 Good afternoon, welcome. Thank you for joining us for this session. I'm glad to be here with you today. For those of you who don't know me, let me introduce myself. I'm Woo Jin Han from Dreamace. Dreamace joined AGL at the Silver Level in 2017, and we are specialized in developing and customizing in 14-month software platform based on AGL and virtualization technologies. And we've been widening our perspectives, ranging from embedded technology to web platform. We've got some interesting topics to show you during session, and we are really excited to share some progress with you about the interactive copy system, as well as web platform. We've been working hard on this. We've got some information to share. If you have any questions, feel free to ask by asking the text chatting system during the speech. Our members, including chief technology innovation officer, Won Ni, are ready to answer your questions. So let's get started. Today we are going to start off by talking about motivations. Last year we shared our experiences about the AGL based container technology applied to mass production of real estate entertainment products. As seen on the screen, most of applications such as music player, home screen launcher, etc. were implemented by AGL. And Android applications such as navigation, game, streaming players were able to learn in virtualization environments by using Linux container. After doing this project, we've been researching and discussing with several clients. By doing so, we decided to widen our perspectives to interact with the system. As we all know, interactive copy system has many advantages ranging from holistic user experiences in the car to cost saving. And especially the number of screens in the vehicle is increasing and the size of screens is getting larger and larger. So it is very important to handle these screens effectively. We think that our container technology should be one of the good way to meet today's needs and interests. The second motivation of this topic is about ecosystem, which contains both Android and web. Our previous research and product has shown possibilities that Android ecosystem can be contained in the AGL system by using container technology. On the other hand, several players in the AGL has worked on web platform to make web applications run effectively in the AGL system. For example, Chromium to Wayland and web application manager, etc. We also believe that web applications can play an important role in extending ecosystem of AGL. We've been working closely with our partner, the web browser team of neighbor, which is one of the top internet service provider in Korea and Asia as well. We've been focusing on how to utilize existing web apps, how to make web developers not hesitate making web applications for vehicle and packaging applications, and how to build web application store in the AGL system. Now let me go to our approach and current status. The first thing I'd like to show is the concept of our system architecture. The blue colored part is standard Azure architecture, and the green colored part is the Android container by using this container. And yellow colored parts shows web platform parts, which is also isolated by this container. Under this concept, since both Android environments and web platform environments are isolated by containers, passengers will be able to use individual applications, regardless of open systems and screens that they use in the car. Furthermore, container has many advantages compared to hypervisor in the aspects of scalability, complexity, and cost saving. Online current activities regarding web platform in AGL, our approach is full browser based web platform to pursue utilize existing web applications. I mean both HTML5 web apps and principle web app contents. I will explain it in detail later. Currently, we are focusing on the infotainment parts. Although instrument-closed expert group has been working hard on cluster parts, the strict conditions such as real-time processing capability and high robustness level are required by car makers. That's why many players have been considering real-time OS such as QNX and integrity with hypervisor solutions. So we've decided that we are going to consider to use hypervisor to isolate cluster parts and infotainment parts until cluster related functions in the AGL will be stable and mature. In the meantime, we will raise degree of completion regarding multi-contain architecture as seen on this page. Now, let's take a look at demo. This is a copy system powered by container and web platform technologies. So 2C is to believe, right? It is the product of our advanced developments. So shall we start? In the past century, automotive has been evolving as a means to help people move quickly and safely. Interior parts also have been evolving rapidly together. Which enables people to receive driving lessons. Now you may see the potential inside. So let's examine this in a little more detail. We've developed this prototype by using Qualcomm's reference board. This board supports three kinds of Qualcomm's next generation automotive SOCs, and we've chose SA8155, which has enough resource and computing power to implement complicated interior copy system requirements. We took AGL 8.0, the Happy Hollywood, and Android QBSPs for implementing AGL and Android container system. Most of Azure applications were developed by using Qt, whereas Kenji UX framework is used in cluster application. Although our goal is to complete 2B architecture in the right side, the blue colored one, currently implemented architecture is shown in the left side of this page. In cluster, AVN and RSE is running in the Azure system and Android container runs in the separated environments. In other words, we've completed all other functions that we planned, but AGL and Android container has not been merged yet. In order to allocate each container to each display, we've been trying to configure multiple direct rendering manager. Unlike the other SOCs we've experienced, it was a bit complicated and there was trial and error, but we think the 2B architecture will be finished in the near future because we are now taking technical support from Qualcomm regarding this issue. As seen on demo, navigation screen of AVN display can be played in the cluster display. We think this kind of multi-screen use cases should be one of the main point of differentiation of the interior copy system. Cluster application consists of two layers. Layer 1 presents cluster functions and streaming server is located in the layer 2. In order to stream the Android navigation screen, we've developed streaming client and server by modifying motion JPEG protocol. We've also modified the surface freeing during Android media framework to implement streaming client function. So when the swipe event is detected, the streaming client starts to capture and send navigation screen to streaming server in the layer 2 of cluster. The specific area of layer 1 is blanked. We can see navigation screen of layer 2 through this blanked area of layer 1. From our experience, motion JPEG is more suitable for stream protocol than H.264. But we think these could be alternatives to do this job and we have a plan to research more to better perform on this end screen use case. The concurrent audio control is one of the important issue in the interior copy system. We think that audio output data from the interior container should be gathered in the control unit in the host OS and could be managed and sounded by the priority policy. This is because there are several sound that needs to be sounded first, while other sounds should be dwindled or muted. For example, when the FCA, the forward collision avoidance system, detects collision threats and warning sounds should be sounded. Other sounds such as music or navigation should be muted to make passenger recognize the urgent situation clearly. Our approach is to implement audio server and clients by using purse audio. To gather audio data from Android container, we've modified audio freeing of Android media framework. Like the end screen case, we think there could be alternatives to this job and we also have a plan to research more to better perform on this audio mixing use case. Now let's go to the web platform. Proof of concept development, we call it phase one development, is completed in the environment as seen on the screen. The Qualcomm SA8155 reference board and AGF 8.0. At this stage, our goal is to explore existing web application store can be applied to the embedded Azure system and to find out whether existing web apps can be installed and operated well or not. We found out that many web applications use APIs or functions of latest Chromium versions, which Chromium 68 included in the AGF does not support. The first tour of our partner can be performed well in the latest Chromium version. Plus, there are many progressive web apps, but Chromium 68 does not support PWA. So we decided that we replace Chromium 68 in AGL with Chromium 83 and fix several bugs in the Chromium 83 version, such as video failure and starting browser, etc. In addition, we found out that there are many web apps that needs keyboard input to use their full functions, and keyboard is also necessary to use web app store for searching web apps. By the way, current web platform of AGL does not support IME, the input method editor yet. So we will decide if virtual keyboard for both web app and Azure applications are needed. One more thing we want to share our experience is that we've brought the existing web app store to Azure system successfully, and web apps are packaged by widget format, which is the same way of packaging Azure applications. Both HTML5 web app and progressive web app are packaged in the form of WGT, the widget, and these widget files can be downloaded and executed well in our system. This means that existing web apps and web app stores are able to jump into the Azure ecosystem. After finishing the phase one development, Dreamace and Naver is decided to proceed more bold idea. This is what we call a browser driven web platform. The first motivation is to invite web developers into AGL. We've been developing browser to AGL interface to communicate with JavaScript library, which web developers are accustomed to. By doing so, there's no need to know about AGL for developing web app while Azure framework and its service layer is fully supported. The second motivation is to utilize web app store. Since our partner, the neighbor, has their own web app store and browser. So we think that it is a good way to deploy their experiences on browser and web app store. In the aspect of internal copy system, we assume that each passenger want to use each screen individually. So we've decided that browser driven web platform is isolated by loose containers, in which we have expertise. Here is a system architecture of our proposed web platform. As mentioned in the previous page, it is based on Azure framework and service layer. Dreamace is proceeding with Azure interface and vehicle info service while neighbor is in charge of the browser and web application store. It is also web developer friendly approach. Dreamace supports web application store that both Azure file apps and progressive web apps can be registered in form of a widget packaging. Let me tell you a little more in-depth about vehicle information service in the next page. We think that it will be great if web applications are able to use device information of car and control device functions. This could be a strong motivation for web developers to jump into automotive web ecosystem. Recently we've been engaged in W3C community and one of the members from the vehicle information service specification suggested us to review vehicle information APIs and its open source GitHub as written in this page. After reviewing this, we think that using this W3C vehicle information service specification helps us to save a large amount of time and resources while we are able to get stable and matured implementation. Now we are working on applying this vehicle information service specification to AGR service layer. This is the concept of web based home screen of our proposed web platform. It is fully browser powered HTML5 home screen launcher. It is under design now and will include many interesting features ranging from the floating widget to the push notifications, etc. Here's the quote, a dream you dream alone is only a dream, a dream you dream together is a reality. To build strong ecosystem, we've been collaborating with prominent players in the automotive industry. First, in order to make stable hardware with performance, we have been working closely with continental automotive Korea to make this all happen. In the aspect of web ecosystem, we've been able to work with real team of neighbor, and it's been our pleasure to work on this great technology. Also, we are going to be collaborating with automotive auto service provider in Korea. Wiper provides on-demand mobile car wash, Karam provides on-demand car maintenance, and Icha provides real-time wallet parking service. They have been actively providing services through the iOS and Android, but since this November, we've started to expand their service experiences to web ecosystem. We are going to be doing innovation together and we are looking for potential partners continuously. So if you are interested in this collaboration, please do not hesitate to contact us. Now let me go on to plans. First, the safety issue. As mentioned earlier, although the prototype is made of multi-containers is functional, it is highly recommended that cluster part and infotainment part should be isolated. So we've decided that we are going to consider to use RTOS and Hypervisor to isolate the cluster part and infotainment part until the cluster really functions in the AGR will be stable and mature. Second, reducing hardware dependency. We've been leaving various kinds of hardware platforms from different SOC manufacturers by the request of potential clients. When it comes to applying our container-based copy system to new platform, it was needed to keep integrating on the newer Azure versions and Android versions and fix and improve some components regarding different device drivers. So if we are able to find a way to reduce hardware dependence, this definitely helps us to save a large amount of time and resources. The third, dynamic resource allocation of multiple containers. Under limited hardware resources in the weaker, it could be a good way to manage resources of containers in different situations, such as concurrent usage of media or game in the internal copy system. We've been studying this topic by collaborating with universities in Korea. Finally, container-based upgrade. We think that container-based upgrade functions will enhance the advantage of using containers. To do this, we are going to be doing research about optimized file system to update each container of the internal copy system in the aspect of memory storage and update speed. For example, upgrade file system on this container could be a good way to achieve our goal. I just want to review that we've seen today. We introduced our advanced research regarding Azure-based copy system with web platform. Under this multi-container-based architecture, since both Android environments and web platform environments are isolated by containers, passengers will be able to use individual applications regardless of operating systems and screens that they use in the car. Unlike the current activities regarding web platform in Azure, our approach is full browser-based web platform to pursue utilizing existing web apps. We've ported web app store to Azure system and there's no need to fully understand how Azure works by supporting browser-to-Azure interface layer. Web developer friendly. We also found out that PWA web app can be installed and operated well in our system. And vehicle information service specification of W3C will be applied to Azure system to make web applications control automotive devices in the car. Finally, we've been working closely with our partners to build Azure ecosystem. We are always welcome to discuss future cooperation. Thank you for all coming our session today and I hope you are interested in our proposed internal copy system with web platform as much as we do. If you have any questions, please let us know. And here is our field of interest. Although our prototype was functional, but we know we have to go the extra mile to be ready for production. As we all know, there are many brilliant players are working on the Azure community. I hope we can have a cooperation opportunity in the near future. So that's all. Thank you again.