 Yes, okay. Okay. So hello everyone. We would like to talk about the OpenStack solution which we use for a radio access network virtualization. My name is Robert Neumann. I'm a cloud engineer for OpenTelecom cloud and this is Professor Andreas Hartmann from the Hochschule for telecommunications in Leipzig about the motivation of this project. We expect for the diversity of the 5G use cases that requires a lot of flexibility that should be adaptable and a programmable architecture. Also the vendors like Deutsche Telekom, Nokia and Huawei work on concepts as well as practical implementations. And our colleague of Andreas, the Professor Michael Einhaus from Hochschule for telecommunication in Leipzig provides a research project for new key technologies in this context. The Hochschule for telecommunications in Leipzig and the Open Telecom cloud analyze OpenStack as an possible cloud architecture. About the OpenTelecom cloud, it's an OpenStack platform and which offers a wide range of services. For this project, we use some services like the compute research that's the elastic cloud server, the image management service to provide some images, the storage and network components for external IP addresses for the connection to the OTC and beside them the routing service and some management security services like firewall and encryption for the volumes. For future work, it is necessary to consider the use of the cloud container engine which is the Kubernetes service of this platform. And now Andreas will explain about the use and the project itself. Yeah, thank you, Robert. So as Robert said, I'm a researcher and my main topic is IT architecture so I'm pretty familiar with cloud architecture concepts. And when Professor Einhaus came to me and said, okay, there's a big change in 5G technology since they are moving from many hardware specific software which runs on the client side to centralized software, they need to change architecture and concepts. And I said, so maybe you're speaking about cloud architecture or don't we use something like OpenStack for this. So to understand what we do, this is just the plain architecture of LTE or radio access network. Important to know is that we have two parts. It's which is on the right hand side, the EPC, the old packet core taking care of the controls and the data exchange between the user environment and what we call antenna. So in the second part, this already access network or the so-called E&B, E-NodeB component. And to bring this to a new level in terms of cloud architecture, I propose three steps to Professor Einhaus. So the first step is we have to virtualize all the components. This has been done by introducing the open air interface. So with the open air interface, we were capable to bring all the difference of the components from the E-NodeBs beginning to the EPC core network on container technology. So in the first place, we did use Docker because we know it, but Docker can be replaced with any other technology that fits the requirements. So this was the first part and this solution still runs on a client PC in our environment. So the second step is to bring the containers into our cloud technology architecture. So next slide please. Thank you. The next step is we choose some of the components and then luckily, we had a cooperation with Dodger Telecom and we brought the containers to the open telecom cloud. And this actually works. So we have proof of concept. We did some measurements. And as you easily understand, there's a big bottleneck in the VPN tunnel. So the more clients you have, the higher the latency times gets. So and we speak of latency lower than one millisecond. So obviously, we have to do something more. But the first thing is all the components in the open air interface that now run on the open telecom cloud can be managed centralized. Just imagine in future we need some more capabilities in the LTE network. And we do this just by pushing a button when we boot some more EPC core network nodes. So we can increase the capabilities of the network. So there's one more step on that's but something we actually work on. We introduce an edge concept. So using the advantages of OpenStack architecture like centralized managing, like automated installation, automated managing, we include the client side, which we call the local end. We include the client side into the cloud concept. So now the BBU and the radio head hardware becomes part of the OpenStack architecture. Now the latency issue is over here. So between the thing we call an antenna and a BBU unit. So when we run the BBU physically at the client side, we can manage those latency times and those latency requirements. And we can still use the advantages of centralized automation and centralized management. Actually, we're still working on that issue. We have to create the local nodes where we use like Neutron and Nova. We have to do some more measurements to see if that works. I'm really looking forward to determine new technologies like Starling X, different container technologies. Unfortunately, Professor Anders couldn't manage to join us today. So he could explain more from the 5G component. So well, okay. The project is still in progress. Next things we do, as I said, we create the edge cloud solution. We install Neutron and Nova. So we have a cooperation with Huawei. We install the nodes right in our environment. We connect this to the OpenTelecom cloud so it all works together. And then we take care of security issues, as well as the connectivity between the EPC core network and the radio access network components. And we try to find how low can we get with the latency times. And just as I said, in future the overall goal or the objective is to automate as many things as we can. And together with Professor Einhaus we still trying to figure out how many software components that now work together can be split into different parts. So we can decide which parts of the software needs to be run physically as close as we can get to the antenna and which parts can be moved to the cloud like centralized software. So this is many much more work to do. And we are always happy about information exchanging experiences. So if you would like to know more about the project, don't hesitate to contact us or don't hesitate to ask questions. Thank you very much for listening. Thank you. Any questions? So it was pretty fast because we only had 10 minutes. Yes. Oh, sorry. Sorry. I can't understand you. No. Oh, yeah, yeah, yeah. So there were some approaches using hardware acceleration in terms of security. Like we can increase performance with hardware encrypting. So, but not more right now. Sorry. We are running all the components, all the EPC core network components. We run them in containers right now. We're using Docker. And the containers are being run in virtual machines on the OpenTail.com cloud. So it's no specific hardware. Is it complete? They have been completely virtualized. And we still do some performance measuring. So we can figure out what gives us better performance. A virtual machine or a container using Docker instead of other technologies. You're asking why we're using containers instead of virtual machines? So right in the moment, it's just for proof of concept. We just wanted to know if it runs. And we still are, as I said, we still try to figure out which technologies work best together. So right. Yeah, the containers are used, but it's a reason because only one container has one service inside. So the VM includes a stack of containers. And so you can manage all servers separately, not by a directly virtual machine. The virtual machine is something like a layer on top of the infrastructure of the OTC. And on top of these, we have the single containers to separate and manage all the infrastructure by the application to separate this. Yeah. So basically we're thinking results in two layers. So we have an infrastructure layer providing virtual machines. And on top of that, an application layer, which is being supported by containers. Thanks. Thank you. Okay.