 2010, which was then dubbed Bluetooth 4.0. Although this initially lowered the maximum theoretical bandwidth of the technology from the low megabyte range to the hundreds of kilobytes, BLE also reduced power consumption by orders of magnitude (from approximately 1 W transmission power into the milliwatt range) and minimized latencies into the low milliseconds while retaining similar range to its predecessors. Furthermore, Bluetooth 4.1 added support for the IPv6 protocol developed to provide enough address space for all the “things” connecting to the Internet. All of this had two effects on Bluetooth’s addressable market. First, it opened up a range of new applications for Bluetooth beyond the traditional consumer space, some of which included commercial and IIoT applications governed by latency and power consumption more so than sheer throughput. Second, it provided an upgrade path for legacy Bluetooth devices, which helped bring consumer economies of scale into these emerging markets – billions of Bluetooth devices are shipped year over year, making the technology extremely cost competitive regardless of application.
installed base of Bluetooth devices. Considering the range, throughput, power consumption, mesh networking, compatibility, and cost advantages of the Bluetooth specification today, many industrial and building automation applications stand to benefit. With the potential benefits of consumer wireless standards for the IIoT space, are long-lived 802.15.4-based technologies set to be replaced? ANDERSSON: The benefits of Bluetooth mentioned previously that extend the technology’s utility as a networking toolbox can, in a number of applications and vertical markets, replace incumbent 802.15.4 devices and protocols, be they standardsbased or proprietary. As IoT systems and architectures in general move increasingly towards IP-based backbones in not only the data center and gateway but also at the edge, Bluetooth will continue to advance the proliferation of TCP/IP through IPv6-based networks. As a happy byproduct, this will also solve many of the network interoperability issues that occur at the intersection of operational technology (OT) and information technology (IT) networks. Of course, there is no one-size-fits-all networking technology when it comes to the IoT, and niche application areas will remain that require extremes in terms of wireless latencies. Here, Bluetooth may not be a fit, but 802.15.4-based solutions may not either, for that matter.
Bluetooth 5 essentially brings more tools to the toolbox for designers of all kinds of connected applications, including increased range (up to 4x) and throughput (2 Mbps link performance compared to 1 Mbps today) that can be tuned on a sliding scale with power consumption based on specific system requirements. Perhaps more important, though, is mesh networking capability. Bluetooth Mesh can be considered a true mesh network in that the specification will not require a network hub or gateway, which both reduces single points of failure to improve network reliability and enables the capacity to support tens of thousands of nodes that can almost limitlessly extend the range of a Bluetooth signal.
However, where it can be used, it’s better to do so than go against the grain. Doing so can hurt device manufacturers in terms of time to market as well as reliability, particularly when using proprietary technology. A good example of this is security. Recently, British and Belgian security researchers found vulnerabilities in the communications protocols of no less than 10 implantable cardiac defibrillators (ICDs). Conversely, Bluetooth has featured “Secure Connections” since version 4.2, when Federal Information Processing Standard-compliant (FIPS-compliant) levels of link layer security were added. The move towards IPv6 technology also enables tested and trusted end-to-end security protocols like TLS/DTLS to be leveraged in Bluetooth-based applications, and Bluetooth 5 will continue this tradition in upcoming releases to meet stringent government-class security requirements. The soonto-be-released Bluetooth Mesh specification also has a large emphasis on security.
Moving forward, even more support can be added with IPv6 over Mesh, all while maintaining compatibility with the
Given time, the choice of Bluetooth technology in most IoT and IIoT edge networking applications will be a simple one.
u-blox
www.u-blox.com/en/short-range-radio-modules
TWITTER @ublox
GITHUB
https://github.com/u-blox
www.linkedin.com/company/u-blox
YOUTUBE
www.youtube.com/c/ublox1
Connecting the Intelligent World from Devices to the Cloud IoT Gateway Solutions
E100-8Q, E200-9B
Compact Embedded Server Appliances
Network, Security Appliances
SYS-5028D-TN4T
SYS-5018D-FN8T (Front I/O)
Cold Storage
SYS-5018A-AR12L
High Performance / Short Depth Solutions
SYS-1028R-WMRT
High Performance / IPC Solutions
SYS-6018R-TD (Rear I/O)
Intel Inside®. Powerful Productivity Outside.
Learn more at www.supermicro.com/embedded Intel, the Intel logo, Intel Core, Intel Quark, Xeon, and Xeon Inside are trademarks or registered trademarks of Intel Corporation in the U.S. and/or other countries. All other brands and names are the property of their respective owners.
03_Q3_SM_USP_070616_MasterFile
• Low Power Intel® Quark™, Intel® Atom™ Intel®, Core™ processor families, and High Performance Intel® Xeon® processors • Standard Form Factor and High Performance Motherboards • Optimized Short-Depth Industrial Rackmount Platforms • Energy Efficient Titanium - Gold Level Power Supplies • Fully Optimized SuperServers Ready to Deploy Solutions • Remote Management by IPMI or Intel® AMT • Worldwide Service with Extended Product Life Cycle Support • Optimized for Embedded/IoT Applications