Skip to Main Content

Strategic Business Insights (SBI) logo

Internet of Things August 2018 Viewpoints

Technology Analyst: Guy Garrud

Qualcomm Unveils 5G Antennas

Why is this topic significant?

Next-generation cellular networks (5G) target, in part, the nascent IoT-device market. Qualcomm's early release of compatible hardware could give it considerable first-mover advantage in the 5G-hardware marketplace.

Description

Semiconductor and telecoms hardware manufacturer Qualcomm has announced the first generation of its modules for use in 5G telecoms devices. The QTM052 is a millimeter-wave antenna module; the QPM56xx is a sub-6-gigahertz (GHz) radio-frequency module. Qualcomm reports that these modules are the first fully integrated 5G New Radio (NR) modules available on the market. Several major smartphone manufacturers—including Samsung, LG, and Xiaomi—are working with Qualcomm to develop 5G-enabled devices.

The 3GPP (3rd Generation Partnership Project) approved the 5G NR standard in December 2017. The radios transmit and receive in a number of frequency bands as high as 40 GHz; stakeholders expect bands near 3.5 GHz to see much use. The highest of these frequency brackets uses signals in the millimeter-wavelength range. Using short-wavelength signals—in comparison with using longer-wavelength signals—means that networks can support higher data-transfer rates but at the expense of being vulnerable to physical obstructions. Millimeter signals are more reliant on line of sight, and signals can be blocked by, for example, a user's hand when gripping a smartphone or a person walking in between a device and a cellular-antenna mast. Qualcomm claims to have circumvented the signal-blocking problem by using an antenna array of four antennas that can direct signals toward the nearest cellular station and even bounce signals off nearby objects in order to transmit past obstacles. The new antennas will operate with Qualcomm's X50 5G modem, which can connect to up to four antenna arrays within a device.

Implications

Technology development for 5G has advanced rapidly, from ratification of the initial radio standard to first-generation hardware offerings in the space of less than a year. Qualcomm is already a large player in the smartphone hardware market, and being first to market with antenna and modem hardware for 5G devices will likely put the company in a strong position in the coming months. In an ideal case, device manufacturers could have 5G flagship phones for sale as early as 2019, to capitalize on the expected rollout of 5G-network coverage by major telecoms providers such as AT&T and Verizon.

Less clear, however, is how well Qualcomm's 5G NR hardware will support IoT applications. The company may have developed a practical workaround for line-of-sight issues with millimeter-wavelength telecoms, but incorporating multiple antenna arrays is a high-cost approach. Such costs may be justifiable for a high-end smartphone, but many IoT applications require large numbers of low-cost sensors and actuators, which may make a multiple-antenna system not cost-effective.

Impacts/Disruptions

The rollout of 5G networks will be an important step for enabling many IoT projects. However, the initial rollout of 5G networks and devices will likely focus more on consumer portable electronics and smartphone connectivity than on industrial applications. More industrial applications will come with phase two of the 5G rollout, which Qualcomm predicts will occur sometime about 2021. In the meantime, millimeter-wavelength telecoms may find use in edge-computing applications such as sensor aggregators or other local network nodes.

Scale of Impact

  • Low
  • Medium
  • High
The scale of impact for this topic is: Medium

Time of Impact

  • Now
  • 5 Years
  • 10 Years
  • 15 Years
The time of impact for this topic is: 5 Years

Opportunities in the following industry areas:

Network infrastructure, electronic-device manufacturing, edge computing, telecoms

Relevant to the following Explorer Technology Areas:

Internet of Cameras

By David Strachan-Olson
Strachan-Olson is a consultant with Strategic Business Insights.

Why is this topic significant?

Advances in machine learning are revolutionizing computer-vision systems, which could have a major impact on the development of the Internet of Things.

Description

In recent years, the capabilities of computer-vision systems have progressed immensely because of advances in neural networks and machine learning (ML). Numerous companies have developed computer-vision systems, but such intelligence is primarily limited to the cloud because of the hardware necessary for hosting complex neural networks. Now, numerous groups have optimized ML frameworks and developed techniques for accelerating neural-network performance on mobile-device processors, allowing companies to create cameras with embedded artificial intelligence (AI) for commercial and consumer Internet of Things (IoT) applications.

In late 2017, Amazon Web Services (AWS) released DeepLens, an AI-powered camera, and Sage Maker, a platform for developing and deploying ML algorithms for the camera and other AWS services. In May 2018, Qualcomm and Microsoft announced a Vision AI developer kit, which uses Qualcomm's Vision Intelligence Platform and Microsoft's Azure machine learning and IoT Edge functionality (which enables use of Microsoft's Azure services at an organization's premises, rather than in the cloud). The included camera demonstrates how companies can use Qualcomm's QCS603 system on chip to support intelligent vision applications. Microsoft also recently announced Project Kinect for Azure, which uses the fourth generation of Microsoft's Kinect depth-sensing technology. The device will function with Microsoft's Azure platform and support many features, including machine learning, IoT Edge, and Cognitive Services (which includes, for example, speech and semantic technologies). Many other companies are also developing software and hardware to enable embedded computer vision.

Implications

Although companies can implement intelligent vision systems in the cloud, on-device intelligence greatly decreases bandwidth requirements and decreases the latency of detection and decision making. Companies are just beginning to embed computer-vision technologies into cameras, but such devices have a massive number of potential applications. Microsoft has demonstrated intelligent cameras for use in construction environments to monitor for spills and misplaced equipment and for tracking workers. Amazon uses computer-vision technology in its Amazon Go stores to enable checkout-free shopping by tracking what shoppers pick from shelves. Numerous companies have demonstrated smart security cameras that monitor for trespassing individuals and vehicles.

Most available computer-vision systems work with simple cameras, but Microsoft's Kinect system exhibits the future of intelligent computer vision. Other types of imaging devices that sense depth information or use nonvisible light could improve detection of specific objects or events, especially in industrial applications. For instance, many companies are developing solid-state lidar for autonomous vehicles, which is dramatically cheaper than traditional lidar. Inexpensive lidar could allow intelligent cameras to create detailed depth maps and thereby improve ability to detect objects. In theory, a system could detect counterfeit products that appear similar to real ones but that have slightly different shapes.

Impacts/Disruptions

The Technology Map identifies enhanced computer vision as a competing technology to the IoT. Many stakeholders' visions for the IoT involve selling hundreds of millions of electrical components—including microprocessors, radio-frequency equipment, batteries, and sensors—to connect many everyday objects. However, in many sensing applications, a network of intelligent general-purpose cameras could monitor and sense the state of the world and many objects without the need to connect every object. This capability is a dramatic shift from the traditional concept of the IoT. Continued advances in computer vision and cost reductions of complex imaging technologies, such as lidar, will only further enable a camera-dominated version of the IoT.

Scale of Impact

  • Low
  • Medium
  • High
The scale of impact for this topic is: High

Time of Impact

  • Now
  • 5 Years
  • 10 Years
  • 15 Years
The time of impact for this topic is: Now to 5 Years

Opportunities in the following industry areas:

Computer vision, surveillance systems, smart cities, retail, manufacturing, warehouses, robots, personal assistants

Relevant to the following Explorer Technology Areas: