The Internet of Things1 (IoT) is upon us. For years, market analysts have been obsessed with how many people use the Internet, but the fastest growing group of “users” are things. More devices than people are already connected. Estimates of Internet-connected devices range from 50 billion to 1 trillion by 2020. The technological and cost barriers to add sensing and networking capabilities to things are falling rapidly. At the 2013 Consumer Electronics Show, innovators displayed services based on connecting everyday things, such as HAPIfork (a fork that monitors eating habits),2 Beam Brush3 (a toothbrush that tracks dental hygiene habits), and others.
As discussed in the article, “Using technology to help customers achieve their goals,” the IoT is creating the potential for a post-transaction relationship with customers in which enterprises take advantage of information from sensors embedded in their products that are linked to smartphones and to the cloud. They use the information to track and then guide the use of products and services that better align to the customers’ goals. Thus far, technology generally introduced digital processes into the design, manufacturing, marketing, and selling of products and services; that is, up to the transaction. Now emerging technologies enable a post-transaction relationship between customers and vendors.
Until recently, embedding sensors and connectivity hasn’t been easy or economical, thereby limiting the possibility for post-transaction relationships to high-value segments, primarily in business-to-business (B2B) scenarios such as monitoring power plants or jet engines. Many emerging technologies are now pushing down the cost curve, in effect democratizing the opportunity to digitize any customer’s consumption process.
But to capitalize on post-transaction relationships, enterprises will need to become service providers, bringing together a distributed system using hardware and software technologies, with the goal of integrating sensing, networking, computing, and the end-user experience into a valuable service for customers. This integration will extend their business model beyond the device itself. “To become a service provider, [an enterprise] must learn to build and own the full stack—from network to device to application. That’s the name of the game at this point,” says Macario Namie, vice president of marketing at Jasper Wireless, a provider of machine-to-machine network management solutions.
This article provides an overview of the emerging technologies and where they fit into the stack. Prior issues of the Technology Forecast cover related advances in social technologies, mobile technologies, analytics, and cloud computing (SMAC), which also contribute to going beyond the transaction.
Historically, the adoption of information technologies accelerates after the layers of functionality are conceptualized and delivered as semi-autonomous, loosely coupled offerings. In essence, the tech industry coalesces around an IT stack of hardware and software capabilities and interfaces that create the full system. The IoT stack is rapidly approaching this level of maturity to become something PwC calls the Thing Stack. It has three layers, as illustrated in Figure 1:
The layers in the Thing Stack
New technologies, solutions, and choices are emerging rapidly— even accelerating—in all three layers. This trend is not surprising, given the broad range of use cases across different industries. At the same time, the service-specific requirements for local processing, networking method, power constraints, size, cost, and other concerns will vary widely based on use case. “A healthcare application that monitors a person’s temperature and blood pressure will be very different from a utility application that monitors a smart meter,” suggests Namie.
For any given enterprise, the context defined by their devices, the use cases, and the knowledge of what information will create value will guide designers toward particular choices of emerging technologies and solutions in the Thing Stack. What follows is a discussion of some of the developments in each of the layers of the Thing Stack.
Sensors have been used for years in industrial, automotive, healthcare, manufacturing, and other contexts. Now they are becoming small enough and inexpensive enough to embed ubiquitously in all devices and the physical environment.
According to IC Insights, worldwide sensor sales will increase by a compound annual growth rate (CAGR) of 18 percent, to $10.9 billion between 2011 and 2016. (See Figure 2.) During the same time, the unit shipments will increase from about 8 billion to 18 billion.4
Worldwide sensor market and unit shipments from 2008 to 2016
Driving this proliferation of sensors is their inclusion in smartphones, which already include the types of sensors shown in Table 1. Sensors in smartphones capture contextual information about location, motion, orientation, light, and other environmental elements, which app developers can use in new services.
Sensor types used in the various mobile operating systems
Besides embedding sensors in smartphones, innovators are bringing to market new devices that contain many more sensors that take advantage of smartphones to capture and integrate data from the cloud and the device. For instance, Sensordrone, a Kickstarter project, packs a dozen environmental sensors—including humidity, pressure, color intensity, and gas sensors—into a keychain-sized dongle that collects information about the surrounding environment and relays it to the smartphone via Bluetooth.5
Such innovation expands the range of services that can be provided by digitizing the use of the phone as well as the physical world around the device.
Another class of sensors is used in watches, wristbands, contact lenses, fabric, and other devices worn or implanted in the body or skin. Nokia recently patented technology for magnetic tattoos that vibrate on incoming calls or messages. This action would be possible by tattooing, stamping, or spraying ferromagnetic material onto a user’s skin and then pairing it with a mobile device.6 In another example, researchers at Microsoft and the University of Washington are expanding the use of the contact lens to provide real-time updates on biochemical fluctuations in the body. Such a device could be used to monitor insulin levels in diabetes patients.7
ABI Research forecasts the wearable computing device market will grow to 485 million annual device shipments by 2018. Much of what makes this growth possible are advances in materials science and augmented reality. New developments in materials science are essential to safely and seamlessly embed sensors in fabrics, implantable devices, and the human body. For example, Proteus Digital Health makes an ingestible sensor that is one square millimeter and embedded in a pill.8 The sensor transmits a signal when it comes in contact with stomach fluids. When used in conjunction with a patch on the arm, the sensor helps track patient compliance with prescription medicine regimes.
Advances in augmented reality will be essential to blend the physical and digital worlds into a seamless experience. For instance, Google’s Project Glass is experimenting with making augmented reality part of individuals’ daily lives by shrinking the size of a head-mounted display to the equivalent of regular eyeglasses. These glasses are capable of capturing and playing audio and video, have a built-in compass and accelerometer, and allow interaction with voice and head movement.9
The size and price of sensors are dropping, thanks to microelectromechanical systems (MEMS). MEMS combine electronics and mechanical components at tiny scale by integrating sensors, actuators, and integrated circuits. Bosch, STMicroelectronics, Panasonic, and Texas Instruments are the largest suppliers of MEMS. The fastest growing segment for MEMS is gyroscopes and accelerometers used in smartphones, where increasing volumes are driving down prices. The price of a three-axis accelerometer dropped by about 80 percent between 2007 and 2010. And the price of a one-axis automotive gyroscope dropped by about 88 percent between 2006 and 2010.10
There are many reasons to expect prices to continue to fall. MEMS use the production techniques of the semiconductor industry and benefit from manufacturing efficiencies at large volumes. As volumes rise, the MEMS industry is also getting better at reducing the cost of testing and packaging.
The state of the art in MEMS typically allows one discrete microsensor to be combined with discrete electronics in a silicon substrate. As techniques to pack multiple sensors and electronics on a single substrate are developed, further reductions in price and size are expected. “Today we are making a tricorder that will fit in a pocket. In the long term, it will be a tricorder in a chip that we can implant in the environment,” says Walter De Brouwer, CEO of Scanadu, a healthcare startup developing a tricorder and aiming to transform the smartphone into a health phone by packing it with various sensor capabilities. “The tricorder will have a complete diagnostic experience so consumers can explore their health.”
Over the longer term, nanoelectromechanical systems (NEMS) hold promise. NEMS integrate electrical and mechanical functionality at nanoscale, and they will provide ample opportunities to further reduce the size and price of sensors.
Sensing information and sharing that information over a network are often handled by separate components in a device. However, another trend is to integrate sensing, networking, and power on a single node, so they can be used liberally in many environments without being constrained by a power source or networking capability. For example, Streetline’s smart parking solution has a networked sensor to detect and transmit the presence or absence of cars in a metered parking spot. The sensors communicate with each other via a wireless mesh network protocol.11
The HP Labs’ Central Nervous System for the Earth (CeNSE) project combines advances in materials, nanotechnology, and MEMS to develop a planet-wide sensing network using billions or trillions of tiny, inexpensive, and tough sensors.12 Such sensor ubiquity will accelerate the integration of the physical and digital worlds, making possible various new use cases, including monitoring the conditions of bridges, forests, and the air people breathe.
The second layer in the Thing Stack provides processing, local storage, and connectivity. The IoT, as its name implies, requires a connection to the Internet to combine sensor information with cloud-resident data. But even before connecting to the cloud, most devices will need local processing capability for quantifying, summarizing, and analyzing. In some cases, the things will do more than sense; they will take action, such as by turning something on or off. Programmable microcontrollers typically serve these functions. Finally, local data storage for staging sensor information usually resides in this layer.
The first key capability of this layer is networking. Networks can be wired or wireless. For devices that are stationary and can access external sources of power, a wired network would be suitable but carries the burden of running cables from a nearby network node. As a result, it is no surprise that various types of wireless networks are by far the most commonly used protocols for the IoT. For instance, Wi-Fi is popular for home devices, wireless mesh networks are popular for smart city applications, and so on. Several choices are available, some of which are listed in Table 2.
Some of the popular wireless networking options in use today and emerging for the future
Device manufacturers must consider a range of use case dependencies: Is their device meant to be fixed or mobile? Will it have a power source or need a battery? Will high or low data rates be necessary? Does connectivity need to be continuous or episodic?
The other key component in this layer of the Thing Stack is the microcontroller, which is essentially a tiny computer on a chip. It includes a processor, a small amount of random access memory (RAM) to hold data, a few kilobytes of erasable programmable read-only memory (EPROM) or flash memory to hold any programs on the device, and some solid-state memory for caching data until it can be uploaded. The microcontroller runs the programming that captures and digitizes sensor data, performs any initial analytics or data summaries, and manages its transfer to a hub or the cloud.
Health and fitness devices used in running, exercise, or bicycling store data during the activity and synchronize with the cloud when in proximity to a hub device, such as a smartphone, PC, or tablet. The vendor offerings for this layer of the Thing Stack today span various networking protocols, architectures, hardware, and software. Given the diversity of use cases and environments, the good news is that manufacturers have many options when adding networking and computing capabilities to their devices. The most important emerging capabilities are the following:
Many vendors are creating solutions to bring networking and computing to things. Table 3 provides a sample. Some vendors focus largely on hardware components; others provide a system of hardware and software solutions. In some cases, the software components begin spilling over into the third layer of the Thing Stack, which PwC calls the service platform and discusses in the next section.
Sample of vendors that provide solutions for embedding networking and computing in things
While new inexpensive microcontroller platforms such as Raspberry Pi and Arduino can simplify the task of getting things on the Internet, they also create new risks. The devices can be attacked and hacked, making it possible for hackers to assume control, falsify information, or change system behavior. The network enabling of devices also increases the need for robust vulnerability testing and threat management of the overall system.
Given diverse devices, networking methods, and use cases, the market is fragmented and there are legitimate concerns about interoperability, data and interface standards, security standards, governance standards, and so on. Consortia are forming so devices and services from different vendors can work with each other and create an intelligent fabric useful to all. For instance, 10 companies, including Logitech and Basis, recently created the nonprofit Internet of Things Consortium to work toward creating interoperability by promoting an open approach to integrating with other companies.13
Adding sensors as well as networking and computing capabilities to any device will add cost to the bill of materials. But the rapid reduction in these costs, and the simplification of networking to a plug-and-play experience, creates a huge potential for a much larger universe of things to embed connected sensors. Wi-Fi chips are available in the $4 to $5 range today compared to more than $16 in 2002. Similarly, the price of Bluetooth chips is less than $1, compared to about $20 when introduced in 2000.
New capabilities are also simplifying the connecting of devices to hubs and networks. Texas Instruments has unveiled a Wi-Fi module, SimpleLink Wi-Fi CC3000, that simplifies the task of connecting devices without screens to a Wi-Fi network by using a smartphone.14 Electric Imp has created the BlinkUp feature for setting up a Wi-Fi network by having a smartphone blink at a device with an Electric Imp board in it. The smartphone’s blinking transmits the information to configure the network to the Imp board. With such features, there is no need to add a display or input device purely for network setup. As a result, the complexity and cost of bringing connectivity to devices can stay low.
Power requirements to keep untethered devices running has been one of the big barriers to greater penetration by smart devices. This challenge is being addressed in several ways, including chips designed to perform with minimal power. Another effort is new network protocols to reduce the power needed to wirelessly transfer data. “Bluetooth 4.0, with its Bluetooth low energy (LE) feature, augments the 3.0 specification by adding a network architecture optimized for power conservation. It does this by transporting data in bursts rather than a continuous stream at high data rates, as is needed for audio media,” says Suke Jawanda, chief marketing officer of the Bluetooth Special Interest Group (SIG). For example, a coin cell in a heart rate monitor that lasts for two to three months with Bluetooth 3.0 can last up to two years with Bluetooth 4.0. “Now you can, in essence, bring things into the connected world and completely change the scenario and the user experience for the consumer, the OEM [original equipment manufacturer], and other service providers,” Jawanda says.
The Internet Engineering Task Force (IETF) has specified an IP-based protocol that small sensors and things can use to participate in a personal area network called 6LoWPAN. It allows things to connect to IP networks without a gateway. Some network protocols in use today tend to be popular in specific markets—ZigBee in building automation and Z-Wave in home automation. Not having actual IP addresses does make it harder to integrate devices using these protocols with Internet applications without using a gateway for translation. 6LoWPAN solves this issue, creating the potential to greatly increase the number of small and low-power devices that define the IoT. Figure 3 compares 6LoWPAN with the Bluetooth and Wi-Fi protocols.
The power and throughput ranges supported by the various wireless networking protocols
The first two layers of the Thing Stack embed sensors and tiny, networked computers in the things in the physical world. In some use cases, such as digital thermometers, nothing else is needed. Many emerging use cases, however, take advantage of a third layer, which PwC calls the service platform. These platforms can include middleware, analytics, and application software that usually combine sensor-originated data with other contextually relevant information. One of the most important roles for service platforms is to establish a feedback loop between the things at the edge of the network and the management systems that monitor, maintain, or upgrade the device firmware.
In most cases, service platforms are in the cloud and take advantage of cloud computing traits such as multi-tenant software architectures that scale efficiently. In many ways, the service platforms are the backbone of creating post-transaction relationships because cloud infrastructure can power business applications and end-consumer applications that greatly extend the utility of raw sensor data.
The key capabilities of these platforms are to perform the following:
The vendors providing solutions for these platforms today can be divided into two groups, mobile network-centric platforms and web-centric platforms, as listed in Table 4.
Sample of vendors that provide service platforms to enable post-transaction relationships
The first group defines a market often referred to as machine-to-machine (M2M) solutions. They support the need for devices to connect to commercial cellular networks, mediating the many technological variations and presenting a consistent, standard access method. In the past, M2M technology was limited to organizations, such as government and transportation and logistics companies, which had the size and capacity to build proprietary data networks. However, with widespread coverage of cellular networks globally, M2M can be used by organizations of all sizes. Figure 4 shows the typical M2M platform arrangement and how these platforms interact with IT systems at an enterprise to support employees or customers.
Use of a mobile network in an M2M deployment
M2M solutions offer a value proposition to both carriers and enterprises. For carriers, they create the potential to address a new market and business model based on connecting devices that are not human mediated, such as mobile phones. For enterprises, they reduce the complexity of using a carrier’s network and bring together a rich set of features that can power post-transaction relationships.
“A big part of what we do is creating an operational standard, so that enterprises can run their connected device initiatives consistently across operators,” says Namie of Jasper Wireless. “Even though we work with multiple operators, we enable enterprises to have one standard way of operating, reporting, provisioning, configuring, and controlling their devices worldwide.”
According to Forrester, in 2016 approximately 450 million connected M2M devices will generate nearly $17 billion in connectivity revenues for carriers globally, a 34 percent CAGR between 2010 and 2016.15 Vendors active in this market are Jasper Wireless, Numerex, RACO Wireless, Wireless Logic, and others.
The second group of service platform solutions uses web-centric technologies and facilitates connections at the data API level to make products smart. These platforms bridge the connection between the sensors, which surface the data, and the web or mobile apps over which services are experienced. Typically, these platforms assume the presence of network connectivity and therefore are network protocol agnostic and do not aim to simplify the complexity of provisioning or configuring network connections. They leave that to the device manufacturer or to the vendors that provide the network and computing layer.
Most vendors in this class are emerging companies that are innovating how post-transaction experiences will be monitored, managed, and facilitated. Cosm, EVRYTHNG, IFTTT, Sensinode, SensorCloud, ThingSpeak, and Wovyn are some of the vendors. A few of the vendors that provide network and computing layers, such as Bug Labs, Electric Imp, and Ninja Blocks, also couple a service platform with their hardware solutions for networking and computing.
The continuing reduction in size and price of sensors and the ability to network every device is propelling companies to more fully engage with customers, going beyond transactions to goal-oriented post-transaction relationships. The digitization of consumption and the convergence of the physical and digital worlds is creating opportunities to evolve single transactions into long-running relationships. To do so, devices need a Thing Stack composed of a sensor layer that captures information from the environment, a network and computing layer that shares that information over computer networks, and a service platform layer that orchestrates the overall experience to deliver value-added services to consumers.
The convergence of the physical world with the digital world will put stress on other parts of business and IT operations. In particular, it will increase the burden of data storage, tagging, management, and analysis and related activities. Solutions emerging from mobile and social technologies and big data analytics will address some of these issues.
Behind any purchase is a customer goal or desired outcome. Customers buy insurance to cover unpredictable losses; they buy running shoes to lose weight; or they use new golf clubs to improve their golf scores. In light of the emerging technologies discussed in this article, the challenge to enterprises is to rethink their relationship to their customers’ goals. By selling customers smart things rather than dumb things, enterprises can surface information that guides their customers to more successful outcomes.
1 The term “Internet of Things” was first used by Kevin Ashton in 1999, according to Wikipedia. The initial use referred to industrial technologies such as radio frequency identification (RFID) for radio tagging items in the supply chain. More recently, the use of the term has expanded to include the machine-to-machine communications market and broadly the ability to embed sensing and connectivity into any physical environment, so it can be used in or acted upon by a service.
2 For more details, see http://www.hapilabs.com.
3 For more details, see http://beamtoothbrush.com.
4 “New Embedded Features Will More Than Double Sensor Sales by 2016,” IC Insights, June 11, 2012.
5 Dario Borghino, “Sensordrone adds more sensory capabilities to smartphones,” Gizmag, June 21, 2012.
6 “Vibrating tattoo alerts patent filed by Nokia in US,” BBC, March 20, 2012.
7 Charlie Osborne, “Microsoft develops glucose-monitoring contact lenses,” SmartPlanet, January 6, 2012.
8 For more details, see http://proteusdigitalhealth.com/technology/.
9 Dara Kerr, “Google Glass development charges ahead,” CNET, January 1, 2013.
10 Paula Doe, “Sharply Falling MEMS Prices Spur Rising Demand,” SEMI, July 6, 2010.
11 “The rising value of linked information,” PwC Technology Forecast 2012, Issue 2.
12 HP CeNSE: Sustainable Brands ’11 keynote.
13 For more information, see http://www.iofthings.org/.
14 Texas Instruments, “Latest TI SimpleLink Wi-Fi CC3000 module simplifies home network setup and improves user experience,” news release, January 3, 2013.
15 Michele Pelino, M2M Connectivity Helps Telcos Offset Declining Traditional Services, Forrester Research Inc., December 2011.