Ericsson CTO Laxdal dishes on virtualization, network slicing and 600 MHz

Glenn Laxdal, Ericsson


with Glenn Laxdal, CTO and head of strategy, Ericsson North America

Glenn Laxdal joined Ericsson in 2012 from RIM, where he served as vice president of global product management. Before that, he held executive roles at Airvana and Nortel Networks. As head of strategy, marketing and chief technology officer for Ericsson's North American region, he is responsible for identifying Ericsson's long-term vision, defining the overall company strategy and driving business value for Ericsson's customers in North America. FierceWirelessTech editor Monica Alleven caught up with Laxdal at the CTIA Super Mobility 2015 conference to find out what he thinks about network slicing, the upcoming 600 MHz auction and 5G. The following Q&A has been edited for clarity and brevity.

FierceWirelessTech: Can you address how the networks are being virtualized?

Laxdal: The first step is operators virtualizing their network resources. So instead of optimizing an application like the mobile packet core, for example, onto a hardware platform to deliver the high capacities that an operator needs, we'll put a virtualization layer on top of an industry standard X86 hardware server, and then the application, the network function, will run on the virtual machine rather than running on the hardware itself. We introduced this abstraction layer, and the value of the abstraction layer sitting between the hardware and the application is that I can now emulate in the virtualization layer the hardware, which means that I can now scale out, with multiple X86 servers that all look identical to one another and run any network function or application on top of that.

These technologies that we call network functions virtualization and software defined networking emerged in the data center -- it's how Amazon scaled its data center. It's how Google scales their data centers. It's how Microsoft scales their data centers. So we simply take those same concepts that emerged in the data center and we're now introducing them into the network.

FWT: Are they different cultures?

Laxdal: It's different cultures but it's a culture that does exist in telecom, it just exists in the IT world. The telecom operator has a large IT function, information technology function where all their back office systems are run, their operational support systems are run, their billing systems are run out of this IT back office, and they have a CTO function that runs the network. The network and the IT functions are now sort of coming together. As we introduce these concepts of network functions virtualization and software defined networking into the core of the network, we make the network much more agile. We can set up and tear down applications much more quickly because the applications are now running not on a specific hardware platform but they're running on a collection of hardware servers.

The important thing is we make the network much more programmable. In the past when I wanted to set up a network link from the data center to an enterprise, I would have to program all of the boxes that sat between them … link between the data center and the enterprise, all the routers would have to be programmed to set that link up. Now, with network functions virtualization and SDN, we can make that entire link very programmable, so I can set up that link dynamically. We call that a programmable network, meaning I don't have to touch each one of these boxes. Through a single control mode through the network, which we call an SDN controller, I can control the network path from the data center to the enterprise or to the consumer or to the machine. If I can program that network path from a single place in the network, from a single control point, and the network becomes programmable, then the logical next step in that is what we call network slicing.

FWT: What exactly is network slicing?

Laxdal: What network slicing refers to is we have the ability in software to assign a set of networking resources across the network that are optimized for the application that's requesting them. I might have a very low bandwidth, low latency network resource that is assigned for a connected car application or for a connected wind turbine application where wind turbines will ultimately be connected to the network and we can optimize the performance of the wind turbine by reorienting the wind turbine to meet the wind conditions in real time -- just one very simple IoT application. I can set a path through the network that is optimized for that low latency, low bandwidth application. I can also set up a path through the network that is optimized for a very high bandwidth application like video. Because there's going to be thousands and thousands and thousands of different kinds of applications that are going to ask for connections to the network when we get into the Internet of Things, we need to be able to set up network bandwidths that are optimized for each of those applications and to do that, it requires a programmable network. In other words, the network resources can be invoked based on the need of the machine or the need of the application.

If I'm a wind turbine application, I ask for access to the network, the network will authenticate me. I've got the right security credentials, I've got the right authentication, then OK, it will allow me on the network. The network will assign a network slice that is optimized for that wind turbine application and you multiply that for the thousand-fold applications that are going to ask for access to the network and you get this concept for network slicing. So each different slice of the network will have a different set of characteristics associated with that.

The reason why we need these programmable networks is because as we get into the world of Internet of Things, it's not just going to be smartphones that are connected to the network. And let's face it, all smartphones are basically part of the same thing, they just require connectivity and bandwidth. But as you get into the multiple applications like connected cars, like connected drones, like connected farm equipment, connected devices of all kinds, you're going to need to have different network slices that are optimized for each one of those connection requests. We see network slicing as sort of being the first step down the path of migrating from 4G to 5G and from where we are today to the Internet of Things.
Another step within 4G is getting operators access to more spectrum, because some of these applications that are asking for resources from the network are quite demanding in terms of bandwidth and operators have a fixed amount of licensed spectrum. If we could give them access to a very small amount of unlicensed spectrum to be paired with the licensed spectrum at the instant and time when you need that bandwidth, then that would be a pretty significant benefit. It just demonstrates the capability for an operator that has a session running on a licensed carrier like 700 MHz and that session demands an instantaneous download of a movie or something that's very high bandwidth, then you can dip into a small portion of the unlicensed 5 GHz band, the one that's used broadly by Wi-Fi today, and we can accelerate the download of that session by pairing the unlicensed bandwidth with the licensed bandwidth. We can increase download speeds by a 150 mbps by just dipping into that unlicensed band when we need it.

FWT: What's the role of 5G in all of this?

Laxdal: We can introduce in 5G now some very important capabilities, like ultra-low latency over the air. We can reduce the latency across the interface by a factor of five, and low latency matters for applications like connected car and autonomous cars that would require ultra low latency -- 2 milliseconds. We can go from 10 milliseconds one way across the air interface to 2 milliseconds, so it's a factor of five reduction, and we can also enable ultra high bandwidth in 5G, that is by leveraging larger channel sizes and bundling those channel sizes together, we can deliver ultra high bandwidth. So ultra low latency, ultra high bandwidth and ultra low battery consumption are sort of the three goals of 5G. Our demonstrations show 5 gigabits per second throughput on 5G, so five times what Google Fiber is delivering with their fiber deployment to the home.

It's a combination of knitting together both the capabilities that are required for the Internet of Things, many many billions of connections all requiring very low power connection to the network and very low latency, all the way up to ultra high bandwidth across the wireless interface.

FWT: Have you had carriers come to you about what they can do with the 600 MHz spectrum that's going to be auctioned next year?

Laxdal: I think there is very broad-based interest in 600 MHz. I think operators realize how important low-band spectrum. Verizon and AT&T have a good amount of 700 MHz spectrum that's really helped them drive their LTE coverage model and now they can move to the densification stage on their LTE networks, so now operators that don't have as much low-band spectrum are really interested in how much low-band spectrum they're going to need, what the deployment model would be.

FWT: How quickly could you provide the equipment?

Laxdal: We'll have equipment for 600 very quickly. It's very easy for us today. We've got a 700 MHz radio, so introducing it in a 600 MHz radio is going to be very straightforward. The challenge is going to be clearing the spectrum. 600 MHz spectrum with broadcasters in it today will have to go through a clearing process, which typically could take a year and a half to two years for an operator to fully clear it. In the meantime, that doesn't mean they can't be deploying networks but they would need a clear network to be able to launch the network. So I'd say the long pole in 600 MHz will be clearing the 600 MHz spectrum and then getting the device on 600, and we will make sure the radios are available in good time.

FWT: What kind of position can the U.S. play in the development of 5G? How could things evolve in the U.S. to be a leader in 5G?

Laxdal: There is another spectrum that will be auctioned off after 600 MHz: That's 3.5 spectrum auction. So that's 150 megahertz of spectrum, and a big part of our ongoing discussions with CTIA and the FCC and other constituents is the need for more spectrum as we head into 5G. We see that it's very important for the North American operator base to jump into 5G. It's great news that Verizon has jumped into 5G and is willing to drive that in a very aggressive way, just like they do with 4G. Verizon really drove 4G, and I think North America benefited from the leadership that Verizon and AT&T showed for 4G, and similarly for 4G, DoCoMo in Japan was an early leader in a pre-standard version of 4G.

I think it's fair to say the ecosystem really rallied behind the standards-based approach that we took to LTE and that's really what Verizon and the North American operators can drive is sort of unifying the community behind a standards-based approach to 5G that captures both the requirements coming from both the Asian operators as well as the requirements coming from the North American operators and I think just by putting really aggressive timelines and goals out there as Verizon has done is a great thing for the industry, and we're going to be fully supportive in marching toward those really aggressive timelines with Verizon. I think it's huge, absolutely huge, and there's no question that Verizon can have a significant influence on the pace of the introduction of 5G.

Ericsson CTO Laxdal dishes on virtualization, network slicing and 600 MHz

Suggested Articles

Their collaboration seeks to simplify the steps and reduce the integration effort involved in creating virtualized RAN solutions.

T-Mobile is touting the world's first nationwide standalone (SA) 5G network.

The "cloudification" of the core network is perhaps the most important part as it essentially functions as the brain or control center of the network.