{"id":32292,"date":"2019-10-31T21:46:11","date_gmt":"2019-10-31T18:46:11","guid":{"rendered":"https:\/\/prohoster.info\/blog\/kak-ustroena-oblachnaya-igrovaya-platforma-dlya-b2b-i-b2c-klientov-resheniya-dlya-otlichnoj-kartinki-i-borby-s-poslednej-milej\/"},"modified":"2019-10-31T21:46:11","modified_gmt":"2019-10-31T18:46:11","slug":"kak-ustroena-oblachnaya-igrovaya-platforma-dlya-b2b-i-b2c-klientov-resheniya-dlya-otlichnoj-kartinki-i-borby-s-poslednej-milej","status":"publish","type":"post","link":"https:\/\/prohoster.info\/en\/blog\/administrirovanie\/kak-ustroena-oblachnaya-igrovaya-platforma-dlya-b2b-i-b2c-klientov-resheniya-dlya-otlichnoj-kartinki-i-borby-s-poslednej-milej","title":{"rendered":"How the cloud gaming platform is structured for B2B and B2C clients. Solutions for excellent visuals and last-mile challenges.","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"

Cloud Gaming is referred to as<\/a><\/noindex> is one of the key technologies to watch right now. Over the next six years, this market is expected to grow tenfold \u2014 from $45 million in 2018 to $450 million by 2024. Tech giants have already jumped into the niche: Google and Nvidia have launched beta versions of their cloud gaming services, while Microsoft, EA, Ubisoft, Amazon, and Verizon are preparing to enter the scene. <\/p>\n

For gamers, this means that very soon they will no longer have to spend money on hardware upgrades and will be able to run powerful games on low-end computers. Is this beneficial for the other players in the ecosystem? We explain why cloud gaming will increase their earnings and how we developed a technology that allows easy entry into this promising market.<\/p>\n

\"How
\n
\n<\/a><\/noindex><\/p>\n

Publishers, developers, TV manufacturers, and communication operators: why do they all need cloud gaming?<\/h1>\n

\nGame publishers and developers are interested in delivering their product to the largest number of players as quickly as possible. Currently, according to our data, 70% of potential buyers never reach the game \u2014 they do not wait for the loading of the client and the installation file, which can be tens of gigabytes in size. At the same time, 60% of users, based on their graphics cards,<\/a><\/noindex>cannot run powerful (AAA-level) games on their computers at an acceptable quality. Cloud gaming can solve this problem \u2014 it will not only maintain the earnings of publishers and developers but also help them grow their paying audience.<\/p>\n

TV manufacturers and set-top box producers are also looking towards cloud gaming. In the age of smart homes and voice assistants, they are increasingly competing for user attention, and gaming functionality is the main way to attract that attention. With built-in cloud gaming, their customers will be able to run modern games directly on their television, paying the service fee to the manufacturer.<\/p>\n

\"How
\n
\nAnother potentially active participant in the ecosystem is telecom operators. Their way of increasing revenue is by providing additional services. Gaming is one of those services that operators are already actively implementing. Rostelecom launched a 'Gaming' plan, and Akado sells access to our Playkey service. This not only applies to broadband internet operators. Mobile operators, given the active spread of 5G, can also make cloud gaming an additional source of income. <\/p>\n

Despite the promising prospects, entering the market is not easy. Existing services, including products from tech giants, have yet to fully overcome the 'last mile' issue. This means that due to network deficiencies within homes or apartments, users do not have sufficient internet speeds for cloud gaming to function correctly.<\/p>\n

\"How
\nTake a look at how the WiFi signal weakens as it spreads from the router throughout the apartment.<\/i><\/p>\n

Players who have long been present in the market and possess powerful resources are gradually moving towards solving this problem. However, launching cloud games from scratch in 2019 requires spending a lot of money and time, and it may not even result in an effective solution. To help all ecosystem participants thrive in the rapidly growing market, we developed a technology that allows for a quick and cost-effective launch of their cloud gaming service.<\/p>\n

How we developed the technology that will allow for an easy launch of a cloud gaming service.<\/h2>\n

\nPlaykey began developing its cloud gaming technology back in 2012. The commercial launch took place in 2014, and by 2016, 2.5 million players had used the service at least once. Throughout the development, we saw interest not only from gamers but also from manufacturers of TV set-top boxes and telecom operators. With NetByNet and 'Er-Telecom,' we even launched several pilot projects. In 2018, we decided that our product potentially had a B2B future. <\/p>\n

Developing a customized cloud gaming integration for each company, as we did in pilot projects, is challenging. Each implementation took between three months to six months. Why? Each has different hardware and operating systems: some need cloud gaming for an Android console, while others require it as an iFrame in the web interface of a personal account for streaming on computers. Moreover, everyone has different designs, billing systems (a completely unique world!), and other specifics. It became clear that we either needed to increase the development team tenfold or create a highly versatile boxed B2B solution. <\/p>\n

In March 2019, we launched Remote Click<\/a><\/noindex>. This software can be installed on company servers to enable a functioning cloud gaming service. How will it appear to users? They will see a button on the familiar website that allows them to launch the game in the cloud. Once clicked, the game will start on the company\u2019s server, and the user will see a stream and be able to play remotely. This is how it might look in popular digital game distribution services.<\/p>\n

\"How
\n
\n\"How
\n<\/p>\n

An active fight for quality. And a passive one, too.<\/h1>\n

\nWe'll now explain how Remote Click tackles numerous technical barriers. The first wave of cloud gaming (for example, OnLive) was doomed by poor internet quality for users. Back then, in 2010, the average internet connection speed in the USA was<\/a><\/noindex> only 4.7 Mbps. By 2017, it had risen to 18.7 Mbps, and soon 5G will be widely available, ushering in a new era. However, despite the overall infrastructure being ready for cloud gaming, there remains the previously mentioned 'last mile' problem. <\/p>\n

One aspect we call objective: users indeed face network issues. For instance, the operator does not provide the promised maximum speed. Or they are using WiFi 2.4 GHz, which is interfered with by microwave ovens and wireless mice.<\/p>\n

The other side, which we call subjective: the user is often unaware of any network issues (they don\u2019t know what they don\u2019t know)! At best, they are confident that since the provider sells them a 100 Mbps plan, they have 100 Mbps internet. At worst, they have no idea what a router is and divide the internet into blue and colored. A real case from the helpdesk. <\/p>\n

\"How
\nBlue and colored internet.<\/i><\/p>\n

But both parts of the 'last mile' problem are solvable. At Remote Click, we use both active and passive mechanisms to address this. Below is a detailed account of how they overcome obstacles.<\/p>\n

Active mechanisms<\/h3>\n

\n1. Effective error-resilient coding of transmitted data aka redundancy (FEC \u2014 Forward Error Correction)
\n<\/b>
\nError-resilient coding is used when transmitting video data from the server to the client. It allows us to recover the original data in case of partial loss due to network issues. What makes our solution efficient? <\/p>\n

    \n
  1. Speed. <\/b>The coding and decoding processes are performed very quickly. Even on 'weak\u2019 computers, the operation takes no more than 1 ms for 0.5 MB of data. Thus, coding and decoding add almost no delay when gaming via the cloud. The importance of this cannot be overstated. \n<\/li>\n<\/ol>\n

    <\/p>\n

      \n
    1. Maximum data recovery potential.<\/b> Specifically, the ratio of the redundant data volume to the potentially recoverable volume. In our case, the ratio = 1. Suppose we need to transmit 1 MB of video. If we add 300 KB of additional data during coding (this is called redundancy), then during decoding, to recover the original megabyte, we will only need any 1 MB from the total of 1.3 MB sent by the server. In other words, we can lose 300 KB and still recover the original data. As can be seen, 300 \/ 300 = 1. This is the maximum possible efficiency.\n<\/li>\n
    2. Flexibility in adjusting the additional volume of data during coding. <\/b>We can set a separate level of redundancy for each video frame that needs to be transmitted over the network. For instance, upon detecting network issues, we can increase or decrease the level of redundancy. \u00a0\n<\/li>\n<\/ol>\n

      \r\n \"Play\r\n
      <\/div>\r\n <\/div><\/center>
      \nWe are playing Doom through Playkey on a Core i3, 4 GB RAM, MSI GeForce GTX 750.<\/i><\/p>\n

      2. Data re-sending
      \n<\/b>
      \nAn alternative way to combat losses is to re-request data. For example, if the server and user are located in Moscow, the transmission delay will not exceed 5 ms. With this delay, the client application can request and receive the lost part of the data without the user noticing. Our system determines when to apply redundancy and when to resend the data. <\/p>\n

      3. Individual Data Transfer Settings
      \n<\/b>
      \nTo choose the optimal way to combat losses, our algorithm analyzes the user's network connection and configures the data transfer system individually for each case. <\/p>\n

      It examines:<\/p>\n

        \n
      • the type of connection (Ethernet, WiFi, 3G, etc.);<\/li>\n
      • the frequency band used by WiFi \u2014 2.4 GHz or 5 GHz;\n<\/li>\n
      • the strength of the WiFi signal.\n<\/li>\n<\/ul>\n

        \nIf we rank connections by losses and delays, the most reliable is, of course, wired. Through Ethernet, losses are rare, and delays on the 'last mile' are extremely unlikely. Next comes WiFi 5 GHz, and only then WiFi 2.4 GHz. Mobile connections are generally poor; we are waiting for 5G.<\/p>\n

        \"How
        \n
        \nWhen using WiFi, the system automatically adjusts the user's adapter, switching it to the mode most suitable for cloud usage (such as disabling power saving).<\/p>\n

        4. Individual Encoding Settings
        \n<\/b>
        \nVideo streaming exists thanks to codecs \u2014 programs that compress and restore video data. In uncompressed form, one second of video can easily exceed a hundred megabytes, while the codec reduces this size by an order of magnitude. We utilize H264 and H265 codecs. <\/p>\n

        H264 is the most popular. All major graphics card manufacturers have supported it in hardware for over a decade. H265 is the audacious young successor. Its hardware support began about five years ago. Encoding and decoding in H265 requires more resources, but the quality of the compressed frame is noticeably higher than in H264. And without increasing the size! <\/p>\n

        \"How
        \n
        \nWhich codec to choose and what encoding settings to apply for a specific user based on their 'hardware'? It's a non-trivial task that we solve automatically. Our intelligent system analyzes the capabilities of the equipment, sets optimal encoder parameters, and selects the decoder on the client side.<\/p>\n

        5. Loss Compensation
        \n<\/b>
        \nWe didn't want to admit it, but even we are not perfect. Some data lost in the depths of the network is irrecoverable, and we can't send it in time. However, there is still a way out in this case. <\/p>\n

        For example, bitrate adjustment. Our algorithm constantly monitors the amount of data transmitted from the server to the client. It records every shortfall and even predicts possible future losses. Its task is to notice in time, and ideally to anticipate, when losses reach a critical level and start causing noticeable disruptions on the user's screen. It then adjusts the amount of data being transmitted (bitrate) at that moment. <\/p>\n

        \"How
        \n
        \nWe also employ error concealment of lost frames and a reference frame mechanism in the video stream. Both tools reduce the number of noticeable artifacts. This means that even with significant data transmission issues, the image on the screen remains acceptable, and the game is still playable. <\/p>\n

        6. Distributed Sending
        \n<\/b>
        \nTime-distributed data transmission also enhances streaming quality. The exact way to distribute depends on specific network metrics, such as loss, ping, and other factors. Our algorithm analyzes these and selects the optimal approach. Sometimes, distributing over a few milliseconds significantly reduces losses.<\/p>\n

        7. Reducing Latency
        \n<\/b>
        \nOne of the key characteristics when gaming through the cloud is latency. The lower it is, the more comfortable the gaming experience. Latency can be conditionally divided into two parts:<\/p>\n

          \n
        • network latency or data transmission delay;\n<\/li>\n<\/ul>\n

          <\/p>\n

            \n
          • system latency (taking control on the client side, capturing the image on the server, encoding the picture, the aforementioned data adaptation mechanisms for transmission, gathering data on the client, decoding the image, and rendering it).\n<\/li>\n<\/ul>\n

            \nNetwork latency depends on infrastructure and is challenging to combat. If a mouse chewed through the cable, no amount of tinkering will help. However, system latency can be reduced significantly, drastically improving the quality of cloud gaming for the player. In addition to the already mentioned error-resistant encoding and personalized settings, we use two more mechanisms. <\/p>\n

              \n
            1. Fast data acquisition from control devices (keyboard, mouse) on the client side. Even on weak computers, 1-2 ms is sufficient for this.\n<\/li>\n
            2. Rendering the system cursor on the client. The mouse pointer is processed not on the remote server, but in the Playkey client on the user's computer, meaning there are no delays. Yes, this doesn't affect actual control in the game, but the key aspect here is human perception. \u00a0\n<\/li>\n<\/ol>\n

              \r\n \"Play\r\n
              <\/div>\r\n <\/div><\/center>
              \nRendering the cursor without delay in Playkey using Apex Legends as an example. <\/i><\/p>\n

              With our technology, at a network latency of 0 ms and working with a video stream at 60 FPS, the entire system delay does not exceed 35 ms.<\/p>\n

              Passive mechanisms.<\/h3>\n

              \nFrom our experience, many users have a poor understanding of how their devices connect to the internet. Interviews with gamers revealed that some do not know what a router is. And that's okay! You don\u2019t need to understand how an internal combustion engine works to drive a car. It's unreasonable to expect a user to have sysadmin knowledge to play. <\/p>\n

              However, it's still important to convey some technical details so that players can eliminate barriers on their side. And we help them with that. <\/p>\n

              1. Indication of WiFi 5 GHz support.
              \n<\/b>
              \nAs we mentioned earlier, we see the Wi-Fi standard \u2014 5 GHz or 2.4 GHz. We also know if the user's device's network adapter supports the capability to operate at 5 GHz. If yes, we recommend using this frequency. We can't change the frequency ourselves for now, as we can't see the router's specifications. <\/p>\n

              2. Indication of WiFi signal strength.
              \n<\/b>
              \nSome users may have a weak WiFi signal, even if the internet is working well and seems to be at an acceptable speed. The problem will manifest during cloud gaming, which puts the network to the real test. <\/p>\n

              The intensity of the signal is affected by obstacles \u2014 for example, walls \u2014 and interference from other devices. Microwaves also emit quite a bit of interference. As a result, there are losses that may not be noticeable while browsing the internet, but are critical for cloud gaming. In such cases, we alert the user to the interference, suggest moving closer to the router, and disabling \"noisy\" devices. <\/p>\n

              3. Indication of traffic consumers.
              \n<\/b>
              \nEven if the network is fine, other applications may consume too much traffic. For example, if video on YouTube is running concurrently with cloud gaming or torrents are being downloaded. Our application identifies these intruders and warns the player about them.
              \n\"How
              \n<\/p>\n

              Fears from the Past \u2014 Debunking Myths about Cloud Gaming <\/h1>\n

              \nCloud gaming, as a fundamentally new way of consuming game content, has been trying to break into the market for almost a decade. Like any innovation, its history is a series of small victories and loud defeats. It's no surprise that over the years, cloud gaming has become surrounded by myths and prejudices. At the dawn of the technology's development, these concerns were justified, but today they are completely unfounded. <\/p>\n

              Myth 1. The picture in the cloud is worse than the original \u2014 as if you were playing on YouTube<\/h2>\n

              \nToday, in technically advanced cloud solutions, the images of the original and the cloud are practically identical \u2014 the differences are imperceptible to the naked eye. Custom encoder settings for the player's equipment and a comprehensive set of mechanisms to combat losses address this issue. On a quality network, there is neither motion blur nor graphical artifacts. We even take resolution into account. There\u2019s no point in streaming at 1080p if the player is using 720p.<\/p>\n

              Below are two Apex Legends videos from our channel. In one case, it is a gameplay recording on PC, and in the other \u2014 through Playkey. <\/p>\n

              Apex Legends on PC
              \n

              \r\n \"Play\r\n
              <\/div>\r\n <\/div><\/center>
              \nApex Legends on Playkey
              \n
              \r\n \"Play\r\n
              <\/div>\r\n <\/div><\/center>
              \n<\/i><\/p>\n

              Myth 2. Unstable Quality<\/h2>\n

              \nNetwork conditions are indeed variable, but this problem has been resolved. We dynamically adjust the encoder settings based on the user's network quality. A consistently acceptable level of FPS is maintained through special image capture techniques. <\/p>\n

              How does it work? The game has a 3D engine that constructs a 3D world. However, the user sees a flat image. For them to view it, a memory picture is created for each frame \u2014 a sort of photograph of how this 3D world looks from a specific point. This picture is stored in the video memory buffer in encoded form. We capture it from the video memory and pass it to the encoder, which then decodes it. And it\u2019s the same for each frame, one after another.<\/p>\n

              Our technology allows capturing and decoding images in a single stream, which enhances FPS. And if these processes are run in parallel (a quite popular solution in the cloud gaming market), the encoder will continuously access the capture, retrieve new frames with a delay, and consequently transmit them with a delay. <\/p>\n

              \r\n \"Play\r\n
              <\/div>\r\n <\/div><\/center>
              \nThe video at the top of the screen was obtained using the technology of capturing and decoding in a single stream. <\/p>\n

              Myth 3. Due to lag in controls, I will be at a disadvantage in multiplayer<\/h2>\n

              \nThe average control delay is a few milliseconds. Generally, it is imperceptible to the end user. However, there may be a tiny misalignment between mouse movement and cursor movement. This does not affect gameplay but creates a negative impression. The aforementioned cursor rendering directly on the user's device eliminates this drawback. Overall, the system's delay of 30-35 ms is so minimal that neither the player nor their match opponents notice anything. The outcome of the battle is determined solely by skills. Evidence below. <\/p>\n