Brinley Macnamara (host) (00:03):

So why will our future 5G and 6G networks need to operate at higher frequencies than past generations like 4G?

Dr. Garry Jacyna (00:13):

Well, primarily it’s all about information needs that require higher data rates and capacity from multiple state high definitions streaming services, remotely controlled medical procedures to giving consumers new levels of control over their homes and cars using the ubiquitous Internet of Things. 6G has even loftier goals such as holographic teleportation. So these applications require lots of bandwidth and bandwidth with scales with frequency. So really long story short, millimeter waves offer new applications, which some are essential, and more importantly, new revenue opportunities.

Brinley Macnamara (host) (00:58):

Hello and welcome to MITRE’s Tech Futures Podcast. I’m your host, Brinley Macnamara. At MITRE, we offer a unique vantage point and objective insights that we share in the public interest. And in this podcast series, we showcase emerging technologies that will affect the government and our nation in the future. Today, I’m going to tell you about Kevin Burke and Dr. Bindu Chandna’s recent MITRE investigation into the unique performance requirements of our future cellular networks i.e., 5G, 6G, and beyond, which we collectively refer to as xG. More specifically, I’m going to tell you about a novel technology, known as millimeter waves, that will be critical to unlocking the bandwidth intensive promises of our future cellular networks, paying special attention to the unique challenges that come with figuring out just how to accurately model a millimeter wave’s performance in a variety of environments. But before we dive in, I want to say a huge thank you to Dr. Kris Rosfjord, the Tech Futures Innovation Area Leader in MITRE’s Research and Development Program. This episode would not have happened without her support. Now, without further ado, I bring you MITRE’s Tech Futures Podcast episode number eight.

Brinley Macnamara (host) (02:24):

To understand what it means for frequency to scale with bandwidth, it’s important to first understand the biggest problem with our current 3G and 4G networks, which is, the spectrum that dominates 3G and 4G in the range of six gigahertz and below, is extremely crowded. And since spectrum is a limited and highly regulated resource, this traffic congestion is making it harder for network operators to handle their users growing demands while also increasing performance. But in recent years, network operators think they found a solution to this traffic congestion problem. And according to Dr. Chandna, a Principal Signal Processing Engineer in MITRE Labs and the Principal Investigator on the Millimeter Waves project, this solution lies in transmitting data over higher frequencies of the electromagnetic spectrum, sort of the spectrum equivalent of taking the back roads.

Dr. Bindu Chandna (03:16):

The higher millimeter wave bands have not been explored. And there are very few systems present there, which means we should be able to use the larger chunks of bandwidth to get high data rates.

Brinley Macnamara (host) (03:30):

Millimeter waves fall into the range of 28 to 300 gigahertz. And they get their name, millimeter waves, from their super small wavelengths, measuring between one to 10 millimeters. But as Dr. Chandna noted, their true value lies in the fact that barely anyone is using millimeter waves right now, so network operators see millimeter waves is a promising option for the expansion of bandwidth in 5G and beyond.

Kevin Burke (03:52):

Use of millimeter waves for commercial communication purposes is relatively new to 5G.

Brinley Macnamara (host) (04:01):

That’s Kevin Burke talking. He’s a Lead Signal Processing Engineer in MITRE Labs and was a Co-Principal Investigator on the millimeter waves project.

Kevin Burke (04:08):

But the other thing about those higher bands is they typically are more useful at shorter ranges because the nature of the propagation is more directional and things like buildings and other obstructions in the environment can block the propagation of the RF energy that’s needed to make the transmission. So, there was a necessity to understand these particular transmission bands better in order to provide the coverage and the capabilities that was envisioned in 5G and going forward from there.

Brinley Macnamara (host) (04:42):

Now, the typical way that researchers go about forming a better understanding of how different parts of the electromagnetic spectrum behave is through a technique called channel modeling. So next question is, when researchers develop channel models for any signal, whether it be SONAR or whether it be for electromagnetic waves that we have in RADAR and our cellular networks, what is usually their goal when they’re developing these channel models?

Dr. Garry Jacyna (05:08):

Well, I think it comes down to understanding the limitations that the environment imposes on a given signal based on its application.

Brinley Macnamara (host) (05:16):

That’s Dr. Garry Jacyna talking. You heard him at the very beginning of this podcast. He’s a Fellow Emeritus in MITRE Labs and a recognized expert in signal processing.

Dr. Garry Jacyna (05:25):

So for example, the channel requirements for a communication receiver may be different than the requirements for RADAR or signal intercept receiver. However, the goal is the same to better understand the constraints and limitations involved in deriving useful information from this signal that has been corrupted by the channel.

Brinley Macnamara (host) (05:46):

So to recap, signal processing engineers like Dr. Jacyna, whose job it is to figure out how we can manipulate signals, such as millimeter waves, to communicate messages like emails and cat videos often refer to a signal’s channel as a function that maps an input signal from a transmitter to an output signal at a receiver. And in a perfect world where things like buildings, inclement weather, and Earth’s atmosphere didn’t exist, channel modeling would be trivial because there would be nothing to corrupt the input signal on its path from a transmitter to a receiver. But we live in an imperfect world that is chock full of pesky channel corruptors like buildings, inclement weather, and Earth’s atmosphere, and while this is certainly a downside for guaranteeing the reliability of our daily activities that depend on the transmission of electromagnetic signals. Think anything you do with your cellphone.

Brinley Macnamara (host) (06:36):

On the upside, it means that signal processing engineers have a whole lot of job security, especially when it comes to developing channel models to predict just how good a particular part of the electromagnetic spectrum will be at transmitting large sums of data. And when developing channel models for millimeter waves, the first key step is determining the large scale path loss of millimeter wave channels. Can you define what large scale path loss actually means in terms of the propagation of a millimeter wave?

Dr. Bindu Chandna (07:03):

There’s nothing specific to millimeter waves here, but large scale path loss models are basically saying that, can we talk of what is going to be the, on a large level, on a big level between if the distance between the transmitter and receiver is so much, what will be the attenuation of the signal?

Brinley Macnamara (host) (07:28):

Thus, any signal, whether it be a millimeter wave, sound wave, or visible light loses power i.e., attenuates as a function of distance. And this phenomenon is known as a signal’s large scale path loss. And for electromagnetic waves, there are a variety of well understood physical laws that govern large scale path loss. Most of these are out of scope for this podcast, but I will reiterate two key things we know about millimeter wave path loss. Number one is that millimeter waves will attenuate more quickly than lower frequency waves as a function of distance, meaning they are worse at carrying data over longer distances. And number two is that millimeter waves attenuate much more quickly when encountering objects in their environment, meaning they are worse at carrying data through objects, like buildings and trees. With this in mind, I asked Dr. Jacyna about why he thinks the development of large scale path loss models is the most logical first step in modeling the performance of a network.

Dr. Garry Jacyna (08:22):

Well, putting on my system engineering hat, you really need a high level system performance model to determine if a proposed network meets the technical requirements under average operating conditions. The what I call gotchas uncovered under extreme conditions or say specific operational areas are examined using more detailed path loss models. But the first step, at least in my mind, is to understand average performance.

Brinley Macnamara (host) (08:56):

Now I should stop here to make the important disclaimer that Kevin and Dr. Chandna did not develop their own channel models as part of their research. Rather, they focused on comparing different millimeter wave channel models that have been developed by researchers outside of MITRE. And through this work, they noticed one big divide between these researchers methodologies. And the crux of this divide is whether researchers used a physics approach or a data fitting approach to developing their channel models.

Kevin Burke (09:21):

There’s the physics based model, which is looks at a fairly simple characterization of having a radiator of electromagnetic energy and what it’s characteristics are versus frequency and as that energy propagates away from the radiator. Some of the models that have been put out for use in 5G are based on this approach. But again, like all models, it has its applicability in a general way without being perfectly precise for a very specific situation. The data fitting approach, if one were to make a lot of measurements, collect a lot of data in a specific area, and then coming up with a model that fits those measurements, it would be very accurate for that particular setting. But if you were to move to another setting that at least at a high level was considered to be similar geographically, you may find that there would be variations from the data fitting approach.

Brinley Macnamara (host) (10:34):

And while the real world comprises an infinite number of unique settings for millimeter wave propagation, there are only a small number of broad environmental settings that will undoubtedly have an impact on the average performance of millimeter waves under average conditions, which are the following. Whether a channel is one, terrestrial or two, aerial, and whether the channel is situated in an urban, number three, or rural environment, number four. In fact, these four settings have such an outsized impact on a signal’s large scale path loss that channel modelers commonly develop models that can be parameterized with combinations of these four settings. At some point in time, we’ve all run into the constraints the characteristic clutter in our urban propagation channels. Ever lost phone service inside of an elevator? That’s urban clutter getting into your way.

Brinley Macnamara (host) (11:21):

Likewise, we probably experienced the downsides of rural propagation channels as well. Ever lost phone service while on an off the beaten path vacation with a gang of crazy relatives? In this instance, one reason you might have lost your lifeline to the outside world is that electromagnetic signals propagate for longer distances in rural channels, which means cell towers are fewer and farther between, usually when it’s least convenient. But the inclusion of terrestrial and aerial channels in millimeter wave channel models made less sense to me. So I asked Dr. Chandna about why she thinks this particular environmental distinction will be equally important in our future networks.

Dr. Bindu Chandna (11:56):

The way the development is happening today, we are looking at kind of a seamless coverage everywhere. And that’s going to involve whether drones and other kinds of methodologies in between our terrestrial to non-terrestrial connectivity overall. There’ll be UAVs. There could possibly be going into xG system, some SATCOM being used for commercial purposes also. But so our kind of plan was to kind of see overall how the systems perform and we needed both terrestrial and non-terrestrial channel models for all our research here.

Brinley Macnamara (host) (12:38):

For a good portion of our interview, Dr. Chandna and Kevin talked a lot about some significant differences they found in the path loss models they compared. And while I can’t elaborate on all of the differences they found today, I do want to highlight one example that jumped out at me. The model in question was of a millimeter wave’s large scale path loss in an urban terrestrial environment. One group use a physics based approach to develop their model while another group use the data fitting approach. To Dr. Chandna and Kevin’s surprised, they found that beyond a thousand meters, the models’ path loss differed by eight decibels. That’s more than a factor of two. Why should we care about these differences in a factor of two? For example, what implications do they have on the decisions that network operators will make about resource provisioning?

Kevin Burke (13:23):

A factor of two may not seem like a lot, but actually the factor of two in terms of large scale path loss might imply that you would need twice the transmit power in one case compared to another, for one particular model, as opposed to the other model, in order to get the same received signal power at the receiver. For commercial carriers, they make very large investments in infrastructure equipment. So they usually invest in sophisticated tools and do a great deal of modeling in order to be able to size and specify the nature of all equipment they need when they’re making an upgrade to a new standard or a new mode of operation. The development of models, it’s certainly… It’s kind of an art and kind of a science. Actually, there’s a quote that said that all models are wrong, but a few are useful. So the art that comes into creating models is in part the realization that none of these are going to be perfect for all situations.

Brinley Macnamara (host) (14:36):

And this idea that all models are wrong, but some are useful, is at the core of what Dr. Chandna and Kevin were trying to do in their comparison in millimeter wave channel models. Knowing that none of these models would be right for every situation i.e., they’re all somewhat wrong, Dr. Chandna and Kevin’s mission was to figure out which of these models would be most useful for which situations, paying special attention to the millimeter wave propagation settings that will be most critical to MITRE’s sponsors. And I couldn’t help, but ask Dr. Jacyna about what he thinks about the now famous George Box quote, “All models are wrong, but some are useful.”

Dr. Garry Jacyna (15:11):

From experience, I totally agree with this because let’s face it, science itself is provisional. And I recall that Karl Popper said that it’s not science if it can’t be shown to be falsifiable. So Newton’s Law of Gravity was shown to be false under certain situations, and obviously was replaced by Einstein’s theory of general relativity. However, Newton’s laws are extremely useful under restricted circumstances. The same goes for any modeling endeavor. In my mind, models are used as pathfinders. They really light the research path and guide us to potential solutions.

Brinley Macnamara (host) (15:58):

This show was written by me. It was produced and edited by Dr. Kris Rosfjord, Dr. Heath Farris, and myself. Our guests were Dr. Garry Jacyna, Dr. Bindu Chandna, and Kevin Burke. The music in this episode was brought to you by Dream Cave, Ooyy, Gavin Luke, Michael Keeps, and Truvio. We’d like to give a special thanks to Dr. Kris Rosfjord, the Technology Futures Innovation Area Leader for all her support. Copyright 2022, MITRE PRS # 22-0259, February 8th, 2022.

Brinley Macnamara (host) (16:35):

MITRE: solving problems for a safer world.