Internet of Things December 2018/January 2019 Viewpoints
2018: The Year in Review
By David Strachan-Olson
Strachan-Olson is a consultant with Strategic Business Insights.
The year 2018 was great for the Internet of Things (IoT): IoT adoption was strong among consumers and in a wide variety of industries, including manufacturing, retail, and medicine. Adopting IoT cloud platforms continued to be the go-to way for companies rapidly to build out their IoT deployments and data-sharing infrastructure. The consumer market for connected devices continued to focus on virtual assistants, which increasingly appeared in new products—including smart displays and vehicles. With the implementation of the European Union's General Data Protection Regulation (GDPR), companies had to become more aware of individuals' data rights and consider the types of data IoT devices might collect. Major players in the telecommunications industry prepared for 5G by designing new devices that support the new radio standard and by planning for network deployments in 2019. Finally, the importance of embedded artificial intelligence became clear, and many companies focused on developing custom processors to improve the performance of machine-learning algorithms in IoT and mobile devices.
Strong Growth in Many Industries
Companies and organizations across a wide range of industries continued to increase adoption of IoT devices and technologies throughout 2018. Although many projects are just trials, corporate surveys show that most companies are thinking about the IoT and taking actions to adopt IoT technologies. Rather than focusing on new hardware, many enterprises' first steps focused on choosing an IoT and data platform to make the most of data that companies already collect. Many companies found significant value just by making existing data easily accessible across their organizations. Cloud IoT platforms, such as Amazon Web Services and Microsoft Azure, continued to grow in 2018 and to offer more services to support IoT deployments and IoT data, such as machine-learning algorithms and security solutions. Start-ups such as C3 and Uptake Technologies also found success in the market for industrial cloud platforms by targeting specific industrial operations. In addition to strong growth, many companies and cloud platform providers were vocal about the importance of security for IoT deployments. However, whether companies are serious about security or just echoing consumer and industry concerns is unclear.
Consumer Devices: Virtual Assistants Lead the Way
Interest in home IoT devices increased throughout 2018. Companies continued to refine existing products through software updates and to introduce new products. Smart speakers that incorporate virtual assistants continued to be a rapidly growing device category. Amazon and Google currently dominate the market for such devices in Europe and North America, but Apple joined the fray with the release of its HomePod in February 2018. A study by Adobe Analytics estimated that 32% of US consumers reported owning a smart speaker in August 2018. Adobe estimates that by the beginning of 2019, just under half of US consumers will own a smart speaker with a virtual assistant. Adobe's study also found that device owners use virtual assistants for a wide range of tasks, including playing music, checking weather forecasts, conducting online searches, and checking the news. Less common use cases included controlling smart-home devices and shopping.
Throughout 2018, companies focused on moving beyond smart speakers to smart displays. Amazon released its version of a smart display in 2017; in 2018, Google and its hardware partners released several smart displays that work with Google Assistant. Later in 2018, Facebook introduced a smart display specifically for video calling. Typically, smart displays enable virtual assistants to display information relevant to user queries and can enable additional functions, such as follow-along recipe instructions. The popularity of smart speakers and smart displays encouraged some households to explore other connected home devices that integrate easily with virtual assistants. Markets for other types of smart-home devices—such as security cameras, door locks, thermostats, switches, and lights—continued to grow in 2018, but product prices are dropping as competition between companies increases.
Data Handling and Data Privacy
In May 2018, the European Union's GDPR came into force. The GDPR is a revised version of previous data-privacy standards but is most notable in that it grants much larger enforcement powers against companies that violate its regulations. Under the GDPR, companies that commit violations could sustain fines of up to 4% of their global annual revenue. The GDPR grants several key rights, including access to one's personal data, portability of data, and the right of deletion. Although the GDPR has mostly affected websites, the principles governing personal data are equally applicable to IoT platforms and services that handle personally identifiable data. Consumer interactions with smart-home devices and sensors in other spaces could generate data about individuals and thus fall under GDPR protections. Decisions about how to handle user data in countries outside the European Union may have a strong shaping effect on international common practice and, ultimately, legislation in non-EU countries. Late in 2018, California passed a consumer-privacy act that is similar to but less sweeping in scope than the GDPR. The act will go into effect in January 2020. Global companies face the choice of making their services EU-compliant worldwide or adopting various data-handling approaches on the basis of where data originate, potentially making back-end infrastructure much more complex.
5G and LTE IoT
The majority of the telecommunications industry spent 2018 preparing for 5G network and device deployments coming in 2019. The 3GPP (3rd Generation Partnership Project) approved the 5G New Radio (NR) standard in December 2017. Late in 2018, Qualcomm announced its first-generation 5G models compliant with the NR standard. The 5G NR standard uses frequency bands as high as 40 gigahertz (GHz), which are higher than those LTE communications use. Higher frequency signals enable higher data-transfer rates at the expense of being vulnerable to physical obstruction, and they are practical only for relatively short-range communications. This first generation of 5G hardware and networks will primarily target enhanced mobile broadband and ultrareliable, low-latency communications, which will find use in mobile devices and mobile hot spots. Concerning the IoT, the first generation of 5G components will target only high-performance IoT use cases—for example, cloud-controlled mobile robots and unmanned vehicles—rather than vast networks of small, low-power devices. In March, Baidu demonstrated an autonomous vehicle that uses a 5G network for communications.
For some IoT applications, companies will require large numbers of devices that are inexpensive and low power and have minimal data needs; 5G NR will not affect these applications for a few years. In the meantime, such applications are being served more frequently by LTE IoT. LTE IoT, which was introduced as part of a previous 3GPP standards release, offers two low-power wide-area narrowband technologies that scale LTE networking down to inexpensive and low-power devices. The two technologies are enhanced machine-type communication (eMTC) which has data rates up to 1 megabit per second, and narrowband IoT (NB-IoT), which has data rates up to 100 kilobits per second. Some US cellular companies already offered eMTC before 2018, and both T-Mobile and Verizon began rolling out NB-IoT networks in 2018. The expansion of LTE networks to support low-power devices will bring competition to the market for low-power wide-area networks.
IoT and AI
The combination of IoT and AI technologies continued to be a significant trend in 2018. IoT sensors have the ability to generate huge quantities of data. The data are useful for engineers, data scientists, and business analysts to explore after the fact to analyze operations and evaluate changes. However, a number of opportunities enabled by the IoT are accessible only if sensor data see analysis in real time, which necessitates complex data processing and conditioning and algorithms and machine learning. Typically, machine-learning algorithms are in the cloud, but 2018 saw a significant shift toward machine-learning algorithms seeing in-device implementation. Many companies—including Qualcomm, NVIDIA, and Intel—are developing specific AI chips and processors for accelerating the performance of neural networks in low-power devices. Embedded AI enables devices to operate with decreased latency, operate temporarily without connections, reduce communications volume, and potentially increase security and privacy. Throughout 2018, a number of companies began exploring the use of embedded AI in connection with IoT and Industrial 4.0 deployments, mainly in the form of AI-powered cameras (see the next section).
AI-powered cameras are perhaps the most compelling example of integration between an IoT sensor and embedded AI. Cameras capture a huge quantity of data, and uploading this data to the cloud requires significant communications bandwidth. At the same time, photo and video analysis is one of the most researched areas of machine learning, and companies are making rapid and significant advances. Embedded machine-learning algorithms enable the device to monitor the environment for significant events and to send notifications and begin streaming video only when significant events occur. This mode of operation could reduce communications traffic greatly and enable low-latency responses to rapidly unfolding events. Applications for AI-powered cameras are far-reaching—including security cameras with object and facial recognition, traffic management, industrial-processes monitoring, and checkout-free shopping.
In late 2017, Amazon Web Services released AI-powered camera DeepLens; in May 2018, Qualcomm and Microsoft announced their Vision AI developer kit, which uses Qualcomm's Vision Intelligence Platform and Microsoft's Azure machine learning and IoT Edge functionality. Many other companies are also developing such technologies and made significant progress in 2018.
Look for These Developments in 2019
- IoT adoption across most major industries—including manufacturing, agriculture, health care, and smart cities—will continue to increase in 2019. Companies will continue to face a challenge in making sure they can extract value from IoT deployments and big-data repositories.
- In the smart-home market, expect high adoption of virtual assistants and for these assistants to encourage homeowners to purchase other connected devices to help automate and control their houses.
- AI will continue to grow in importance for IoT deployments, and device manufacturers will adopt chips and processors that accelerate the performance of on-device machine-learning algorithms.
- 5G networks will see significant deployments in 2019. In 2019, devices will ship that will support 5G NR standards, enabling high-bandwidth and low-latency communications for high-performance IoT applications.
- ARM will continue to dominate the market for IoT-device processors, but stakeholders should monitor ongoing developments that relate to the adoption of the RISC-V open-source instruction-set architecture.