Solving network's mission impossible

You’re one of the lead inventors of SRCON – Simulated Reality Communication Networks. What is this, exactly? 

SRCON is a “digital twin” for mobile networks. A digital twin is basically a digital image of a real physical structure or entity. By adding different inputs to a digital twin, you can create models that let you predict what will happen in the real world. SRCON helps us predict what will happen on a real-world network.

For example, everyone has a cell phone. When you watch a video or movie on your phone, you connect to a mobile network. But sometimes, when you move around, the connection is bad, and you get poor service. We want to tune the parameters of the network so that everyone gets the best possible experience. But it’s hard – impossible, really – to do that on a real-world network. 

What’s hard about it?

You can’t possibly try out every combination of inputs on a real network. There are too many combinations, and testing them all would take forever. Plus, if you didn’t do everything just right, network service would deteriorate badly – it would be a disaster for users. We can’t risk that.

Instead, we try out different combinations of inputs on a digital model – SRCON – and get the information we need to optimize the network without putting customer experience at risk. 

So SRCON is evaluating customer satisfaction? 

Not directly, no. It looks at certain metrics: data rate, throughput, service delay, and so forth. If you know those metrics, you automatically have a pretty good idea of what your customers are experiencing. Polling customers, for example, is too inefficient and takes too long. We can get the same information with digital twins – and we can get it much faster. For example, SRCON can evaluate the performance of a mobile network covering a metropolis like Shenzhen within 30 seconds.

Because there are so many different combinations of parameters, the numbers get really big.

In a large city, for example, a network might have 500,000 base stations that send and receive wireless signals from phones and other connected devices. Five hundred thousand is a lot – but it’s nothing compared to the combinations of network parameters you’d need to test, in order to optimize how those base stations perform. That number is ten to the power of 2,000,000. So you can’t do an iterative, trial-and-error process. You need some kind of powerful technology to do the work at scale. 

You’re the lead inventor of SRCON. Does that mean you invented it yourself?

We have a large team. I would say about 100 people. The team consists of front-line engineers with years of experience, plus researchers like me who have Ph.Ds. We also have university professors working with us. One principal contributor to SRCON, and in fact the one who proposed the idea of building a digital twin for mobile networks, is Professor Tom Zhi-Quan Luo from the Chinese University of Hong Kong in Shenzhen.

I’m on the algorithm design team, which has about 10 people. We developed the core intelligent design of SRCON.

What did that involve?

To simulate a real-world network, you need to model multiple factors. One key factor is the wireless propagation environment: how radio signals are transmitted to a cell phone or other user terminal. We have to model those transmission patterns in the digital twin, so we can predict what will happen when some parameters in the base stations change. I’m responsible for developing that part.

When did you start working on this?
Right after joining Huawei, a little more than two years ago. Before that, I was in school, getting a Ph.D. in wireless communications at Tsinghua University [one of China’s top research universities].

Is SRCON a breakthrough technology?

Yes, I believe it is. SRCON builds on pre-existing research on wireless communications that has been going on since Claude Shannon published his work on information theory more than 70 years ago. But until now, not a lot of people have understood that the network-level problem was so important.

Wireless communications tend to focus on things like signal processing – basically, trying to make the wireless connection stable, so the signal doesn’t drop out, and to make wireless transmission work as close to its upper limit as possible. We had mostly focused on one user, connected to one base station. We hadn’t put so much effort into making the parts of an entire network, consisting of tens of thousands of base stations, work well together.

There are several reasons for this, but one reason was that it’s really difficult! You have to know how the base stations interact with each other and with the environment. That’s super-complicated, and there was no mature model for us to study. So getting base stations to work well together was Mission Impossible.

SRCON lets us design rules for how to optimize the performance of the whole network. That means not just improving user experience, but doing so with minimal energy consumption.

SRCON helps the environment?

Yeah, base stations consume a ton of energy. But actually, there are certain hours where lots of users are connected and need service, and other times when there aren’t many users at all. Keeping base stations active all the time is a waste of energy. It’s better to shut down the idle units. SRCON lets us evaluate the best hours to shut down some of the base stations.

Couldn’t you do that manually? Just turn off 80% of them in the middle of the night, when most people are asleep?

Well, how do you know which units to shut down? If you turn off the wrong 80%, some users won’t get good service – or any service at all.

And which hours would you shut them down for? Maybe for business, the idle hours are 1am to 5am, but for households, it’s actually 9pm to 5am. There’s no one rule.

SRCON does a better job figuring out how to minimize energy use – with little compromise on network performance.    

So SRCON is a big deal. Have other inventors come up with anything similar?
Everyone will claim they have something similar. But are their solutions as good as SRCON? I don’t think so. We ourselves had tried other solutions in the past, and they just weren’t good enough.

For example, right now people are deploying high-definition maps to see how radio signals propagate in different parts of a city or rural area – how the signals reflect, how they scatter.

The best result has generated known accuracy that’s much lower than that of SRCON.

What drew you to this line of work?

Since I was little, I’ve been fascinated by research, engineering, knowing how things work and how to make them work better. I’ve always been good at math and physics. For me, this was kind of a natural choice.

As a successful inventor, do you have a particular technique for “thinking outside the box”?

I just forget about the job for a while, and go talk to people – stop thinking and take a break.

When you go out and see other people, and other things, you’ll say, “Oh, here’s something new...” And then sometimes, if you’re lucky, you’ll find a connection between that new thing and the work. It doesn’t happen that often, but you only need it to happen once in a while to be creative.  

You’re one of the lead inventors of the SRCON patent portfolio. Can you give us a sense of what that means?

We have four patents in the SRCON portfolio. The basic patent is about radio signal propagation. That patent has four listed inventors. I’m the primary contributor, which means I was in charge of developing that patent, and I wrote the patent application.

But our team consists of people from lots of different backgrounds. Again, there are people like me, who have Ph.Ds., people know math and AI and the theoretical parts of algorithms. But there are also team members who are experienced field engineers. They’re the ones with the knowledge and experience to identify important problems that need be solved in the real world.

And it’s not just Huawei people. We’re working with academic researchers from outside the company. We’ve formed a joint lab – Huawei, plus the Chinese University of Hong Kong in Shenzhen, and the Shenzhen Research Institute of Big Data. We’ve got at least half a dozen professors and outside researchers working in this joint lab.

What’s the most exciting breakthrough possible – the Holy Grail for SRCON?

Right now, we’re all just thinking about how to make SRCON more powerful. The goal is not just to produce the best network performance, but also to reduce energy consumption. The technology is not perfect. We’re still working on making every little bit of it more accurate, more precise.

The cool thing is that we’re doing something that will benefit the entire telecom industry – and lots of other industries as well. As a matter of fact, fantasies about virtual reality and augmented reality in science fiction and futurism – including the metaverse – build on a high-speed network. For that reason, SRCON is expected to accelerate real-world creation of “extended reality”: the combination of VR and AR. It’s great to be working on a technology that has such a broadly beneficial impact on society.