6G: It's All About Networks---Human and Machine
It’s 2030, and you’re networking with colleagues in a crowded banquet hall, where you hear parts of hundreds of conversations. Everyone in the room has a connected device (or two, or three, or 10), which are also communicating with each other and with their base stations. And even with all this noise, people and devices are communicating seamlessly.
Whether it’s people or machines shaking hands (or bumping elbows), networks are critical to Lingjia Liu’s research.
Liu, ECE professor and Associate Director of Wireless@VT, is working on the technologies needed for the next iteration of wireless communications—6G. Critical to his success are not just communications networks, but also his network of relationships with industry partners, government research labs, former students, and colleagues.
The Path to 6G
Although we talk about wireless communications in terms of 4G, 5G, and now 6G, Liu explains that these are iterative steps rather than huge leaps. There are intermediate steps between each large release, he notes. Between 4G and 5G, for example, we saw LTE-Advanced Pro. Before it’s time for 6G, we’ll see 5G-Advanced.
Still, researchers like Liu, who contributed to the 4G LTE-Advanced specification, are looking two—or more—steps ahead.
According to Liu, 6G will bring us “higher data rates, higher reliability, higher density, and higher intelligence.” The big push, he notes, is for ubiquitous, efficient, and intelligent connectivity—for all devices everywhere. Connectivity no longer just means our cellphones need to stay connected to a cell tower as we drive down the highway—in 10 years, we’ll be dealing with cellphones, self-driving cars, smart cities, CubeSats and the innumerable connected devices that make up the Internet of Things.
“5G is starting to pave the way for smart cities and autonomous vehicles, but 6G will capitalize on those initial ideas,” says Liu. “At the end of the day, we should be able to connect any device, wherever we are in a smart and efficient way.”
One enabling technology for connecting all these devices is the massive multiple input multiple output (MIMO) technology, where each device incorporates a massive number of antennas. Unlike humans, these devices can then communicate different messages to different places at the same time. According to Liu “it’s easy in theory, but in reality the hardware and complexity is difficult to scale to hundreds or thousands of simultaneous communications.”
Liu is tackling the algorithms behind massive MIMO devices. Similar to when people all converse at once in a crowded room, “when you need simultaneous communications on the same frequency, those signals will interfere with each other,” he explains. Liu and his team are developing algorithms for transmitters to minimize this interference.
Another challenge is scaling these algorithms to hundreds or thousands of users—when you can only communicate with a fraction of those users at once. “You have to determine which users to choose to maximize both throughput and fairness, which can present conflicting objectives,” says Liu. Traditionally, these problems are solved by modeling, but as we connect more and more devices, the complexity becomes too great for traditional models to be sufficient.
Machine Learning Joins the Team
This is where machine learning comes in. Machine learning requires a lot of training, overhead, and computational power, explains Liu. But when the models become this complex, and must adapt to a rapidly changing environment, machine learning rises to the challenge.
Even for 4G and 5G systems, there are tens of thousands of variables to account for. “Even with the perfect equation, there are thousands of parameters. It’s too complicated. This is where machine learning can provide benefits.”
Liu also uses machine learning to optimize high frequency communications—millimeter wave, or even Terahertz. For such high frequencies, Liu explains that “the devices are far from being ready. When you move to those high frequencies, there are a lot of imperfections in the devices. Machine learning can help you learn and compensate for these imperfections.”
To handle all these communications, our communications infrastructure will also become far more diverse. We already have communications satellites that can deliver high speed internet to areas that are not well served by cell towers and other methods. As we see more of these, as well as communications from high altitude balloons and aircraft, one challenge becomes how to integrate them together.
“A key feature of cellular communication is mobility—you don’t feel interrupted when you move from one tower to another. This is called mobility control,” explains Liu. But when you need to move not just from cell tower to cell tower, but from cell tower to high altitude balloon, to satellite, to airplane, and back to a cell tower, it becomes more complicated.
In addition to moving between different types of nodes, there are additional challenges for these new nodes: higher velocities, Doppler spread, and environmental factors that can interfere with communications.
Nothing is off limits for finding a solution, including changing the waveform itself. “The current waveform may not be enough for these different nodes to work together,” says Liu. “We’re looking at how the fundamental waveform could be redefined.”
Although his designs might be preparing for our future communications needs, Liu is looking beyond theoretical models. “Our work is not only in theory. We do a lot of prototyping, and it’s one of the most important features of our research,” he says. When you show that your ideas are working in hardware, he continues, “you don’t have to argue for your ideas or assumptions.” It just works.
In early stages of new technologies, researchers traditionally simplify their models with various assumptions, working on crafting theoretical frameworks and solutions. These complex communications challenges, even in the not-yet-realized 6G, can’t afford simplification, notes Liu.
These technologies must work with all the messiness that is the real world. “In industry, you cannot make too many idealized assumptions, you have to make things work,” notes Liu. “We work very closely with our industry partners to make sure our ideas and innovations will solve the challenges of reality, not just address theoretical models.”
As with so many challenges, Liu notes that the hardest part is defining the problem appropriately. And to do that, he relies on his industry contacts. “Industry knows 5G and 6G systems best. To really make an impact, you need to collaborate with industry leaders and become a trusted partner so you can identify the problems.”
Core to Liu’s group’s success is his ties to the telecommunication industry, both from his own network and close ties to his former students. “I want my students to feed back into my group,” says Liu. “I want them to work with us, provide internships, and network—with current and past students.”
And there are plenty of former students for the group to connect with. Even during the pandemic, Liu’s team graduated six Ph.D. students. “It’s very rewarding to see how hot those students were in the job market, even during the pandemic,” says Liu.
All these connections are vital, according to Liu. “If you know the problems, and have the industry connections, you know what is coming, and you know what challenges to solve.”
When Liu and his team find themselves in that crowded room, they are part of all those conversations, solving the challenges that will take us to the next level of connectivity.