Let the machines talk

For most of us, the allure of 5G—the fifth generation of cellular technology—has been faster speeds to download a movie, send a photo, or dash off a message.

Researchers know that what’s really important about 5G has little to do with our phones. It is the promise for machine-to-machine communication.

We conceive of cellular networks as handling our calls, texts, and streaming video, but there are more crucial users we don’t see: smart devices with their embedded sensors, robots and other forms of industrial automation, drones, and mobile health applications. Within two years, half of connected devices will be machines talking only to other machines, according to Electronics360, an online publication that covers communications.

These increased demands have strained the networks where these devices operate: 4G, 4G LTE, and WiFi. This is why 5G is so important: it expands the electromagnetic spectrum used by devices from the radio frequency band into the neighboring microwaves, taking advantage of millimeter waves and using frequencies of 30 to 300 gigahertz—10 to 100 times higher than what’s used for WiFi and 4G communications. In this region, higher communication rates and lower latency are possible.

For us, that means downloading a movie in a matter of seconds, but that’s not what matters to researchers. “From a pure technological perspective, the shift is from internet browsing and streaming of data to machine-to- machine communication and the Internet of Things,” said Daniela Tuninetti, a professor of electrical and computer engineering. She had her colleagues in UIC’s NICEST Lab—Besma Smida, Hulya Seferoglu, and Natasha Devroye—are working on how to best communicate data over networks, from theory to practice.

Machine-to-machine communication is far more computationally intensive than human communication, bogging down an already strained network, Tuninetti explained. Intelligent autonomous systems, for example, use techniques based on artificial intelligence and machine learning, which a device may not be able to process on its own. It would offload these computations to a cloud computing center, where distributed computers each solve various parts of the problem and communicate the solution back to the device. And machine-to-machine applications often require near-instantaneous calculations: Should the driverless car turn? Is the robot malfunctioning?

The shorter microwaves used in 5G cannot be transmitted via the previous generation of communication infrastructure. They measure between 1 and 10 millimeters—by comparison, radio waves are centimeters long—and are easily absorbed by buildings, foliage, and other objects. Transmitting them requires many more base stations located close together, unlike traditional cell towers.

Now that the technology is here and infrastructure finally is being built, the NICEST Lab experts are refining approaches that will make the most of it, with advanced machinery in mind.

Smida studies full-duplex communication, a technique that could double the use of the available communication spectrum. Currently, messages must be separated by time  or frequency to avoid interference, meaning they cannot be sent and received simultaneously. Smida likens the problem to trying to talk and listen at the same time. She is trying to resolve it using backscatter modulation, which allows her to estimate self-interference and cancel it before it occurs.

“If I can subtract that interference from the receiving side, I can cancel the echo in the electromagnetic wave,” she said. “It’s always about efficiency in wireless communications.”

Meanwhile, Seferoglu is adding efficiency by finding ways to capture and process data as close to the source as possible, including at the edge of a system, using a method known as edge computing. This represents a distinct advantage for the Internet of Things.

“IoT devices are constantly gathering huge amounts of data, which can be sent to the cloud for processing, but there is usually a bottleneck between these devices and the cloud in terms of connectivity,” Seferoglu said. “This makes edge computing a crucial part of the 5G platform.”

Smida, Tuninetti, and Devroye are establishing a framework to determine the fundamental tradeoffs among delay, transmission rate, and reliability for data moving over wireless networks. They’re concerned not just with the technologies that currently exist, but also those that have  yet to come online. “5G is not only about the mobile user experience,” Devroye reinforced.

One of the group’s areas of interest is the optimal assortment of transmissions across the power bands that are available: low, middle, and high. For instance, a hospital may need 5G for high-power and high-speed applications such as robotic surgery, which would require multiple base stations for transmission. Housecat videos on YouTube, however, could safely stay on a lower-frequency band.

Tuninetti said the NICEST research team is keeping an eye on security questions associated with 5G, too, including issues inherent in IoT devices and networks and issues related to 5G base stations. The base stations may be manufactured by companies outside of the United States and, if used for critical infrastructure, could present geopolitical security issues.

“These are very complex pieces of equipment, for which it is difficult to exactly know what everything does,” Tuninetti said. “This vulnerability poses hardware and software security concerns, which industry and academia are hard at work to address.”

“These are things that rightfully people are concerned about,” she added. “If, for example, malware were injected into critical controls for self-driving cars or an energy plant, there could be dire consequences.”

Still, the lab members are optimistic about 5G and about resolving challenges through the next generation of wireless communications. What may not be realized in 5G may be feasible in 6G. At this moment, they are excited to leverage the possibilities of massive connectivity for the machines we have—and those that are to come.