Building the Internet of Things

Executive summary

To take full advantage of IoT opportunities in predictive maintenance, you need to think strategically about the many elements of IoT. For example, one should consider connectivity pathways and types, transport-layer and application-layer protocol choices, device interaction and communication patterns, and how to design for the vast scale of IoT. It is especially critical to understand the complex issues of data security and regulatory compliance, which can expose the enterprise to legal difficulties if they are not handled properly. You also should think about how the enterprise’s communications systems will ingest data, including message types, sizes, formats, and priorities, conditional and contextual messaging, message batching, bandwidth, and how to scale a messaging system.

Another pivotal set of questions to ask relate to the data: where will data be stored and how will it be distributed or potentially sold, and what is the longevity of the data, the right format, and the associated cost to do that? What is the most efficient way to analyze Big Data, how can you best take advantage of possibilities, such as alarm processing, complex-event processing, Big Data analysis, machine learning, and data enhancement? Because data that seems at first uninteresting can be very valuable to the right audience, how do you find that audience to monetize the insights gained from processing it?

The elements that are needed for security, communication, and scale in an IoT solution make it very challenging to build one from scratch. To succeed with any IoT solution, it will very likely require the implementation of a reference architecture that can help accelerate the use of massive data from millions or even billions of devices. Modeling the system’s capacity to scale, and calculating the costs to do so for related aspects, such as ingress (device to cloud) and egress (cloud to device, cloud to system) paths and system processing, is paramount. Depending on the company background, a classic buy vs. build vs. hybrid decision should be made, based on what you are already using, what is available, and what will be available in the near future at a price that is acceptable to your business. This white paper introduces and describes all of these considerations and provides you with the tools necessary to estimate the operational cost of an implemented reference architecture in production.

IoT and predictive maintenance

At Microsoft, we hear constantly from customers who say that the Internet of Things (IoT) is one of the most exciting trends in IT.¹ Many of our customers are interested in deploying sensors and devices in every part of their businesses in order to capture information from the physical world and act upon the knowledge gained from refining it. They will buy or build systems that can deliver these capabilities in order to optimize their bottom line, keep customers satisfied, and explore new revenue potential.

Predictive maintenance is an IoT scenario where a device can provide data that leads to insightful, proactive maintenance before the likely failure can take place. Predictive maintenance offers a new revenue stream for device manufacturers, and it is very interesting to their customers because it enables better business continuity, which usually generates extra revenue. In this way, the cost of a new service from a device manufacturer is justifiable to customers, given the cost and impact on them of unplanned downtime of the device.

a.The Internet of Things

An expert in radio-frequency identification named Kevin Ashton first used the term “the Internet of Things” in 1999,² though the idea had been around at least a decade earlier. As with many terms in technology, IoT is a loaded term that people interpret differently depending on their viewpoint and purpose. For example, Gartner defines it as “The network of physical objects that contain embedded technology to communicate and interact with their internal states or the external environment.”³ Formulating this differently:

The Internet of Things is a metaphor for a set of systems in which direct human intermediation is dramatically reduced by equipping distributed systems with sensors that let us acquire information, make decisions, and control things in the physical world.

Figure . Foundational activities, composable within and between devices and systems

Based on this definition, IoT consists of a set of four composable activities:

Acquiring data. Using sensors to record information about the physical world. Examples include measuring location, humidity, temperature, light, heart rate, blood pressure, brain waves, current, and gas detection.

Processing information. Take action based on data captured and on contextual information retrieved previously or sourced from other systems. This processing could involve using actuators that can alter the state of the physical world, such as opening valves, switching machines on and off, sounding alarms, controlling servos, closing doors, and many other things.

Storing information. To enable trend analysis, forecasting and insight-driven decision making, historical information and context is needed. Storing the information retrieved in its contextual form (for example, including the location where it was captured, the date and time it was captured, the state of the system at the time it was captured, and so on.) is critical for this process.

Publishing insights. When embedded sensor data is combined with both internal and external data from other systems, additional insight from analyzing the data can be learned and acted upon. Exposing that insight can also drive additional value for other stakeholders outside the immediate needs of the current system, allowing for the monetization of this knowledge.

On top of familiar devices, such as phones for input and presentation, a set of core components to support those activities is needed, though business goals and technical constraints will drive those that are required. Core components may include:

Sensors: the components that translate a value from the physical world into bits. Examples include sensors that measure pressure, humidity, heart rate, gas levels, and acceleration.

Devices: networked, physical, special-purpose systems that emit telemetry data, accept external information, request external information, and execute remotely-issued commands. Examples include factory floor equipment, environmental pollution sensors, and control modules in vehicles.

Bridges: systems that act as communication brokers between a device and a gateway, typically by translating data traffic between different link protocols or methods, for instance between short-range and long-range wireless protocols. A bridge can also be a connectivity infrastructure that manages a nationwide or world-wide wireless network on one side, and a bridge to a cloud system on the other. A bridge might also perform intelligent preprocessing of data, or act as an autonomous local communications hub in addition to its bridging function relative to a cloud system⁴. Bridges are often also referred to as gateways, but we reserve the term gateway for a network-based service with which a bridge communicates.

Gateways: network-based services that manage connectivity and connections with devices either directly or through bridges. The service establishes a trusted communication relationship with a device, deals with ingestion and routing of telemetry data, and provides access to command and notification data destined for the device. On top of these services, it provides data pipeline processing, possible containing transformation, complex event processing capabilities, data analytics components, machine learning, and so on.

Machine learning: computational algorithms that can analyze large sums of data and extract patterns from it to help a system act and “learn” from that data to drive more intelligent system responses in the physical world.

Interconnections: different systems sharing learnings and data that in turn form composite systems.

We have read thought-provoking papers about IoT. Two that we found especially valuable in providing context to the concepts and opportunities of IoT are:

“Recommendations for the Strategic Initiative INDUSTRIE 4.0.”⁵

“Industrial Internet: Pushing the Boundaries of Minds and Machines, a European Perspective.”⁶

IoT enables you to build, enhance or extend a business model based on data-driven insights from pervasive sensors that help you optimize resource use and reduce cost and environmental impact. IoT also helps you maintain a closer relationship with customers beyond the point of sale of physical products by enabling contextual, remote actions automatically and intelligently. Examples include remote servicing, proactive sales, best-practices guidance, and more.

Business value

At least 26 billion devices will be connected on the Internet by 2020, and organizations in every sector will use them.⁷ Billions of connected devices will help businesses to:

Reduce cost. Businesses can use the increased insight into manufacturing and delivery processes to optimize those processes and reduce cost. For example, reducing the number of scheduled visits a technician must make by scheduling service visits based on duty cycles and expected product lifespans informed by actual usage.

Create new revenue streams. Using the ability to sense from and actuate in the physical, new business models are emerging. Business can capitalize on these new opportunities and create new innovate revenue streams. Some examples would be monetizing newly collected datasets, offering APIs to create new business partnerships, increasing service revenue by notifying and offering improved convenience to customers, offering differentiating SKUs based on usage patterns, supplying optimized configuration services, and so on.

Increase customer satisfaction and retention. By knowing how customers of physical products use them, opportunities exist to extend the customer experience into scenarios of higher value, and retain and extend the customer base. Capturing data on how customers actually use products, and ensuring that they do not experience frustrating service issues helps companies retain customers.

In the blog post “10 reasons businesses need a strategy for the Internet of Things now,”⁸ the author identified a concise set of benefits that a company can realize by adopting an IoT strategy.

Megatrends

The world faces many challenges, such as changes in wealth distribution, resource scarcity, and an aging population in developed countries. The authors of the book “From Machine-to-Machine to the Internet of Things: Introduction to a New Age of Intelligence” analyzed these megatrends and capabilities in detail.⁹ They found that these megatrends are driving a proliferation of embedded devices with sensors, which in turn require new capabilities for new market scenarios, as the graphic below shows.

Figure . "Megatrends." From Machine-To-Machine to the Internet of Things: Introduction to a New Age of Intelligence. Amsterdam, Netherlands, Elsevier, January 2014.

Among the list of megatrends listed in the previous figure, we want to explain in this paper how some of them relate to the Internet of Things:

Natural resource constraints. The world population is growing at a high rate, with a projected peak population of 9.22 billion in 2075.¹⁰ Given this growth and the impact it has on the growth of the worldwide economy, the world will increasingly have to do more with less, and optimize the way that we produce. IoT can support the optimization of production, loss reduction, and the efficiency of the necessary supply chain.

Economic shifts. Much like the shift in IT, going from packaged products to as-a-service solutions, the global economy is moving from a product-oriented to a service-oriented perspective.¹¹ For a viable service-oriented economy to come into existence, it needs to be supported by a large set of devices that provide context to the customer environment for the system in order to offer the right service, at the right price, and at the right time.

Changing demographics. With the world population, especially in more-developed countries, increasingly aging, the change in demographics will need smart solutions that can help elderly people remain self-supporting.

Climate change. The impact of human activities on the environment, although debated at length, is detrimental to the sustainability of the world. In recent years, there has been a growing movement of “green” technologies and services, ranging from electric cars to corporate and government policy changes. IoT can be a supporting factor in both providing footprint insight and reduction.

Technology enablers

The ever-decreasing cost and size of components, such as accelerometers, Wi-Fi radios,¹² GPS, microcontrollers, and Bluetooth radios is also enabling the Internet of Things (IoT). It allows components and devices to be used in new settings, such as wearables, on-person devices, and even smaller equipment.

As shown in Figure , IoT depends on several other major technologies and trends. Some of these technology enablers as well as others warrant clarification:

Ubiquitous connectivity. Low-powered wireless networking enables devices to talk to a gateway, among each other, or directly to the outside world. A foundation for IoT implementations, connectivity must be managed carefully. To learn more, see the Connectivity section in this paper.

Cloud computing. For systems that connect hundreds of millions of devices, cloud computing is the technology that allows for vast scale and acceptable costs, providing the ability to store large amounts of machine generated data at low cost and perform Big Data analytics and machine learning.

Small, low-power, low-cost microcontrollers. Microcontrollers today can perform tasks at very low power and have a battery life of many years.¹³ For example, the Texas Instruments MSP430 runs at less than 100µA/MHz and can operate on a single coin battery for more than 20 years.¹⁴ (Device battery life always depends on components and application cycle use). The memory embedded in this microcontroller is ferroelectric read-only memory (FRAM), an improvement on flash memory that sports very high data throughput at a power consumption three times lower than flash memory and 99 percent lower than comparable dynamic random-access memory (DRAM).

Power supply and storage technologies. Given the tiny size of many new devices, their deployment location, and the vast number of them that will be deployed, changing batteries is often impractical or impossible. Besides optimizing hardware design for these scenarios,¹⁵ enhancing circuitry by limiting their quiescent current (I_q) will further improve battery life. Also, with energy harvesting techniques, such as solar power supplies, devices can recharge their built in batteries as long as there is a minimal charge left.

Embedded operating system platforms. With the vast number of devices that will be installed, cost and energy consumption per device become decisive. Engineers will create devices that cost less and that are more energy-efficient, even if they have limited processing capabilities and memory. CPU cycles spent, and the memory allocated will become important factors for choosing operating system platforms, installed components, and security configurations. There is a plethora of good general-purpose operating systems, ranging from Windows Embedded and Embedded Linux to real-time operating systems, such as FreeRTOS, ThreadX, Integrity, Nucleus, Qnx, Atomthreads, AVIX-RT, ChibiOS/RT, ERIKA Enterprise, TinyOS, Thingsquare Mist/Contiki, and others.¹⁶

In sum, IoT is gaining momentum because of growing customer and enterprise needs meeting technology enablers at the right cost.

Standardization efforts

Throughout the world, many organizations are working on the standardization of IoT, based on specific technology or holistically on reference architectures. Examples of this work include:

ITU-Telecom (ITU-T), Internet of Things Global Standards Initiative (IoT-GSI).¹⁷

European Union, Internet of Things Architecture (IoT-A).¹⁸

In addition to these efforts, there is a lot of work going on in depth in many different technology areas, such as the standardization of protocols. Protocol choices, both at the transport as well as the application layer, are discussed later in this document.

Directory: download
download -> Performance Tuning Guidelines for Windows Server 2008 May 20, 2009 Abstract
download -> Northern England’s set-jetting locations
download -> Bbc archives: Beyond the programme download
download -> Occupational health and safety
download -> Dynatest 3031 lwd light Weight Deflectometer
download -> Forth coming competitive exams
download -> English Olympiad Tasks: 10 th form Аудіювання, 10 клас. Текст
download -> Brush High School Morning Announcements Friday, September 05, 2014 all school
download -> Year 9 The Arts — Music
download -> Years 7 and 8 standard elaborations — Australian Curriculum: Music draft

Download 258.06 Kb.

Share with your friends: