Robotics Software

<p>Buildings account for about 40 percent of U.S. energy consumption, and are responsible for one-third of global carbon dioxide emissions. Making buildings more energy-efficient is not only a cost-saving measure, but a crucial climate change mitigation strategy. Hence the rise of “smart” buildings, which are increasingly becoming the norm around the world.</p><p>Smart buildings automate systems like heating, ventilation, and air conditioning (HVAC); lighting; electricity; and security. Automation requires sensory data, such as indoor and outdoor temperature and humidity, carbon dioxide concentration, and occupancy status. Smart buildings leverage data in a combination of technologies that can make them more energy-efficient.</p><p>Since HVAC systems account for nearly half of a building’s energy use, smart buildings use smart thermostats, which automate HVAC controls and can learn the temperature preferences of a building’s occupants.</p><p>In a&nbsp;<a href="https://doi.org/10.1016/j.apenergy.2020.115451" target="_blank">paper</a> published in the journal&nbsp;<em>Applied Energy</em>, researchers from the MIT Laboratory for Information and Decision Systems (LIDS), in collaboration with Skoltech scientists, have designed a new smart thermostat which uses data-efficient algorithms that can learn optimal temperature thresholds within a week.</p><p>“Despite recent advances in internet-of-things technology and data analytics, implementation of smart buildings is impeded by the time-consuming process of data acquisition in buildings,” says co-author Munther Dahleh, professor of electrical engineering and computer science and director of the Institute for Data, Systems, and Society (IDSS). Smart thermostat algorithms use building data to learn how to operate optimally, but the data can take months to collect.</p><p>To speed up the learning process, the researchers used a method called manifold learning, where complex and “high-dimensional” functions are represented by simpler and lower-dimensional functions called “manifolds.” By leveraging manifold learning and knowledge of building thermodynamics, the researchers replaced a generic control method, which can have many parameters, with a set of “threshold” policies that each have fewer, more interpretable parameters. Algorithms developed to learn optimal manifolds require fewer data, so they are more data-efficient.</p><p>The algorithms developed for the thermostat employ a methodology called reinforcement learning (RL), a data-driven sequential decision-making and control approach that has gained much attention in recent years for mastering games like backgammon and Go.&nbsp;</p><p>“We have efficient simulation engines for computer games that can generate abundant data for the RL algorithms to learn a good playing strategy,” says Ashkan Haji Hosseinloo, a postdoc at LIDS and the lead author of the paper. “However, we do not have the luxury of big data for microclimate control in buildings.”</p><p>With a background in mechanical engineering and training in methods like RL, Hosseinloo can apply insights from statistics and state-of-the-art computing to real-world physical systems. “My main motivation is to slow down, and even prevent, an energy and environmental crisis by improving the efficiency of these systems,” he says.</p><p>The smart thermostat’s new RL algorithms are “event-triggered,” meaning they make decisions only when certain events occur, rather than on a predetermined schedule. These “events” are defined by certain conditions reaching a threshold — such as the temperature in a room dropping out of optimal range. “This enables less-frequent learning updates and makes our algorithms computationally less expensive,” Hosseinloo says.&nbsp;</p><p>Computational power is a potential constraint for learning algorithms, and computational resources depend on whether algorithms run in the cloud or on a device itself — such as a smart thermostat. “We need learning algorithms that are both computationally efficient and data-efficient,” says Hosseinloo.</p><p>Energy-efficient buildings offer additional advantages beyond reducing emissions and cutting costs. A building’s “microclimate” and air quality can directly affect the productivity and decision-making performance of building occupants. Considering the many large-scale economic, environmental, and societal impacts, microclimate control has become an important issue for governments, building managers, and even homeowners.</p><p>“The new generation of smart buildings aims to learn from data how to operate autonomously and with minimum user interventions,” says co-author Henni Ouerdane, a professor on the Skoltech side of the collaboration. “A learning thermostat can potentially learn how to adjust its set-point temperatures in coordination with other HVAC devices, or based on its prediction of electricity tariffs in order to save energy and cost.”</p><p>Hosseinloo also believes their methodology and algorithms apply to a diverse range of other physics-based control problems in areas including robotics, autonomous vehicles, and transportation, where data- and computational efficiency are of paramount importance.</p><p>This research was a Skoltech-MIT Joint Project conducted as part of the MIT Skoltech Next Generation Program.</p>

A diagram from a new paper detailing a smart thermostat shows the interaction of weather conditions, control systems, sensors, and learning algorithms to optimize microclimates in buildings. Credits:Image courtesy of the researchers.

A smart thermostat quickly learns to optimize building microclimates for both energy consumption and user preference.

Making smart thermostats more efficient

<h2>Air quality monitoring and people’s well-being</h2><p>Proper lighting and indoor air quality go hand in hand when it comes to productivity and people’s well-being. Indeed, both factors have a direct impact on the way we feel and interact at work or learn in the classroom. It is usually quite straightforward to identify situations where lighting needs improvement. However, it is much more difficult for people to assess the quality of the air inside a room. It is hence not uncommon to see meeting rooms and classrooms with poor ventilation, especially during the heating period.</p><p>According to <a href="https://eartharxiv.org/repository/view/527/" rel="noopener noreferrer" target="_blank">studies</a>, the level of carbon dioxide (CO2) in indoor spaces can directly affect people’s well-being. Poor air quality due to high levels of CO2 is linked to poor decision making, lack of focus, and drowsiness. Usually, people start experiencing some physical effects at 900 ppm but it is not rare to measure levels higher than 1000 ppm with peaks above 2000 ppm.</p><p>Good ventilation of schools and office spaces has also become a global concern with the recent COVID-19 outbreak. In this post-pandemic area, facility managers need to be able to measure the air quality on their premises and take appropriate measures in order to ensure a healthy environment for everyone.</p><h2>An innovative smart lighting solution</h2><p>To provide a viable solution to this situation, Astra LED and <a href="https://akenza.com/?utm_source=wevolver&utm_medium=referral&utm_campaign=airmonitoring" rel="noopener noreferrer" target="_blank">Akenza</a> have joined forces and developed an air monitoring smart lighting solution. The system can be split into two main building blocks: a lamp with an embedded sensor, and an Internet Of Things Connectivity Platform.</p><p>First, a sensor embedded in the lamp measures the air quality parameters of the room. These are typically the Carbon Dioxide (CO2) and Volatile Organic Compounds (VOC) levels, the temperature and the humidity. These parameters are then transmitted to the IoT platform via a low-power, long-range radio communication protocol (LoRaWAN connectivity). The IoT platform then monitors the air quality levels continuously and displays the data on a dedicated dashboard.</p><p>If the measured levels reach a predefined threshold, the
lamp changes color and intensity accordingly (green: acceptable, orange: high levels, red: critical levels). This visual
notification indicates that the room needs to be ventilated. In
the case of schools, this is used as a playful approach to raise
awareness about the topic and lead students to act.</p><p><span class="fr-img-caption fr-fic fr-dib" style="width: 300px;"><span class="fr-img-wrap"><img src="https://s3-us-west-1.amazonaws.com/wevolver-project-images/froala%2F1607948747554-smartlightred.jpg"><span class="fr-inner" spellcheck="false">The Smart Light changes to red in case of poor air quality</span></span></span></p><p>The smart lamp and sensors have been developed by Astra LED. The company develops and produces integrated and adaptive lighting and sensor solutions that increase well-being and reduce operating and maintenance costs.</p><p>The IoT software infrastructure is provided by <a href="https://akenza.com/?utm_source=wevolver&utm_medium=referral&utm_campaign=airmonitoring" rel="noopener noreferrer" target="_blank">Akenza</a>.
The platform makes it possible to connect to the cloud any
type of sensor via different connectivity technologies (e.g.
LoRaWAN, Sigfox, NB-IoT, LTE-M, 5G). The device can hence
be managed in one place, and the sensor’s data made available to the cloud in a dedicated application or ERP system.</p><p><span class="fr-img-caption fr-fic fr-dib" style="width: 300px;"><span class="fr-img-wrap"><img src="https://s3-us-west-1.amazonaws.com/wevolver-project-images/froala%2F1607947363656-smartlightroom.jpg"><span class="fr-inner" spellcheck="false">Smart Light in an office space</span></span></span></p><h2>The case of the city of Zug, Switzerland</h2><p>One of the first implementations of the system has been made
in the city of Zug. The depicted solution has been put in
service in several schools of the city to monitor the air quality
inside the classrooms.
The smart light solution is also in service in the townhouse of the city, to monitor the air quality in meeting rooms.</p><p><span class="fr-img-caption fr-fic fr-dib" style="width: 300px;"><span class="fr-img-wrap"><img src="https://s3-us-west-1.amazonaws.com/wevolver-project-images/froala%2F1607946977129-SmartLightMeetingRoom.jpg"><span class="fr-inner" spellcheck="false">Smart Light in a meeting room</span></span></span></p><h2>Towards a healthier indoor environment</h2><p>Fostering a healthy environment in schools and workplaces
is a critical precondition to ensure the well-being of students
and the workforce as well as to ensure high productivity
levels. With the current solution, school and facility managers
have a viable tool to track and monitor the air quality in their
premises.</p><article><p>For more information about indoor air climate monitoring, visit <a href="https://akenza.com/?utm_source=wevolver&utm_medium=referral&utm_campaign=airmonitoring" rel="noopener noreferrer" target="_blank">www.akenza.com</a>.</p></article>

Smart Light with Air Quality Monitoring Sensor

Proper lighting and indoor air quality go hand in hand when it
comes to productivity and people’s well-being. Discover how a Smart Light Solution for Indoor Air Monitoring helps schools and companies improve the air quality in their premises.

Improving indoor air quality thanks to IoT

<section><p><strong>In order to keep up with the growth in - mobile - data traffic, wireless signals with an increasingly shorter wavelength are required. Doctoral candidate <a href="https://research.tue.nl/en/persons/alvaro-morales-vicente" target="_blank">Álvaro Morales Vicente</a> investigated how you can generate and transmit so-called millimeter waves with the use of photonic chips.</strong></p></section><section tabindex="-1"><p>Those of us working from home who -&nbsp;out of necessity or by choice -&nbsp;have taken a closer look at their Wi-Fi network recently, will know that the router sends signals over two frequencies: 2.4 Gigahertz and 5 Gigahertz. The latter connection is faster, but also has a shorter range unfortunately, due to the shorter radio waves (6 centimeters instead of 12.5) -&nbsp;which is why an extra Wi-Fi point is sometimes needed to enhance the signal.</p><p>In order to provide everyone with ultrafast internet indoors and outdoors, to allow self-driving cars to communicate with each other, or to enable wireless data transfer within data centers, ‘data rates’ of several Gigabits per second are required. For this, even higher frequencies are needed, ranging from 30 to 300 Gigahertz, corresponding to wavelengths of a few millimeters.</p><h2>Beat</h2><p>This poses several challenges, doctoral candidate Álvaro Morales Vicente from Spain says. “Since short waves cover a smaller range, the signals need to be steered in a very specific direction over a short distance. That can be done with several antennas in a row.” In addition, generating such high frequencies with traditional electronics is no easy task in itself, Morales Vicente emphasizes. “The main problem is the limited bandwidth of the modulation that can be handled by traditional electronics, which means that you cannot achieve the desired high bit rates.”</p><p>There is an alternative: using laser light to very accurately generate the desired radio frequencies. Laser light has a much higher frequency than radio waves, but combining two laser beams with slightly different ‘colors’ leads to a periodic variation in light intensity caused by interference, Morales Vicente explains. “That difference frequency can be converted to a radio signal with the use of a light detector.” This phenomenon is comparable to the ‘beating’ you can hear when you pluck two strings of a slightly out of tune guitar.</p><p><span class="fr-img-caption fr-fic fr-dib" style="width: 788px;"><span class="fr-img-wrap"><img src="https://s3-us-west-1.amazonaws.com/wevolver-project-images/froala%2F1606906402771-csm_Sluitstuk+detailfoto+Javier+P%C3%A9rez+Santacruz_ba8a309c78.png"><span class="fr-inner">Photonic chip that steers the radiation direction of millimeter-wave radio signals for high-speed wireless communications. Foto: Javier Pérez Santacruz&nbsp;</span></span></span></p><h2>Optical Chip</h2><p>The next essential step is ‘aiming’ the millimeter waves to bring the signal from A to B. “That too works with interference. When the signal is transmitted by a row of antennas, each with a slightly different delay, the radio signal is steered in a certain direction.” Morales Vicente managed to realize this effect at different wavelengths in the lab by creating the signals on an optical chip.</p><p>Optical generation of millimeter waves is energy-efficient and flexible, but this technology is still in its infancy. “The major challenge is to combine all the necessary steps in a handy device. That includes generating a signal with the right frequency, modulating this to store data in it, and steering the millimeter waves in a specific direction to deliver said data.”</p><p>The idea is that all this can eventually be done on an optical chip, a technology in which TU/e is leading. Morales Vicente: “I still used separate lasers to generate the signals, but I have already carried out a study on how to realize this on an optical chip.”</p><hr><p><style type="text/css"></style><style type="text/css"></style><span data-sheets-userformat='{"2":769,"3":{"1":0},"11":3,"12":0}' data-sheets-value='{"1":2,"2":"This article was first published here"}'>This article was first published here:&nbsp;<a href="https://www.tue.nl/en/our-university/departments/mathematics-and-computer-science/the-department/news/news-overview/30-11-2020-radio-from-laser-light/" rel="noopener noreferrer" target="_blank">https://www.tue.nl/en/our-university/departments/mathematics-and-computer-science/the-department/news/news-overview/30-11-2020-radio-from-laser-light/</a></span></p></section>

Álvaro Morales Vicente will defend his thesis on December 7th at the department of Electrical Engineering of the TUe

Radio from laser light

A new approach to image search uses images returned by traditional search methods as nodes in a graph neural network through which similarity signals are propagated, achieving improved ranking in cross-modal retrieval.

Web image search gets better with graph neural networks

<p>This webinar provides an overview of Nordic Semiconductor's latest prototyping platform for cellular IoT applications</p><p>Content:</p><ul><li>Importance of Proof of Concepts (PoC)</li><li>Cellular IoT with LTE-M and NB-IoT</li><li>Nordic Thingy:91</li><li>The nRF9160 SiP for LTE-M/NB-IoT + GPS</li><li>Asset tracker example</li><li>Hybrid applications with cIoT and Bluetooth</li></ul><p>Don't forget to check out the <a href="https://www.nordicsemi.com/Software-and-Tools/Prototyping-platforms/Nordic-Thingy-91?utm_campaign=2021%20wevolver" rel="noopener noreferrer" target="_blank"><strong>Thingy:91</strong></a><a href="https://www.nordicsemi.com/Software-and-Tools/Prototyping-platforms/Nordic-Thingy-91?utm_campaign=2021%20wevolver" rel="noopener noreferrer" target="_blank"><strong>&nbsp;product page</strong></a>, where you can also <strong>click to buy this product</strong> from our distribution partners</p><p>The webinar was live August 22 at 10:30 CEST.</p><p><strong>Questions &amp; answers&nbsp;</strong>from the live session and a PDF of the presentation can be found <a href="https://devzone.nordicsemi.com/nordic/nordic-blog/b/blog/posts/q-a-for-introduction-to-nordic-thingy-91-webinar?utm_campaign=2021%20wevolver" rel="noopener noreferrer" target="_blank">here</a>.&nbsp;</p>

This webinar provides an overview of Nordic Semiconductor's latest prototyping platform for cellular IoT applications

Introducing The Nordic Thingy:91 Cellular IoT prototyping platform

<p>This webinar takes you through the steps to get started with cellular IoT development on Nordic Thingy:91, from unboxing your kit to writing your own code.</p><p><strong><u>Content:</u></strong></p><ul><li>What is Nordic Thingy:91?&nbsp;</li><li>Unboxing&nbsp;</li><li>Installing nRF Connect SDK</li><li>How to compile and flash</li><li>How to modify an example</li><li>Q&amp;A</li></ul><p>The webinar was live February 24th&nbsp;at 2 PM CET.</p><p>The presentation used in the webinar can be downloaded <a href="https://devzone.nordicsemi.com/f/nordic-q-a/58281/where-can-i-find-the-presentation-used-in-the-cellular-iot-development-on-nordic-thingy-91-webinar?utm_campaign=2021%20wevolver" rel="noopener noreferrer" target="_blank">here</a>.</p>

This webinar takes you through the steps to get started with cellular IoT development on Nordic Thingy:91, from unboxing your kit to writing your own code.

Cellular IoT development on Nordic Thingy:91

<p>nRF Connect SDK is Nordic's software development kit for developing both short-range wireless and cellular IoT applications. nRF52 Series SoCs are now supported for development, in addition to the nRF5340 SoC and the nRF9160 SiP.</p><p>This webinar recording introduces the nRF Connect SDK.</p><p><strong><u>Content:</u></strong></p><ul><li>Overview</li><li>Enterprise Technologies and Tools</li><li>Repositories and source code</li><li>How to manage and configure source code</li><li>What is an RTOS?</li><li>The nRF Connect platform</li></ul><p>We had two webinars on the same topic on May 12th 2020. This is the recording of the presentation from one of them, but includes Q&amp;A sessions from both.</p><p>The presentation used in this webinar can be downloaded <a href="https://devzone.nordicsemi.com/f/nordic-q-a/61402/where-can-i-find-the-presentation-used-in-the-nrf-connect-sdk---the-next-generation-for-nordic-wireless-solutions-webinar?utm_campaign=2021%20wevolver" rel="noopener noreferrer" target="_blank">here</a>.&nbsp;</p>

nRF Connect SDK is Nordic's software development kit for developing both short-range wireless and cellular IoT applications. nRF52 Series SoCs are now supported for development, in addition to the nRF5340 SoC and the nRF9160 SiP.

nRF Connect SDK - The next generation for Nordic wireless solution

<p>LTE-M and NB-IoT enable everything to be connected to the Internet. Learn how the nRF9160 SiP realizes this and makes it possible to build very small devices with many years of battery life.</p><p><strong><u>Content:</u></strong></p><ul><li>Overview of the nRF9160 SiP</li><li>Latest feature additions</li><li>Low power optimized designs&nbsp;</li><li>How to use the nRF9160 application processor&nbsp;</li><li>Getting started with cellular IoT application development&nbsp;</li><li>Q&amp;A</li></ul><p>The webinar was live June 29th at 5 PM CEST.</p><p>The presentation used in the webinar can be downloaded <a href="https://devzone.nordicsemi.com/f/nordic-q-a/63364/where-can-i-find-the-presentation-used-in-the-nrf9160-sip-optimized-for-ultra-low-power-and-small-cellular-iot-applications-webinar?utm_campaign=2021%20wevolver" rel="noopener noreferrer" target="_blank">here</a>.&nbsp;</p>

LTE-M and NB-IoT enables everything to be connected to the Internet. Learn how the nRF9160 SiP realizes this and makes it possible to build very small devices with many years of battery life.

nRF9160: Optimized for ultra low power and small cellular IoT applications

<p>Exciting innovations in environmental conservation-related technology comes to rescue for some of the deforesting, equivalent of 27 soccer pitches every minute, going on in the world.</p><p>Forests are essential to global biodiversity. They provide food and habitats for wildlife and indigenous populations and help reduce the impact of climate change. But across the world, forests are at risk of decimation by combining human-driven and natural threats.</p><p>Innovative wireless conservation technologies are emerging to combat this concerning trend and encourage greater action against deforestation.</p><h2>The threat to the world's forests</h2><p>Increasing global demands for food, materials, and energy cause widespread deforestation and forest degradation. As the number of trees on our planet dwindles, potentially irreversible harm is done to terrestrial biodiversity. How precarious is this reality? In a landmark report, the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services (IPBES) published these figures: Approximately 2.9 million km2 of native forest cover was lost from 1990 to 2015 due to clearing and wood harvesting. Intact forests suffered a seven percent reduction from 2000 to 2013 in developed and developing countries.</p><p>Today, things are not getting any better; the World Wildlife Fund (WWF) claims, on average, the world loses 75,700 km2 of forests annually — that's the equivalent of 27 soccer pitches every minute.</p><p>Protected land zones represent a vital weapon in the global fight against deforestation. However, successfully enforcing borders, regulations, and restrictions is much easier said than done. Local authorities and conservationists everywhere need more support to identify and restrict human-driven threats to forests. Typically, illegal logging activity and natural events such as wildfires might otherwise go unchecked. Ultimately, meaningful change relies on closely monitoring the activities occurring within a tropical rainforest's designated protected areas.</p><h2>Wireless to the rescue</h2><p>Protected status alone cannot save forests across the globe. Thankfully, exciting innovations in environmental conservation-related technology have joined the battle. Utilizing short- and long-range wireless networks, sensor-based devices, like remote monitoring systems connected to Cloud platforms and smartphone apps, are at authorities' hands. Thus, they can constantly listen to all the sounds made in a forest, even if they're kilometers or continents away.</p><p><a href="https://www.codeofconscience.org/" rel="noopener" target="_blank">Code of Conscience</a> is a leading initiative supporting efforts to limit illegal deforestation. Launched in September 2019, the initiative aims to track the type of heavy-duty vehicles often used to rip through the forest in both legal and banned logging operations.</p><p>Code of Conscience's solution autonomously restricts crews from entering protected zones (as determined by the UN World Database on Protected Areas). The open-source software uses publicly available mapping data with existing GPS tracking technology installed in construction vehicles. The GPS-based "geofence" capabilities interlock with the forest-clearing machinery's fuel pumps, enabling automatic shutdown if the equipment moves into a restricted area.</p><p>Integrated cellular connectivity enables notifications and audits of the machinery's position during regular operation and provides a method for over-the-air map data updates. A small, low-cost chip has also been developed to equip older, non-GPS vehicle models with the same capability.</p><h2>Listening out for deforestation</h2><p>Code of Conscience has been piloted in the Amazon. Using Nordic Semiconductor's&nbsp;<a href="https://www.nordicsemi.com/Software-and-tools/Prototyping-platforms/Nordic-Thingy-91?utm_campaign=2021%20wevolver" rel="noopener noreferrer" target="_blank">Nordic Thingy:91</a> multi-sensor cellular IoT prototyping platform, they demonstrate forestry and agricultural vehicles tracking on land and water. The Nordic Thingy:91 offers cellular connectivity 'out-of-the-box.' OEM partners can easily integrate the technology into their machinery for future applications and remote installations.</p><p>Another solution can highlight the dangers of deforestation to local communities by detecting loud sounds in forests. The unique device was developed by the young winner of the not-for-profit&nbsp;<a href="https://microbit.org/" rel="noopener" target="_blank">Micro:bit Educational Foundation</a>'s "do your:bit" BBC micro:bit challenge (North America), inspired by the UN Global Goals initiative.</p><p>The system includes a Raspberry Pi computer with a connected camera and a microphone that communicates with the Micro:bit (a Nordic SoC-based pocket computer). It wirelessly relays an audio signal then automatically takes a photo and posts it to a public Twitter account. A local park ranger can then review the picture to determine the noise source and respond accordingly — whether it's the crack of lighting or something more sinister, for example, the buzz of a chainsaw.</p><h2>Light at the end of the canopy</h2><p>Could solutions providing remote access to actionable information and real-time alerts help individuals, organizations, and communities in their quest to save the forests and, in turn, the planet? With support from robust wireless networks, and if the world is willing to get on board, there certainly appears to be light at the end of the canopy.&nbsp;</p><hr><p>This article was first published on Nordic's&nbsp;<a href="https://blog.nordicsemi.com/getconnected" rel="nofollow" target="_blank"></a><a href="https://blog.nordicsemi.com/getconnected?utm_campaign=2021%20wevolver" rel="noopener noreferrer" target="_blank">Get Connected Blog</a>.</p>

Exciting innovations in environmental conservation related technology comes to rescue for some of the deforesting, equivalent of 27 soccer pitches every minute, going on in the world.

Saving the forests with wireless tech

<p>Small manufacturers play a critical role in keeping the world’s supply chain moving, yet adoption of Industrial IoT solutions by the group is lagging. What can be done?</p><p>Industrial IoT is already bringing huge benefits to multinational manufacturers who have made the investment. But how can small-and-medium sized enterprises (SMEs) with fewer resources start to benefit?</p><h2>IoT adoption rate lags in small businesses</h2><p>If you look at any of the surveys done on Industrial IoT adoption, they all tell a similar story. Deployment of operational IoT solutions in large companies is around 20%, and in some industries as high as 80%. But in small to medium enterprises (SMEs), adoption is closer to 10% with low levels of general awareness reported.</p><p>In 2016, 94% of US manufactured exports came from SMEs at an export value of approximately USD $180 billion. 98% of US-based manufacturing firms have less than 500 employees, while three-in-four employ less than 20.</p><p>This matters, because the health of some of the world’s biggest manufacturing companies depends on the health of this complex network of SME suppliers. While an aircraft manufacturer may implement the latest IoT technology throughout its assembly plant, they are still reliant on many smaller suppliers across the world.</p><p>Unplanned downtime can cripple a small manufacturer who may not have adequate backup equipment or spare capacity to make up for the loss of production. In industries that embrace a lean approach to manufacturing, such failures can quickly propagate throughout the supply chain.</p><p>Given the cost pressures faced by many SME’s, Industrial IoT solutions with a proven ROI should be an easy sell. So why aren’t they?</p><h2>A lack of resources to focus on IoT</h2><p>Building up Industrial IoT expertise in a small company from scratch is at best an extremely difficult task. SMEs don’t possess the same financial or human resource options as multinationals. With little background in the connected world, such SMEs don’t know what they don’t know.</p><p>With this in mind, there is a risk of a growing divide opening up between the haves and the have-nots when it comes to Industrial IoT.</p><h2>The problem with solution providers</h2><p>Some full service providers market themselves toward the SME market. And the benefits are compelling: Access to expertise, ready-to-use solutions and so on. But there are significant downsides, in particular the high cost of convenience.</p><p>Full service providers are not necessarily independent. SMEs run the risk of unwittingly tying themselves into a system or platform that may not last the course.</p><p>A quick web search is all you need to see the rapid rise of solution providers peddling IoT solutions to smaller enterprises. But many of these are merely simple web shops, with little or no expertise behind them.</p><p>This compounds the problem. Even an SME with basic knowledge of IoT doesn’t have the time or skills to distinguish the good from the not-so-good. For someone without IoT knowledge, the task quickly becomes impossible.</p><h2>Are company builders the solution?</h2><p>Outsourcing Industrial IoT development to a specialist ‘company builder’ is an intriguing new alternative. The concept involves a collaborative approach to innovation and value creation. Both the risks and the benefits are shared between the existing company and a group of entrepreneurs.</p><p>The person (or more likely, team) has the core competencies required, including:</p><ul><li>Overview of current technology standards and trends</li><li>Overview of current methodologies</li><li>Knowledge of current implementation best practices</li><li>Overview of current products, and those in the pipeline</li><li>Experience with entrepreneurship and launching new ventures</li><li>Marketing and sales expertise from other projects</li></ul><p>That said, the company builder should also have knowledge of the unique challenges of SMEs and preferably the specific market.</p><h2>Tackling the big problems of small business</h2><p>There’s a real need for Industrial IoT solution providers that work for SMEs. The market potential is there. While the revenues are likely to be lower, SME’s tend to operate on a much leaner model. This means quicker decision-making, faster time to deployment, and improved cash flow.</p><p>Industrial IoT solutions need to be made easier to understand and easier to access for SMEs because of their crucial role in the world’s supply chain.</p><p>What such companies need most are monitoring systems that can identify issues early, and/or cost-cutting systems such as a smart lighting system. This can become the backbone for a mesh network that eases their path to a more connected facility in the future.</p><p>While large multinationals can afford the big consulting companies, they are not feasible for the typical SME. A consultant, solution provider or company builder that can offer these kind of services with a business model that works for small business could see massive returns.&nbsp;</p><hr><p>This article was first published on Nordic's&nbsp;<a href="https://blog.nordicsemi.com/getconnected" rel="nofollow" target="_blank"></a><a href="https://blog.nordicsemi.com/getconnected?utm_campaign=2021%20wevolver" rel="noopener noreferrer" target="_blank">Get Connected Blog</a>.</p>

Small manufacturers play a critical role in keeping the world’s supply chain moving, yet adoption of Industrial IoT solutions by the group is lagging. What can be done?

How SMEs Can Get Started With Industrial IoT

<p>By 2050, seven out of ten people are expected to live in a city. The smart city concept offers the promise of a healthier, safer, and greener lifestyle than that enjoyed by today's urban population.&nbsp;A future built on interconnected smart cities is one with enormous potential for all. To date, the evolution of this concept has been slowed by inevitable roadblocks – primarily the prohibitive infrastructure cost.</p><h2>Smart cities: dream or reality?</h2><p>Smart cities are difficult to fund and complex to build. In the case of air quality monitoring infrastructure, for example, calculating the impact of clean air on health services is no simple task, and as a result, difficult to justify in a budget. Consequently, the concept of integrated city-wide digital operations has remained mostly out of reach, but that situation is now changing.</p><p>We have the technology to make smart cities a near-future reality. Cellular 4G networks are already in place with dense cell tower coverage over most cities in the world. NB-IoT and LTE-M typical use cases will feature low activity, making it possible to connect several tens of thousands of devices to a single base station. Moreover, cell towers host several base stations, multiplying deployment density and promoting massive scale.</p><h2>Street lighting for scalability</h2><p>To reach their full potential, cities will need to piggyback on street lights' power and scale. Street lighting affords a platform to make cities smart and unite multiple applications on a grander scale for the first time.</p><p>Cities may achieve unprecedented levels of productivity, efficiency, and intelligence about city functions and environmental conditions. All from converting street lights to power-efficient LED bulbs, plugging-in cellular IoT connectivity, and adding useful supporting devices such as air quality sensors.</p><p>Soon, the next vital stages of advanced air quality analysis and congestion management, among other desirable use cases, should be attainable.</p><h2>Switching to LED</h2><p>According to research by smart infrastructure analyst Northeast Group, around 90 percent of the 363 million street lights globally will use LEDs by 2027.</p><p>Lighting and sensor makers suggest smart lighting delivers a 50-70 percent reduction in the management and maintenance costs associated with the infrastructure. Around half of those savings will come from just switching to LED. The rest comes from connecting and controlling the luminaires and delivering intelligence sparks about their working status across the lighting network.</p><h2>Dual-mode IoT connectivity</h2><p>Various applications make various demands in terms of coverage, throughput, and security; thus, street lighting has developed around unlike technologies in different markets.</p><p>The combination of different IoT networking technologies in street lighting platforms further expands smart cities' potential scope and impact. Short-range technologies – such as Bluetooth LE, interoperable with almost all smartphones on the market – are cheaper and offer greater throughput for developing 'lighting as-a-Platform' (LaaP) setups.</p><p>Short-range technologies provide robustness that make them well-suited for meshing neighborhood-wide sensors, but a denser mesh has its caveats:</p><ul><li>Architectural complexity</li><li>Higher energy demand on interconnected point-to-point sensors</li><li>Range limited to a couple of hundred meters,</li><li>The need for a gateway to get the signal back to the Cloud.</li></ul><p>Integrating a cellular connection provides range and simplicity. Smart lighting providers' trend is to go with point-to-Cloud cellular, offering coverage of five-to-15 kilometers from gateways or sensor devices.</p><h2>Advantages of tighter integration</h2><p>By embedding twin short- and long-range radios in the same hardware, LaaP developers can bring tighter technological integration and lower acquisition costs to their design – which are prime considerations for smart cities.</p><p>Nordic provides multi-mode short- and long-range products ideally suited to supporting applications for interconnected smart cities. The&nbsp;<a href="http://nordicsemi.com/Products/Low-power-short-range-wireless/nRF52840?utm_campaign=2021%20wevolver" rel="noopener noreferrer" target="_blank">nRF52840 System-on-Chip (SoC)</a> supports Bluetooth LE, Bluetooth mesh, and Zigbee, as well as Thread and proprietary 2.4 GHz systems.</p><p>Nordic's cellular-based <a href="https://www.nordicsemi.com/Products/Low-power-cellular-IoT/nRF9160?utm_campaign=2021%20wevolver" rel="noopener noreferrer" target="_blank">nRF9160 SiP</a> meanwhile offers both LTE-M and NB-IoT. The Nordic Thingy:91 prototyping platform combines both the nRF52840 SoC and nRF9160 SiP, allowing developers to immediately prototyping smart cities and LaaP solutions.</p><hr><p>This article was first published on Nordic's&nbsp;<a href="https://blog.nordicsemi.com/getconnected" rel="nofollow" target="_blank"></a><a href="https://blog.nordicsemi.com/getconnected?utm_campaign=2021%20wevolver" rel="noopener noreferrer" target="_blank">Get Connected Blog</a>.</p>

By 2050, seven out of ten people are expected to live in a city. The smart city concept offers the promise of a healthier, safer, and greener lifestyle than that enjoyed by today's urban population. A future built on interconnected smart cities is one with enormous potential for all.

Street lights to support interconnected smart cities

<p>In the drive towards a carbon-neutral world, households and businesses could, and should, be looking at hydrogen as a source of clean energy. And now, thanks to the rise of advanced smart metering based on cellular IoT wireless connectivity, this ambition is no longer just a pipedream.</p><p>As part of the campaign for greater action to address global climate concerns, there's a growing push to transition gas network infrastructure towards net-zero carbon emissions. Key to reaching this goal is the 'green gas' hydrogen. It burns when combined with atmospheric oxygen to release its energy with only water as a byproduct - no soot, no nitrous oxides, and perhaps best of all, no carbon dioxide with its potential for environmental warming.</p><h2>Cutting carbon emissions</h2><p>When combusted, hydrogen produces no carbon emissions. It is the simplest and most common element in the universe. On earth, it can be cleanly obtained in abundance in two ways:</p><ul><li>Through renewable-energy-powered electrolysis of water (the process of using electricity to split water into oxygen and hydrogen gas)</li><li>By "reforming" methane, with the carbon dioxide byproducts captured, compressed, and stored.</li></ul><p>The combination of zero-carbon production of hydrogen and its clean-burning nature would drastically cut carbon emissions and help protect the planet.</p><h2>A massive worldwide transition</h2><p>Hydrogen yields the most energy per unit volume of any fuel source. But this, in turn, means it is highly flammable, which has tended to cause safety concerns among consumers. For a massive worldwide transition to hydrogen, stakeholders must first overcome public skepticism and educate the consumer that risks are very low with the right safety protocols in place.</p><p>But for now, the plan determined by several developed nations is to introduce hydrogen by stealth, initially blending up to 23 percent of the gas with conventional fuel to curb carbon emissions, with no discernible change for consumers. When used on a gas stove, for example, the user is unaware that hydrogen is in the mix. Combining up to 23 percent of hydrogen is an acceptable and realistic target. A higher amount would call for greater investment in infrastructure, in part to prevent leaking of the ultra-lightweight gas.</p><h2>The hydrogen smart meter economy</h2><p>In the U.K. and other countries, pilot schemes with hydrogen blending are underway as a means to meet carbon-emission targets. One challenge is that no domestic hydrogen gas meters are currently available to measure and charge for gas usage.</p><p>Conventional meters measure gas flow using a mechanical system based on a diaphragm. Because of the small hydrogen molecular size, hydrogen diaphragm meters would need to be about three times the conventional meter size – an impractical and costly limitation.</p><p>One option is to replace mechanical diaphragm meters with ultrasonic devices. These rely on accurate 'Time of Flight' (ToF) measurement of the speed of sound to determine the gas's flow rate. But because the speed of sound in hydrogen is around three times faster than that in natural gas, significantly better timing precision is needed.</p><h2>New technology, new technique</h2><p>In response to this challenge, U.K.-based MeterTech has launched a smart meter that, unlike conventional gas meters with diaphragms, uses advanced ultrasonic functionality to make hydrogen metering practical.</p><p>The new MeterTech smart meter not only uses a more advanced technique for the "recovery" of the ultrasonic signals in hydrogen, but it also has timing circuits capable of measuring ToF in hydrogen. Furthermore, the meter has been designed to meet the highest safety requirements in Europe and elsewhere in the world.</p><h2>Cellular IoT benefits</h2><p>All smart meters require robust and reliable connectivity. And because hydrogen meters will be the most advanced domestic meters available to date, they need to incorporate the latest, ground-breaking data communications, according to Eric Beattie, CTO of MeterTech.</p><p>For the MeterTech hydrogen smart meter, the company has integrated cellular IoT—an LPWAN technology supported by two versions, LTE-M and NB-IoT—for wireless connectivity. Cellular IoT represents the simplest, most secure, standardized, and future-proofed connectivity solution for the smart metering sector.&nbsp;</p><p>The technology also provides the most reliable Cloud connectivity from the meter to the utility. And by eliminating manual meter reading, accurately billing customers, initiating dynamic pricing, monitoring grid losses, and optimizing grid performance, utilities and governments can achieve optimal efficiency and maximize profit potential. They can also rely on this robust wireless technology for large scale IoT smart metering deployments accelerating rollouts.</p><h2>Nordic-enabled cellular connectivity</h2><p><a href="https://www.nordicsemi.com/Products/Low-power-cellular-IoT/nRF9160?utm_campaign=2021%20wevolver" rel="noopener noreferrer" target="_blank">Nordic's nRF9160 System-in-Package (SiP)</a> is integrated into the MeterTech product to provide cellular connectivity enabling utilities to remotely check what's happening at the meter and gather accurate information over an extended period.</p><p>Traditional ultrasonic-based meters consume, on average, around 30 to 40 microamps. The nRF9160 has been designed from the ground up for low power consumption to extend battery life. This partly contributes to MeterTech succeeding in reducing the average current of its meter to less than 15 microamps—even though MeterTech's new technique is computationally much more powerful, according to Beattie. Given a meter needs to operate for around two decades on a single battery, every microamp of continuous average power saved is vital.&nbsp;</p><hr><p>This article was first published on Nordic's&nbsp;<a href="https://blog.nordicsemi.com/getconnected" rel="nofollow" target="_blank"></a><a href="https://blog.nordicsemi.com/getconnected?utm_campaign=2021%20wevolver" rel="nofollow" target="_blank">Get Connected Blog</a>.</p>

In the drive towards a carbon-neutral world, households and businesses could, and should, be looking at hydrogen as a source of clean energy. And now, thanks to the rise of advanced smart metering based on cellular IoT wireless connectivity, this ambition is no longer just a pipedream.

Cellular IoT smart meter boosts hydrogen energy sector

<p><span dir="auto">This webinar was a Nordic Partner webinar hosted by Fractus Antennas and gives an introduction to Embedded Antennas for IoT. &nbsp;&nbsp;</span></p><p><span dir="auto">Questions and answers from the live session and a PDF with slides have been posted online: <a href="https://devzone.nordicsemi.com/nordic/nordic-blog/b/blog/posts/q-a-for-embedded-antennas-for-iot-webinar?utm_campaign=2021%20wevolver" rel="noopener noreferrer" target="_blank">https://devzone.nordicsemi.com/nordic/nordic-blog/b/blog/posts/q-a-for-embedded-antennas-for-iot-webinar?utm_campaign=2021%20wevolver</a></span></p><p><span dir="auto">Content: - Practicalities - Nordic Thingy:91 Brief - Fractus Antennas presents: Embedded Antennas for IoT&nbsp;</span></p><p><span dir="auto">The webinar was live October 17 at 10 AM&nbsp;</span><span dir="auto">CEST.&nbsp;</span><a dir="auto" href="https://www.youtube.com/redirect?q=https%3A%2F%2Fwebinars.nordicsemi.com%2Fwebinars&redir_token=QUFFLUhqbkVZNTh3eV94c0RtTzNHa0hqYUFoUlc0eV8zZ3xBQ3Jtc0tudHFaQURaekFGdjVtSTloR0xseGpqRGZiOWxueS1EZmlJTWpMTDZybGdMb2lXSHRER196NHJ0amF6U0dFa0o4eVc5V0lvcE5kcWZZVHlmWW4wWk9UVm0yaFBlSlMwRUJfeC1JQ2RhNnEyVi1fczgxYw%3D%3D&v=rjPd0sFuAyU&event=video_description" rel="nofollow" spellcheck="false" target="_blank">https://webinars.nordicsemi.com/webinars</a><span dir="auto">?utm_campaign=2021%20wevolver</span></p><p><span dir="auto">For more info about Nordic Thingy:91 go to:&nbsp;</span><a href="https://www.nordicsemi.com/Software-and-Tools/Prototyping-platforms/Nordic-Thingy-91" rel="noopener noreferrer" target="_blank">https://www.nordicsemi.com/Software-and-Tools/Prototyping-platforms/Nordic-Thingy-91?utm_campaign=2021%20wevolver</a></p><p><span dir="auto">For more info about Fractus Antennas go to:&nbsp;</span><a dir="auto" href="https://www.youtube.com/redirect?q=https%3A%2F%2Fwww.fractusantennas.com%2F&redir_token=QUFFLUhqbGZIMVNwdng1ZlJPQW10WWlIV0c1UHBQNURDQXxBQ3Jtc0tsUTE3WGdDRXk4clA2TlJ5QTI2X3BIRURNaGRBNEdYTUpJZHlyendtSzlFNERWTmY5Y2lqQ09xUlAwRl9RUVAycEJtREhOUUhPZjdIYUdxVXA5cXptTWRwQ1p2djZneXA2X3ctLXo3a3ctOHhOWlRHOA%3D%3D&v=rjPd0sFuAyU&event=video_description" rel="nofollow" spellcheck="false" target="_blank">https://www.fractusantennas.com/</a></p>

This webinar was a Nordic Partner webinar hosted by Fractus Antennas and gives an introduction to Embedded Antennas for IoT.

Embedded Antennas for IoT

<h1 dir="ltr">Internet of Things</h1><p dir="ltr">IoT or the Internet of Things is the term used to describe the network of physical objects or &ldquo;things&rdquo; that are embedded with sensors, software, and other technologies for connecting and exchanging data with other devices and systems over the Internet. The low cost of chips and the availability of wireless networks mean almost anything can become part of the <a href="https://www.wevolver.com/article/why-ai-is-set-to-reshape-the-iot" rel="noopener noreferrer" target="_blank">IoT network</a>. Adding sensors to otherwise &lsquo;dumb&rsquo; objects gives them a level of intelligence that can be used to build smart systems such as your connected home, automated workspaces, or environmental sensing systems. However, sensors are only one part of an IoT system which also requires a network to connect and exchange data and a method of interpreting and reading that data, which is done by a software application.&nbsp;</p><h2 dir="ltr">IoT networks</h2><p dir="ltr">IoT is possible thanks to the advent of wireless networks. There are many different network types, each network type has particular properties and characteristics suitable for different IoT scenarios. The current maturity and continued development as well as the deployment of this array of networks directly reflect what types of IoT scenarios can be implemented and where.&nbsp;</p><p dir="ltr">These network types range from low power cellular networks such as NB-IoT and long-range radio networks such as LoRaWAN, Sigfox to local personal networks such as Bluetooth and RFID. The selection of a connectivity option depends on the IoT applications it is serving, with each requiring drastically different functions. The IoT connectivity option that is amazing for one application may be unsuitable for another.&nbsp;</p><p dir="ltr">The majority of IoT devices, especially those in smart city and industrial sectors, don&rsquo;t require the same speed and bandwidth of consumer cellular devices to communicate. However, they do have other specific requirements such as longevity and low power consumption.&nbsp;</p><p dir="ltr">This article focuses on Low Power Wide Area Networks (LPWAN). It will discuss the specific type of networks in use under this umbrella term and how LPWANs are integral to the continuing innovation and relevant application of IoT.</p><h1 dir="ltr">LPWAN Technology Overview</h1><p dir="ltr">LPWANs are designed to allow long-range communications at a low bit rate. &nbsp;LPWAN is not a standard but a broad term encompassing various implementations and protocols, both proprietary and open-source, that share common characteristics, namely low power, and wide area. It is important to note these properties in comparison to other wireless network protocols such as local RF networks that include Bluetooth and RFID, cellular M2M, or mesh networks like ZigBee.</p><p dir="ltr"><img class="fr-fic fr-dib" src="https://images.wevolver.com/eyJidWNrZXQiOiAid2V2b2x2ZXItcHJvamVjdC1pbWFnZXMiLCAia2V5IjogImZyb2FsYS8xNjA1NzEzNTM0NzE2LXNtYWxsIGdyYXBoLnBuZyIsICJlZGl0cyI6IHsicmVzaXplIjogeyJ3aWR0aCI6IDk1MCwgImZpdCI6ICJjb3ZlciJ9fX0=" style="width: 788px;"></p><p dir="ltr">LPWANs are useful for IoT scenarios where devices need to send small data over a wide area while maintaining battery life for an extended period of time. There are several connectivity technologies in the LPWAN space the most prominent include:</p><p dir="ltr"><img class="fr-fic fr-dib" src="https://images.wevolver.com/eyJidWNrZXQiOiAid2V2b2x2ZXItcHJvamVjdC1pbWFnZXMiLCAia2V5IjogImZyb2FsYS8xNjA1NzE0OTkwODQyLXRhYmxlIGFrZW56YS5wbmciLCAiZWRpdHMiOiB7InJlc2l6ZSI6IHsid2lkdGgiOiA5NTAsICJmaXQiOiAiY292ZXIifX19" style="width: 788px;"></p><h1 dir="ltr">Why do LPWANS matter to IoT?</h1><p dir="ltr">LPWANs are critically important to the overarching concept of IoT and its expansion. In many ways LPWANs are the backbone of a value-driven IoT network as they provide the opportunity for devices that usually don&rsquo;t have sensors or their own power supply, &nbsp;to become truly connected. By transforming almost anything into a data-gathering device with low cost, low power sensors, and using LPWANs to connect and distribute this data, smart networks are created that can improve efficiency, data collection, industrial process and be used in research and environmental monitoring - meeting many of the goals of the IoT vision.&nbsp;</p><p dir="ltr">Consumer IoT relies on the high-speeds of 4G and <a href="https://www.wevolver.com/article/an-introduction-to-5g" rel="noopener noreferrer" target="_blank">5G wireless services</a> to provide applications such as connected smart homes, remote car retrieval, and entertainment. Often the items in these IoT networks are already relatively high tech, have their own power supply, and can have some information retrieved from them regardless of their connection. These networks are in frequent use with high bandwidth.</p><p dir="ltr">In contrast, LPWANs offers IoT applications the opportunity for devices to send small amounts of data periodically over remote networks that span large distances and use battery-powered devices that need to last many years. The devices in these networks can include typically non-powered, non &lsquo;smart&rsquo; items such as agricultural gates, bins, bridges, or walkways.&nbsp;</p><p dir="ltr">LPWANs achieve their range and low power capabilities by only sending small packets of information periodically upon waking from an external trigger or at a pre-programmed interval.&nbsp;</p><h2 dir="ltr">Opportunities and challenges</h2><p dir="ltr">The advantages of LPWANs are clear in applications where networks require periodic or inconsistent data transfer over long distances for a significant amount of time. Due to the range and simplicity of LPWAN data packets, sensors can even report from underground, in challenging climates, and from significant distances from gateways or towers.&nbsp;</p><p dir="ltr">However, there are limitations to LPWANS, they are unsuitable for use cases requiring data to be transferred frequently and/or in large volumes. So detailed images, videos or lengthy audio is unsuitable to be transferred over an LPWAN.&nbsp;</p><p dir="ltr">Some non-cellular LPWANs may also be unsuitable for applications where the transfer of data is considered &lsquo;mission critical&rsquo; due to possible interference due to band sharing. Open-air LPWAN devices such as something on top of a tower that is sending faint signals in sub-GHz open bands may encounter interference from high-energy signals operating just above the GHz boundary.</p><p dir="ltr">Non-cellular LPWANs sometimes use unlicensed bands - these are the bands left open by government for industrial, science, and medicine so they must share access to the radio spectrum with other devices. This sharing requires regularization to ensure fair spectrum access for all devices and to prevent potentially harmful interference as described above.&nbsp;</p><p dir="ltr">Cellular LPWAN networks are inside the licensed spectrum and are offered by mobile network operators. They were developed with the 3rd Generation Partnership Project&nbsp;(3GPP)&nbsp;and GSMA.</p><p dir="ltr"><strong id="isPasted">Recommended reading:&nbsp;</strong><span style="background-color: transparent;"><a href="https://www.wevolver.com/article/critical-asset-monitoring-with-lpwan" rel="noopener noreferrer" target="_blank">Critical Asset Monitoring with LPWAN</a></span></p><h2 dir="ltr">Maturity of the technology</h2><p dir="ltr">With the advent of the cellular LPWAN, there is now more flexibility in the definition of &ldquo;low-power&rdquo; and &ldquo;wide-area&rdquo;. &nbsp;As the options for connectivity technology grow it&rsquo;s essential there isn&rsquo;t a temptation for one choice to be predicted as a winner. Different technologies suit different applications and the industry must take care to avoid the hype around particular protocols diluting a nuanced approach with a view to local contexts.</p><p dir="ltr"><img class="fr-fic fr-dib" src="https://images.wevolver.com/eyJidWNrZXQiOiAid2V2b2x2ZXItcHJvamVjdC1pbWFnZXMiLCAia2V5IjogImZyb2FsYS8xNjA1NzE1MTA0MzI5LWJpZyBncmFwaCBmb3IgYWtlbnphLnBuZyIsICJlZGl0cyI6IHsicmVzaXplIjogeyJ3aWR0aCI6IDk1MCwgImZpdCI6ICJjb3ZlciJ9fX0=" style="width: 788px;"></p><h2 dir="ltr">Market share &amp; European landscape</h2><p dir="ltr">Critical to IoT success is the wide deployment of LPWAN networks. LoRa has been rapidly deployed in most countries in the world, with<a href="https://lora-alliance.org/" target="_blank">&nbsp;all but one European country</a> connected. Sigfox has strong coverage in western Europe and has <a href="https://www.sigfox.com/en/coverage" target="_blank">rollout plans</a> for Scandinavia, Austria, and Poland as well as North and South America. NB IoT and LTE enjoy wide global coverage, <a href="https://www.gsma.com/iot/deployment-map/" target="_blank">this map</a> shows the mobile IoT commercial launches to-date.&nbsp;</p><p dir="ltr">In Europe, IoT solutions where cross-border networks are likely to be common such as logistics and waste management there need to be particular attention paid to the network type and provider to ensure the continuity of the relevant LPWAN network. In this regard, Sigfox enjoys an advantage as it is a single product operated by a single company with approximately 60 countries partially or totally covered to date. &nbsp;</p><p dir="ltr">LoRaWAN networks, on the other hand, are operated by different private companies in each European country, for example, Belgium and Luxembourg have Proximus, in Switzerland its Swisscom, the Netherlands uses KPN. In France, there are two operators, Orange and Bouygues while the UK has Connexin and IOT Networks. Germany has a choice between 3 public operators Digimondo, Minol Zenner, and Netzikon. Italy has 4 operators: A2A Smart City, Axatel, Telemar, and Unidata. But despite the rich landscape, most of these operators are not interconnected which may make it difficult for a sensor to switch from one network to another. This potential issue with continuity of connection is being addressed through roaming agreements between providers and with regulation at a national level.&nbsp;</p><h1 dir="ltr">LPWANs IoT case studies</h1><h2 dir="ltr">LoRaWAN</h2><p dir="ltr">A common application for LoRaWAN is the Smart Facility / Smart Building field. ISS, one of the world&rsquo;s leading facility services companies is embracing the use of IoT based smart facility services across their services. For example, smart facility management enables ISS customers to track and analyze workspace occupancy therefore optimizing space management. By installing occupancy sensors in buildings, it becomes possible to improve the management of workspaces with new digital solutions. Infrared sensors mounted under personal desks or in meeting rooms allow property and office managers to see at a glance which areas are used more frequently than others (Figure 4). This type of system will become increasingly important as room occupancy rates and contact tracing become essential knowledge in a post-pandemic world.&nbsp;</p><p dir="ltr"><span class="fr-img-caption fr-fic fr-dib" style="width: 788px;"><span class="fr-img-wrap"><img src="https://images.wevolver.com/eyJidWNrZXQiOiAid2V2b2x2ZXItcHJvamVjdC1pbWFnZXMiLCAia2V5IjogImZyb2FsYS8xNjA1NzE1MTgzNzg0LWRlc2sgb2NjdXBhbmN5X2FrZW56YS5qcGciLCAiZWRpdHMiOiB7InJlc2l6ZSI6IHsid2lkdGgiOiA5NTAsICJmaXQiOiAiY292ZXIifX19"><span class="fr-inner" spellcheck="false">Figure 4.</span></span></span></p><p dir="ltr">Other smart building applications include:</p><p dir="ltr"><strong>Air monitoring</strong></p><p dir="ltr">Air monitoring uses stationary or mobile IoT-enabled sensors to map and monitor the air quality inside high-occupancy spaces such as schools and hospitals. This data can inform cities of building management to make better decisions about ventilation, hygiene, occupancy rates, or similar.&nbsp;</p><p dir="ltr"><strong>Smart cleaning&nbsp;</strong></p><p dir="ltr">Smart cleaning applications might include services such as those offered by Soobr which offers clients the ability to plan cleaning choreographies based on levels of higher use and urgency. &nbsp;In these instances, LoRa is a solid LPWAN choice as the data being transferred has a low bandwidth and long-distance. LoRa also allows the user to set up their own network in situations where data security or ownership are critical.&nbsp;</p><h2 dir="ltr">Sigfox</h2><p dir="ltr">Sigfox is suitable for a variety of applications. International company <a href="http://ambergloglay.com/" target="_blank">Amberg Loglay</a> uses Sigfox to develop service on demand systems for logistics on construction sides. On large construction sites with many moving parts and large numbers of workers, there may be several freight lifts on the site, each of which must be operated by lift attendants. With a service-on-demand system, there is not one attendant for each lift but one person operates several. The button solution serves to call the lift attendant/driver to the respective lift. While simple enough sounding this can provide huge efficiency increases on large infrastructure and building projects.&nbsp;</p><p dir="ltr">As Sigfox is operated by a private enterprise it is excellent for larger businesses who need completely remotely managed networks. A critical deciding factor between Sigfox and LoRaWAN is the network coverage. In some areas, you may have one but not the other, which will influence the decision-making process.&nbsp;</p><h2 dir="ltr">NB-IoT</h2><p dir="ltr">NB-IoT is used in solutions for waste disposal companies, utilities, and transport companies. By attaching sensors with low energy consumption and long service life to the waste containers, it is possible to centrally monitor and predict the container&#39;s filling level. This enables improved planning of disposal tours and thus increases the efficiency of waste management. A concept study completed by Swiss logistics company Live Track aimed to evaluate the savings and potential of an IoT based waste monitoring system. The study which monitored glass collection in west-central Switzerland showed huge benefits in cost reduction, environmental impact, and collection efficiency. The total distance traveled by waste collection vehicles fell from over 800 km to approximately 540 km (over 30% decrease) in the evaluated period of 16 weeks while the estimated costs of the system were reduced by approximately 30%. Full details of the study can be read&nbsp;<a href="https://akenza.com/blog/optimized-logistics-in-glass-recycling-through-smart-waste-solution-a-case-study-by-live-track-akenza?utm_source=wevolver&utm_medium=referral&utm_campaign=wevolverblog&utm_content=lpwanblogpost" rel="noopener noreferrer" target="_blank">here.</a></p><h2 dir="ltr">LTE Cat-M1</h2><p dir="ltr">This network technology is extremely energy-efficient and also supports mobile cell handover for non-static applications and voice functionalities (VoLTE). LTE-M has been implemented into several life-saving medical applications. For example in AED-defibrillators, the portable devices used to deliver a safe electric shock to a person having cardiac arrest - when the device is used, perhaps at a public beach or swimming pool, a message is automatically sent to the relevant medical authorities. LTE-M has also been used in mobile telemetry systems that require high reliability such as voice-controlled emergency wristwatches that need to be able to transmit small data volumes in critical timeframes. When the wearer utters a trigger word, the device sends an alert message.&nbsp;</p><h1 dir="ltr">Future adoption</h1><p dir="ltr">LPWANs will continue to emerge as the fastest-growing IoT communication technology with LoRa and NB-IoT currently the most commonly used. However, for the continued expansion of IoT especially into new fields, services, and industries without previous IoT or other networks systems smart software solutions are essential. &nbsp;Users planning IoT projects are currently faced with a fragmented market that will usually require a diligent vendor to offer solutions that focus on the value of the system over the hype of the IoT news machine.&nbsp;</p><p dir="ltr">Services must make clear the true cost, maintenance, and value of IoT for their business in order to reduce IoT project failure. There is a huge opportunity for IoT software infrastructure to assist in the bridge between the user, no matter the size and the value proposition of the IoT system.</p><h2 dir="ltr">Ensuring a focus on value</h2><p dir="ltr">The reason many IoT projects fail is the failure to understand how such a system can bring value. Enterprise software companies are going a long way to simplify IoT for businesses of all sizes and from all sectors. Zurich based company, Akenza, addresses the complexity of IoT by offering a completely agnostic IoT platform. Customers select the products, networks, storage and processing location, and software interface that is right for them independent of other commercial influences. <a href="https://akenza.com/?utm_source=wevolver&utm_medium=referral&utm_campaign=wevolverblog&utm_content=lpwanblogpost" rel="noopener noreferrer" target="_blank">Akenza solutions</a> work with LoRa, Sigfox, NB-IoT, and LTE-M LPWANs, and can use off the shelf, or custom-built software applications.&nbsp;</p><p><img class="fr-fic fr-dib" src="https://images.wevolver.com/eyJidWNrZXQiOiAid2V2b2x2ZXItcHJvamVjdC1pbWFnZXMiLCAia2V5IjogImZyb2FsYS8xNjA1NzE1Mzk3OTM2LWFrZW56YSBpbGx1c3RyYXRpb24gcHVibGlzaGVkLmpwZyIsICJlZGl0cyI6IHsicmVzaXplIjogeyJ3aWR0aCI6IDk1MCwgImZpdCI6ICJjb3ZlciJ9fX0=" style="width: 788px;"></p><h1 dir="ltr">Conclusion</h1><p dir="ltr">The convergence of the trend towards sensors increasing in quality while decreasing in size and cost with the expansion of the spectrum of LPWAN protocols mean IoT networks can provide real value to enterprises, communities, and cities. The use of low power and wide range networks with smart end-user software solutions opens the possibilities of the potential of IoT to more and more applications and more people. The future of IoT based on long battery-life devices connected by low power wide range networks is bright and will only be enhanced by everyone in the ecosystem recognizing its ongoing complexity and dedicating solutions to value.&nbsp;</p><hr><h1 dir="ltr"><em>About the sponsor: Akenza</em></h1><p dir="ltr"><a href="https://akenza.com/?utm_source=wevolver&utm_medium=referral&utm_campaign=wevolverblog&utm_content=lpwanblogpost" rel="noopener noreferrer" target="_blank"><em>Akenza</em></a><em>&nbsp;is dedicated to helping companies build their own smart IoT solutions. The Akenza IoT platform can be seen as a bridge between the physical world and the cloud. With a simple and secure management of smart devices, connectivity, and data, it enables the rapid market introduction of innovative, digital technologies. The technology agnostic architecture avoids technology lock-in and makes it possible to register any type of device via any connectivity technology, process the data, and make it available to any application in the cloud.&nbsp;</em></p><p dir="ltr"><em><img class="fr-fic fr-dib" src="https://images.wevolver.com/eyJidWNrZXQiOiAid2V2b2x2ZXItcHJvamVjdC1pbWFnZXMiLCAia2V5IjogImZyb2FsYS8xNjA1NzE1NjA4MjQwLWxvZ28gYmFubmVyIGJvdHRvbSBha2VuemEuanBnIiwgImVkaXRzIjogeyJyZXNpemUiOiB7IndpZHRoIjogOTUwLCAiZml0IjogImNvdmVyIn19fQ==" style="width: 788px;">&nbsp;</em></p>

IoT hinges on the ability to send periodic data long distances with low power.

What are LPWANs and why do they matter to IoT?

<h2></h2><h2 data-element-id="headingsMap-5-0">Introduction</h2><p>Groq is on a mission to build the computer for the next generation of high-performance machine learning. Founded by former Google engineer, Johnathan Ross the company has doubled in size since its inception and is on track to double its headcount again by the end of 2021. In late September, Ross presented as part of the online AI summit 2020. His keynote gave key insights on the industry, the market, and the Groq architecture capable of producing the world’s first PetaOp chip. We’ve pulled out some of the highlights from the presentation.</p><h2 data-element-id="headingsMap-6-0">An answer to the end of Moore’s Law</h2><p>In 1971, Intel co-founder Gordon Moore predicted that the number of transistors we can pack into a microchip would double every 18-24 months. A statement that has proved itself true for a time, but as Moore himself noted “the nature of exponentials is that you push them out and eventually disaster happens”. Computing now finds itself on a precipice where cost, power, and efficiency compete for priority to deliver the computing ambitions of the next generation.</p><p>Rock’s Law or Moore’s second law states that the cost of a semiconductor chip fabrication plant doubles every four years. In line with this prediction, next-generation factories are estimated to cost more than $20 billion.</p><p>Realizing the enormity of what these two laws mean for the future of computing placed Jonathon Ross, <a href="https://groq.com/" rel="noopener noreferrer" target="_blank">CEO, and Founder of Groq</a>, on his current path.</p><p>Ross recounts when working for Google back in 2012, Jeff Dean gave a presentation to the leadership team, containing just two slides. The first slide was good news; machine learning finally works. The second slide was bad news; It would require another 20 to 40 data centers to be able to afford it - and that was <em>just</em> for speech recognition. At the end of the presentation, Ross started the TPU (Tensor Processing Unit) at Google.</p><p>Fast forward several years and Ross, now the head of Groq, unveils the world’s first and only PetaOp chip. The chip’s performance is equivalent to one quadrillion operations per second, or 1e15 ops/s. Groq's architecture is also capable of up to 250 trillion floating-point operations per second (FLOPS).</p><p><img src="https://images.wevolver.com/eyJidWNrZXQiOiJ3ZXZvbHZlci1wcm9qZWN0LWltYWdlcyIsImtleSI6ImZyb2FsYS8xNjAzODIwODc2NzkxLUdyb3EgYmFubmVyLmpwZyIsImVkaXRzIjp7InJlc2l6ZSI6eyJ3aWR0aCI6OTUwLCJmaXQiOiJjb3ZlciJ9fX0=" style="width: 300px;" class="fr-fic fr-dib"></p><h2 data-element-id="headingsMap-7-0">Architecture over speed</h2><p>“The industry often focuses on faster speeds,” begins Ross. “That’s what so many chip makers are focused on today. Everyone in this field is trying to make a faster chip. That approach of throwing more transistors onto a die means cost increases with incremental performance improvements. Following a trend that is running out of steam isn't a win for customers who are planning for the future. What we've done is come up with something revolutionary. I call this deterministic or predictable computing. A radically simplified approach which is scalable for customers. This is what enabled us to create the first and only PetaOp processor.”</p><h2 data-element-id="headingsMap-8-0">The shortest path to performance</h2><p>“How did we get there?” Ross reflects at the start of his keynote address. &nbsp;“We got there by doing things differently. A good example of this, we built on a 14-nanometer process when others were building on a 7-nanometer process. We simply didn't need the bleeding edge lithography to produce the fastest chip. Instead of adding more transistors, our performance came from our architecture. Everyone here believes Moore's law will run out. We talk about it and no one is acting any differently than they did before.”</p><p>Ross asserts that the real question is not the end of Moore’s law but what happens if ML accelerator hardware doesn't get any better. The answer, Ross believes, is in designing and delivering a new architecture not based on the familiar systems of FPGA, CPU, or GPU. “It’s something completely different,” says Ross, “we call it the TSP or the tensor streaming processor.”</p><h2 data-element-id="headingsMap-9-0">Compute as power</h2><p>“Compute is a utility like power,” states Ross. “The country that controls compute will be in an advantaged position.” The future of computing is likely to need a careful mix of private enterprise, government, and academic input, research, and funding to ensure both its <a href="https://www.wevolver.com/article/a-global-collaboration-to-move-artificial-intelligence-principles-to-practice" rel="noopener noreferrer" target="_blank">continued development and shared access</a>. It isn’t only owning or controlling the technology that will give a nation an advantage but rather the overall deliberate direction and ambition around legislation, use, access, and dominance.</p><p>“The future isn't guaranteed,” says Ross, “but you do get the future that you stand up for.” For Ross being deliberate about your approach is key to success. But he also understands this needs to happen across the industry to be effective for as many as possible.&nbsp;</p><p>“At Groq we believe in the shortest path to performance,” continues Ross. “We have to stand up for the future we want by creating new trends with our approach. Trends like:</p><ul><li>Inference being different than training and being bigger than training.</li><li>Batch Size 1</li><li>Determinism and predictability, and</li><li>Performance and software first, which to us means the complexity and intelligence of a task is managed in the software stack.</li></ul><p>“Defining these trends originated with what we’re accomplishing at Groq but more importantly, these are going to enable customers' AI applications in production, that today are simply too costly.”</p><h2 data-element-id="headingsMap-10-0">Hopes and Challenges for the Future of AI</h2><p>Key investors and stakeholders in the AI field joined Ross in the presentation to share their hopes and challenges for the future. &nbsp;Each speaker pointed to where their priorities are for the development of the technology and gave insights into where their organization is directing resources. Speakers included Nicolas Savage, the Managing Director of TDK ventures the, corporate VC arm of TDK, CEO of Canadian quantum computing company, &nbsp;1QBit, &nbsp;Andrew Fursman, &nbsp;Teddy Gleeser, &nbsp;a partner at D1 Capita and CTO of the ASIC business unit at Marvell, Igor Arsovski. Their full remarks can be viewed in the video excerpts below:</p><h2 data-element-id="headingsMap-11-0">Predictability and reliability</h2><p>“I’m personally excited about future advancements like bioengineering, finance, government applications, and especially autonomous vehicles because it will save lives,” shared Ross. &nbsp;“My cousin was a passenger in a car that got into an accident. The driver of the other car had been drinking and he slammed into the side of my cousin's car. This didn't have to happen and if we had autonomous cars it wouldn't have happened. I want our chip in those cars because we need the level of predictability that I know our chip can bring but there's something else that AI brings you. When cars can make deliveries to you when auto-editing makes you a better photographer, when auto-sorting your inbox or when speech recognition keeps an EMT's hands-free so they can focus on their patient in an emergency.”</p><p>“One major challenge in today's AI's systems is when you look at AI accelerators which typically push both performance but at the same time try to drop the voltage as low as possible so they can achieve really high energy efficiency. This tops per watt metric is kind of critical. Furthermore, these accelerators typically get integrated into data centers and automotive applications that have critical uptime and reliability requirements. I know reliability is not as sexy as the tops per watt metric and because of that it's taken a backseat to many AI discussions and many AI conferences but reliability is really critical to deliver useful systems.”</p><h2 data-element-id="headingsMap-12-0">Time is a Limited Resource</h2><p>“Silicon Valley has gotten a lot of criticism for focusing on convenience but I’m not going to apologize for that whether it's a safety issue or convenience time is a limited resource and we only get so much of it. What are the obstacles keeping us from that new reality? You’re probably thinking about speed but that's only part of the equation. Performance is a big challenge but reliability, predictability, and certainty in the real world is key. Chip makers like us need to deliver that performance not just in the lab but also in the real world,” details Ross.</p><p>“Training cost isn't an issue. Train the model once and use it as much as you want. With inference, each response has to be real-time and it costs money and energy that's a scaling issue. Now instead of scaling with the number of ML researchers, you're scaling with the number of users and the number of queries. Inference at scale is harder than training.”</p><p>“When I was at google roughly 80% of the ML models that were trained ended up having to be abandoned because they were too expensive to run in production. AI success is hinged on speed for inference as well as training but for the workloads of the future total cost of ownership is dependent on scalability. Just like when the team and I created the TPU at Google and it saved billions of dollars, we as an industry have just begun tackling these challenges.”</p><p><img src="https://images.wevolver.com/eyJidWNrZXQiOiJ3ZXZvbHZlci1wcm9qZWN0LWltYWdlcyIsImtleSI6ImZyb2FsYS8xNjAzODc0MjkzNzg4LWdyb3EgaW4gYm9keS5qcGciLCJlZGl0cyI6eyJyZXNpemUiOnsid2lkdGgiOjk1MCwiZml0IjoiY292ZXIifX19" style="width: 788px;" class="fr-fic fr-dib"></p><h2 data-element-id="headingsMap-13-0">The future isn't guaranteed</h2><p>One of the guiding mantras at Groq is that the ‘future isn’t guaranteed’. This is essentially a guiding principle that the future is both unknown yet directable. The intention within the development of technologies such as AI and computing must be deliberate and understood within a wider socio-economic context. Ross takes the responsibility that he and his company, and others like it have the power to change the future, very seriously. It's a way of thinking that is embedded in the company's methods and products.</p><p><strong>“</strong>At Google, the team and I created the TPU for existing workloads that we could make better as well as new and emerging workloads,” says Ross. &nbsp;“To plan for the future means understanding those workloads. Those were yesterday's challenges and we have new challenges today like power efficiency. To Groq those obstacles are the path and that's where we're going.”</p><p>“Today, we're shipping to our customers both as individual PCIe cards and systems with eight cards each and there's even more on the roadmap to come. I know 2020 has been a crazy year. Together we're pushing through and making great strides. Groq is going to grow our <a href="https://groq.com/careers/" rel="noopener noreferrer" target="_blank">headcount by 4X by the end of next year.</a> We’re not going to lose who we are because it's scrappy maneuverability that lets us innovate faster and more affordably than everyone else. That’s how we built the first PetaOp chip and that's how we're going to build what's next.”</p><h2 data-element-id="headingsMap-14-0">Conclusion</h2><p>The power of computing is at a tipping point where major players are making decisions about the direction they will pursue that will have huge impacts on the future of technological innovation and potential. Groq is one of these companies that understands it has the potential to completely change the direction of computing and massively influence the potential of AI. With that potential comes responsibility - something Ross and his team have demonstrated they are ready to shoulder. Interested in learning more about Groq? Head to their website <a href="http://www.groq.com" target="_blank">www.groq.com</a> to learn more about the future of computing and review their <a href="https://www.linkedin.com/company/groq/jobs/" target="_blank">multiple career opportunities</a>.</p><hr><h2 data-element-id="headingsMap-15-0"><em>About the sponsor: &nbsp;Groq</em></h2><p><em>Building the computer for the next generation of high performance machine learning. Groq hardware is designed to be both high performance and highly responsive. Groq’s new simplified architecture drives incredible performance at batch size 1. Whether you have one image or a million, Groq hardware responds faster. &nbsp;</em></p><p><em><img src="https://images.wevolver.com/eyJidWNrZXQiOiJ3ZXZvbHZlci1wcm9qZWN0LWltYWdlcyIsImtleSI6ImZyb2FsYS8xNjAzODc5MDE2MTE0LWdyb3EgIGxvZ28gYmFubmVyLmpwZyIsImVkaXRzIjp7InJlc2l6ZSI6eyJ3aWR0aCI6OTUwLCJmaXQiOiJjb3ZlciJ9fX0=" class="fr-fic fr-dib"></em><br></p><p><br></p><p><br></p><div language="JavaScript"><br></div>

Jonathan Ross, Founder, and CEO of Groq delivered a keynote speech as part of the AI Summit on September 29, 2020, describing insights on the industry, the market, and the unique Groq architecture capable of producing the world’s first PetaOp chip

The shortest path to performance requires a new approach

AI Infrastructure Engineering: Building Enterprise-Scale Machine Learning Systems

This article covers advanced architectural patterns, performance optimisation strategies, and critical operational considerations for building production-grade AI infrastructure systems!

Unexpected shutdowns? Costly errors? SCADA systems are the solution industries rely on to stay in control. Let's understand how this system allows owners to run industries through PLCs, RTUs, and other field devices, allowing them to make smarter decisions in real-time

What is SCADA?: The Basics you Need to Know

<p dir="ltr" id="isPasted">Specialized hardware accelerators like Neural Processing Units (NPUs) and Tensor Processing Units (TPUs) have revolutionized artificial intelligence (AI) hardware by providing capabilities for accelerating machine learning tasks. Both are designed to optimize performance for tasks like machine learning, deep learning, and neural network computations, but they cater to distinct use cases and requirements.&nbsp;</p><p dir="ltr">This article discusses the difference in architectures, performance, practical applications, and challenges of these powerful processing units, offering actionable insights for professionals.</p><p dir="ltr"><em><strong>Recommended reading:&nbsp;</strong></em><a href="https://www.wevolver.com/article/tpu-vs-gpu-in-ai-a-comprehensive-guide-to-their-roles-and-impact-on-artificial-intelligence" target="_blank"><strong><em>TPU vs GPU in AI: A Comprehensive Guide to Their Roles and Impact on Artificial Intelligence</em></strong></a></p><h2 dir="ltr">Introduction</h2><p dir="ltr">The rapid advancement of AI has led to an increasing demand for specialized hardware capable of efficiently handling complex computations. NPUs and TPUs have emerged as pivotal technologies in this space, each tailored to optimize tasks involving complex machine learning algorithms in unique ways. NPUs, with their low-power design, enhance edge computing performance, making them indispensable for IoT and mobile devices. TPUs, on the other hand, excel in scaling deep learning workloads across massive datasets, driving breakthroughs in cloud-based AI.</p><p dir="ltr">This article aims to provide professionals with a detailed comparison of these technologies, offering practical insights into their architectures, applications, and limitations. Understanding these distinctions can guide informed decision-making and optimize AI deployments for diverse requirements.</p><h2 dir="ltr">What Are Neural Processing Units (NPUs)?</h2><p dir="ltr">NPUs are specialized processors designed to optimize computations for neural networks, enabling efficient processing in real-time. With their ability to efficiently handle workloads like image recognition, voice processing, and natural language tasks, NPUs are an integral part of modern central processing units (CPUs) and System-on-Chip (SoC) designs. Companies like Huawei, Samsung, and Apple are leading the way in integrating NPUs into their devices to enhance AI capabilities without sacrificing power efficiency.</p><p dir="ltr"><span class="fr-img-caption fr-fic fr-dib" style="width: 832px;"><span class="fr-img-wrap"><img src="https://images.wevolver.com/eyJidWNrZXQiOiJ3ZXZvbHZlci1wcm9qZWN0LWltYWdlcyIsImtleSI6ImZyb2FsYS8xNzM3NTY2MzI0NTQ0LWltYWdlMi5wbmciLCJlZGl0cyI6eyJyZXNpemUiOnsid2lkdGgiOjk1MCwiZml0IjoiY292ZXIifX19" style="width: 100%px;" class="fr-fic fr-dib" alt="apple-neural-engine-devices"><span class="fr-inner">Apple's NPU, branded as the Neural Engine, is a key component of Apple's A-series and M-series chips. It is designed specifically for accelerating on-device machine learning and AI tasks</span></span></span></p><p dir="ltr" id="isPasted"><em><strong>Recommended reading:&nbsp;</strong></em><a href="https://www.wevolver.com/article/npu-vs-gpu-understanding-the-key-differences-and-use-cases" target="_blank"><strong><em>NPU vs GPU: Understanding the Key Differences and Use Cases</em></strong></a></p><h2 dir="ltr">What Are Tensor Processing Units (TPUs)?</h2><p dir="ltr">TPUs are hardware AI accelerator application-specific integrated circuits (ASICs) developed by Google to accelerate tensor-based computations, which are foundational to deep learning. Their design is optimized for mathematical operations like matrix-multiplication and parallel computing, making them ideal for tasks such as training large-scale neural networks and performing batch processing.</p><p dir="ltr">Here is a block diagram showcasing the architecture of TPUs:</p><p dir="ltr"><span class="fr-img-caption fr-fic fr-dib" style="width: 832px;"><span class="fr-img-wrap"><img src="https://images.wevolver.com/eyJidWNrZXQiOiJ3ZXZvbHZlci1wcm9qZWN0LWltYWdlcyIsImtleSI6ImZyb2FsYS8xNzM3NTY2Mzc3NTczLWltYWdlNS5qcGciLCJlZGl0cyI6eyJyZXNpemUiOnsid2lkdGgiOjk1MCwiZml0IjoiY292ZXIifX19" style="width: 100%px;" class="fr-fic fr-dib" alt="TPU-architecture-simplified-block-diagram"><span class="fr-inner">A simplified block diagram of a TPU</span></span></span></p><p dir="ltr" id="isPasted">The TPU architecture is designed to optimize AI workloads by integrating specialized components for efficient computation, storage, and communication. At the core is the Matrix Multiply Unit, which performs rapid large-scale matrix operations critical for AI tasks. Data flows through the Unified Buffer, a local storage unit for managing activations, ensuring quick access and reducing dependency on external memory. The Control Unit orchestrates data movement and synchronizes operations between components. The PCIe Interface and Host Interface facilitate communication between the TPU and external systems, enabling smooth data transfer and coordination. The DRAM Ports provide access to external memory for managing large datasets that exceed the buffer's capacity. The Activation Pipeline processes intermediate results, applying transformations essential for refining AI computations. Additionally, the Accumulator stores partial computation results temporarily for aggregation and further processing. Finally, the Miscellaneous I/O manages input/output tasks, ensuring seamless integration with the broader system. This balanced architecture ensures high efficiency and performance for AI tasks in data centers.</p><p dir="ltr"><span class="fr-img-caption fr-fic fr-dib" style="width: 832px;"><span class="fr-img-wrap"><img src="https://images.wevolver.com/eyJidWNrZXQiOiJ3ZXZvbHZlci1wcm9qZWN0LWltYWdlcyIsImtleSI6ImZyb2FsYS8xNzM3NTY2MzkwMzg5LWltYWdlNC5wbmciLCJlZGl0cyI6eyJyZXNpemUiOnsid2lkdGgiOjk1MCwiZml0IjoiY292ZXIifX19" style="width: 100%px;" class="fr-fic fr-dib" alt="Google-TPU-v4"><span class="fr-inner">Google TPUs are primarily used in servers within data centers to accelerate large-scale AI workloads, such as training and deploying deep learning models.</span></span></span></p><p dir="ltr" id="isPasted"><em><strong>Recommended reading:&nbsp;</strong></em><a href="https://www.wevolver.com/article/tensor-cores-vs-cuda-cores" target="_blank"><strong><em>Tensor Cores vs CUDA Cores: The Powerhouses of GPU Computing from Nvidia</em></strong></a></p><h2 dir="ltr">Comparing the Performance of NPUs vs TPUs</h2><p dir="ltr">NPUs and TPUs exhibit distinct performance characteristics, making them suitable for different AI workloads. Following section compares both the AI chips based on different parameters.</p><h3 dir="ltr">Speed and Latency</h3><p dir="ltr">NPUs&nbsp;are designed for&nbsp;real-time, low-latency&nbsp;tasks, making them ideal for&nbsp;edge computing&nbsp;scenarios where immediate responses are required. For instance,&nbsp;NPUs&nbsp;enable instant image processing in&nbsp;smartphones&nbsp;and&nbsp;IoT devices</p><p dir="ltr">TPUs often excel in speed and scalability for large datasets, leveraging their specialized architecture for tensor computations. They are optimized for throughput rather than latency, making them suitable for batch data processing tasks like training massive deep learning models. In contrast, NPUs are designed for real-time, low-latency tasks on the edge.&nbsp;</p><h3 dir="ltr">Power and Energy Efficiency</h3><p dir="ltr">Power consumption is a critical factor for both NPUs and TPUs, but their optimization differs. NPUs provide better efficiency for edge computing and low-power applications, optimizing computing tasks with limited computational resources. &nbsp;This ensures extended battery life for small devices while performing complex AI tasks. TPUs, while efficient for their scale, are designed for environments like data centers where power availability is less constrained but efficiency remains essential for handling extensive computations.</p><h3 dir="ltr">Scalability and Flexibility</h3><p dir="ltr">NPUs&nbsp;are less scalable compared to&nbsp;TPUs, as they are tailored for specific edge&nbsp;AI&nbsp;applications. However, their&nbsp;flexibility&nbsp;in&nbsp;real-time&nbsp;processing tasks makes them invaluable for&nbsp;localized AI workloads.&nbsp;TPUs, on the other hand, are built for&nbsp;scalability, capable of handling&nbsp;large-scale&nbsp;AI&nbsp;models and&nbsp;datasets&nbsp;in&nbsp;cloud-based&nbsp;or&nbsp;distributed environments.</p><h2 dir="ltr">Architectural Differences</h2><h3 dir="ltr">Core Design</h3><p dir="ltr">NPUs&nbsp;are engineered to handle&nbsp;neural network activations&nbsp;and&nbsp;real-time&nbsp;matrix computations with specialized cores that focus on efficient, localized processing. These cores typically feature lightweight architectures optimized for low power consumption and latency-sensitive tasks.&nbsp;TPUs, in contrast, utilize systolic array architectures with a high density of computational cores designed for parallel matrix multiplications.</p><h3 dir="ltr">Programming Frameworks and Ecosystem</h3><p>NPUs are compatible with general-purpose AI frameworks such as TensorFlow, PyTorch, and ONNX, allowing developers to integrate AI tasks into various devices. As they are integrated within larger processors, they are more general-purpose and are compatible with a number of different computing tasks. TPUs, in contrast, are tightly integrated with Google’s TensorFlow ecosystem, making them an ideal choice for developers already working within Google’s AI infrastructure. The trade-off lies in flexibility versus specialization.</p><p><span class="fr-img-caption fr-fic fr-dib" style="width: 832px;"><span class="fr-img-wrap"><img src="https://images.wevolver.com/eyJidWNrZXQiOiJ3ZXZvbHZlci1wcm9qZWN0LWltYWdlcyIsImtleSI6ImZyb2FsYS8xNzM3NTY2NDEzOTA4LWltYWdlMS5wbmciLCJlZGl0cyI6eyJyZXNpemUiOnsid2lkdGgiOjk1MCwiZml0IjoiY292ZXIifX19" style="width: 100%px;" class="fr-fic fr-dib" alt="intel-core-ultra-gen-1-series-with-NPUs"><span class="fr-inner">Intel Core Ultra processors integrate NPUs to deliver efficient and enhanced performance for modern AI desktop applications.</span></span></span></p><h2 dir="ltr">Use Cases and&nbsp;Practical Applications</h2><p dir="ltr">NPUs and TPUs cater to distinct scenarios based on their specialized capabilities.</p><p dir="ltr">NPUs find their strengths in edge computing, enabling AI tasks directly on devices like smartphones, cameras, and IoT sensors. Examples of&nbsp;NPU Applications&nbsp;include:</p><ul><li dir="ltr"><p dir="ltr">Real-time object detection in edge devices, such as security cameras.</p></li><li dir="ltr"><p dir="ltr">Voice recognition in smart assistants operating on low-power devices.</p></li><li dir="ltr"><p dir="ltr">AI-powered healthcare devices for monitoring patient vitals and diagnosing conditions.</p></li><li dir="ltr"><p dir="ltr">Industrial <a href="https://www.wevolver.com/article/building-automation-systems">automation systems</a> requiring real-time decision-making capabilities.</p></li></ul><p dir="ltr">TPUs are the backbone of AI operations on large amounts of data. Examples of&nbsp;TPU Applications&nbsp;include:</p><ul><li dir="ltr"><p dir="ltr">Training large-scale machine learning models in cloud environments.</p></li><li dir="ltr"><p dir="ltr">Batch processing for natural language understanding tasks in virtual assistants.</p></li><li dir="ltr"><p dir="ltr">Optimizing recommendation systems in large-scale e-commerce platforms.</p></li><li dir="ltr"><p dir="ltr">Enabling real-time analytics in large datasets for financial and scientific domains.</p></li></ul><h2 dir="ltr">Challenges and Limitations of NPUs and TPUs</h2><h3 dir="ltr">Engineering Challenges in Implementation</h3><p dir="ltr">Deploying NPUs and TPUs presents several technical challenges, including:</p><ul><li dir="ltr"><p dir="ltr"><strong>Compatibility Issues:</strong> Existing software ecosystems, such as libraries and frameworks, often require extensive adaptation to fully leverage the capabilities of NPUs and TPUs.</p></li><li dir="ltr"><p dir="ltr"><strong>Hardware Integration:&nbsp;</strong>Ensuring seamless integration with existing system architectures can be complex, particularly in legacy systems.</p></li><li dir="ltr"><p dir="ltr"><strong>Optimization Challenges:&nbsp;</strong>AI models often need to be specifically optimized to achieve peak performance on these specialized processors.</p></li><li dir="ltr"><p dir="ltr"><strong>Cost-Effectiveness:&nbsp;</strong>While powerful, the initial investment in NPU or TPU hardware can be prohibitive for smaller organizations.</p></li></ul><h3 dir="ltr">Limitations in Real-World Applications</h3><p dir="ltr">Despite their capabilities, NPUs and TPUs face limitations in specific environments:</p><ul><li dir="ltr"><p dir="ltr"><strong>Latency-Sensitive Systems:</strong> TPUs, designed for high-throughput batch processing, may not be ideal for real-time applications that demand low latency.</p></li><li dir="ltr"><p dir="ltr"><strong>Hybrid Requirements:</strong> Certain applications, such as autonomous vehicles, may require a combination of TPU and graphics processing unit (GPU) capabilities to balance precision and speed effectively.</p></li><li dir="ltr"><p dir="ltr"><strong>Scalability Constraints:&nbsp;</strong>While TPUs scale well in cloud environments, NPUs are less adaptable to large-scale deployments due to their edge-specific design.</p></li></ul><p dir="ltr">Here is a table summarising the discussion on NPUs and TPUs:</p><div dir="ltr" align="left"><table><tbody><tr><td><p dir="ltr"><strong>Feature</strong></p></td><td><p dir="ltr"><strong>NPUs</strong></p></td><td><p dir="ltr"><strong>TPUs</strong></p></td></tr><tr><td><p dir="ltr"><strong>Primary Use Case</strong></p></td><td><p dir="ltr">Real-time AI tasks on edge devices (e.g., smartphones, laptops, IoT, robotics)</p></td><td><p dir="ltr">Large-scale AI workloads in data centers (e.g., training deep learning models)</p></td></tr><tr><td><p dir="ltr"><strong>Key Applications</strong></p></td><td><p dir="ltr">Image recognition, natural language processing, voice recognition, augmented reality</p></td><td><p dir="ltr">Language models, natural language processing, large-scale image classification</p></td></tr><tr><td><p dir="ltr"><strong>Architectural Focus</strong></p></td><td><p dir="ltr">Low-latency, power-efficient processing for real-time AI tasks</p></td><td><p dir="ltr">High-performance parallel processing and matrix multiplication for deep learning</p></td></tr><tr><td><p dir="ltr"><strong>Core Design</strong></p></td><td><p dir="ltr">Lightweight architectures optimized for low power and real-time processing</p></td><td><p dir="ltr">Systolic array architectures optimized for massive parallel tensor operations</p></td></tr><tr><td><p dir="ltr"><strong>Scalability</strong></p></td><td><p dir="ltr">Limited scalability, designed for specific edge AI tasks</p></td><td><p dir="ltr">High scalability for training large datasets and deploying AI models</p></td></tr><tr><td><p dir="ltr"><strong>Power Consumption</strong></p></td><td><p dir="ltr">Optimized for energy efficiency in portable devices</p></td><td><p dir="ltr">Designed for efficiency in data centers with less strict power constraints</p></td></tr><tr><td><p dir="ltr"><strong>Programming Ecosystem</strong></p></td><td><p dir="ltr">Supports general-purpose AI frameworks (e.g., TensorFlow, PyTorch, ONNX)</p></td><td><p dir="ltr">Tightly integrated with TensorFlow and Google Cloud services</p></td></tr><tr><td><p dir="ltr"><strong>Manufacturers</strong></p></td><td><p dir="ltr">Companies like Apple (Neural Engine), Huawei, Samsung, and others</p></td><td><p dir="ltr">Google</p></td></tr><tr><td><p dir="ltr"><strong>Integration</strong></p></td><td><p dir="ltr">Embedded in SoC designs for smartphones and IoT devices</p></td><td><p dir="ltr">Used as standalone accelerators in server environments</p></td></tr><tr><td><p dir="ltr"><strong>Key Strengths</strong></p></td><td><p dir="ltr">Real-time processing, low latency, energy efficiency</p></td><td><p dir="ltr">Exceptional performance for large-scale AI training and inference</p></td></tr><tr><td><p dir="ltr"><strong>Key Limitations</strong></p></td><td><p dir="ltr">Less suited for large-scale AI workloads</p></td><td><p dir="ltr">Higher power requirements, less suited for edge applications</p></td></tr></tbody></table></div><h2 dir="ltr">Conclusion</h2><p dir="ltr">NPUs and TPUs play complementary roles in the advancement of AI technology, each excelling in specific environments. While TPUs offer unparalleled performance for large-scale machine learning tasks in cloud environments, NPUs are perfect for real-time inference and energy-efficient edge applications.&nbsp;</p><h2 dir="ltr">Frequently Asked Questions</h2><ol><li dir="ltr"><strong>What Is the Primary Difference Between NPUs and TPUs?</strong><br>NPUs are specialized for low-power, real-time tasks, making them ideal for edge devices.<br>TPUs are optimized for large-scale tensor computations, excelling in cloud-based machine learning.</li><li dir="ltr"><strong>What Are the Key Roles of NPUs and TPUs in AI?</strong><br>NPUs focus on real-time, low-power AI processing in edge devices like smartphones and IoT systems, enabling efficient inference for tasks such as image recognition and voice commands. TPUs specialize in high-performance, large-scale tensor computations, excelling in data centers for AI training and deployment.</li><li dir="ltr"><strong>How Do NPUs and TPUs Compare to GPUs?</strong><br>NPUs are specialized for efficient AI/ML tasks, particularly inference, and are optimized for low-power edge devices like smartphones. TPUs excel in tensor-based operations for deep learning, offering high-speed training and inference, primarily in Google's ecosystem. GPUs, versatile in design, handle a wide range of workloads, including graphics rendering, ML training, and scientific computing, making them ideal for parallel processing. While GPUs are general-purpose, NPUs and TPUs focus on AI-centric optimizations.</li></ol><p dir="ltr"><em><strong>Recommended reading:&nbsp;</strong></em><a href="https://www.wevolver.com/article/understanding-nvidia-cuda-cores-a-comprehensive-guide" target="_blank"><strong><em>Understanding Nvidia CUDA Cores: A Comprehensive Guide</em></strong></a></p><h2 dir="ltr">References</h2><p dir="ltr">[1] Google. Cloud TPU documentation. Available from: <a href="https://cloud.google.com/tpu/docs" target="_blank">https://cloud.google.com/tpu/docs</a></p><p dir="ltr">[2] Apple. Deploying Transformers on the Apple Neural Engine. Available from: <a href="https://machinelearning.apple.com/research/neural-engine-transformers" target="_blank">https://machinelearning.apple.com/research/neural-engine-transformers</a></p><p dir="ltr">[3] Van Belle JL. Why neural processing units (NPUs) are the next Big Thing in AI. 2024. doi:10.13140/RG.2.2.35619.87845. Available from: <a href="https://www.researchgate.net/publication/383184868_Why_neural_processing_units_NPUs_are_the_next_Big_Thing_in_AI" target="_blank">https://www.researchgate.net/publication/383184868_Why_neural_processing_units_NPUs_are_the_next_Big_Thing_in_AI</a></p><p dir="ltr">[4] Sanmartín D, Prohaska V. Exploring TPUs for AI Applications. IE Robotics &amp; AI Club, IE University, Madrid, Spain. arXiv preprint arXiv:2309.08918. Available from:<a href="https://arxiv.org/pdf/2309.08918" target="_blank">&nbsp;https://arxiv.org/pdf/2309.08918</a>.</p>

NPUs are integrated units that excel in real-time AI tasks on edge devices like smartphones and IoT systems with low power consumption. TPUs are standalone processors designed for large-scale AI workloads in data centers, delivering exceptional performance in deep learning tasks.

NPU vs TPU: Understanding the Key Differences in AI Hardware Accelerators

GPUs excel in parallel processing for graphics and AI training with scalability, while NPUs focus on low-latency AI inference on edge devices, enhancing privacy by processing data locally. Together, they complement each other in addressing different stages of AI workloads efficiently.

NPU vs GPU: Understanding the Key Differences and Use Cases

Fine-tuning large language models adapts pre-trained models to specific tasks or domains using tailored datasets, while Retrieval-Augmented Generation (RAG) combines retrieval systems with generative models to dynamically incorporate external, up-to-date knowledge into outputs.

RAG vs Fine-Tuning: Differences, Benefits, and Use Cases Explained

Large Language Model (LLM) training involves teaching AI models to understand and generate human-like text by processing vast amounts of data, significantly enhancing their language comprehension and production capabilities.

Robotics Software

Technical Guides

AI Infrastructure Engineering: Building Enterprise-Scale Machine Learning Systems...

What is SCADA?: The Basics you Need to Know

NPU vs TPU: Understanding the Key Differences in AI Hardware Accelerators

NPU vs GPU: Understanding the Key Differences and Use Cases

RAG vs Fine-Tuning: Differences, Benefits, and Use Cases Explained

LLM Training: Mastering the Art of Language Model Development