使用高级分析, AI, and Machine Learning to Mitigate Risk in the Modern Broadcast Environment

现代广播公司面临的挑战

虚拟化改变了媒体行业, 视频工作流每天都在以新的方式交叉, 允许敏捷, cost-effective business models and the ability for content owners to monetize content across more platforms. With vast amounts of data being transmitted across increasingly complex supply chains, 广播工作流变得更加难以管理, 而且问题更难诊断. As media organizations connect more on-prem and cloud resources together—equipment from different vendors, sources from other business units and partner organizations—and distribute to new device types, 测井和遥测数据的数量在不断增加. 大多数数据都被忽略了. 为什么? Broadcast engineers have more 信息 than they can effectively process, and they often silence alerts and alarms because they are just too frequent. 你不可能知道什么是重要的，什么是不重要的, 而这往往会让他们陷入黑暗, 只是抱着最好的希望.

Advanced analytics can help with these problems—making sense of the avalanche of data, 帮助人类操作员知道该关注哪里, and revealing where issues are likely to occur before failures are noticed. 紫溪 helps media companies take advantage of the unprecedented opportunity to leverage these toolsets for sophisticated event correlation, 数据聚合, 深度学习, 以及几乎无限的应用程序来改善广播工作流程.

紫溪智能数据平台防患于未然

的 紫溪 Software-Defined Video Platform (SDVP) delivers live broadcast-quality video at the lowest latency and the highest reliability over any IP network. With more than a decade of experience pioneering live video delivery using the 紫溪 protocol and the ZEN Master control plane, 紫溪 has unique insight into the dynamics of the network and how video data is transmitted.

紫溪智能数据平台(IDP) 这是多年来对数据架构投资的高潮吗, 先进的分析, 机器学习, AI, 和事件相关的工作，使用户能够监督, 管理, and deeply understand their inputs and outputs by providing ultimate visibility using learning algorithms to work with extremely large data sets and customize intelligence. 虽然人工智能和机器学习并不是新鲜的概念, they have been largely underutilized in the broadcast and media industries up to this point.

智能数据平台是如何工作的?

的 IDP consists of a data bus that aggregates data from all 紫溪 endpoints, 包括编码器, 云服务, 网络——250多个内置紫溪的设备和系统中的任何一个. This telemetry data is sent to 紫溪’s cloud 机器学习 platform and fed into continuously updated models where events are correlated, 发现了一些模式. 的 system uses historical data to form a model of normal behavior and generates an anomaly score. 的 IDP also uses a predictive model to determine the likelihood of unrecoverable dropped packets or stream failure within the next 2 hours—the 紫溪健康评分. Both the Anomaly and the Health Scores are communicated to ZEN Master where they are shown as indicators and graphs over time.

通过每天使用超过30亿个数据样本, 紫溪’s 机器学习 models are constantly being trained for accuracy and consistency, continuing to improve over time as more data is aggregated across the platform. SDVP提供世界上最高价值的直播, the IDP has unique access to network and video quality telemetry and can leverage these and additional strengths to produce proactive data insights in a variety of ways.

紫溪智能数据平台核心特点

紫溪健康评分

• 在2小时的窗口内预测降级的视频和丢失的数据包
• 离线状态预测
• Utilizes event correlation and a variety of data inputs to produce a tangible, numerical value
• Continuously improving ML, based on 100 correlated measurements per source, in real-time
• 快速解决问题的根源分析指标

异常检测

• 每个源检测到的异常随时间变化的连续图
• ML-based model learns normal behavior to identify deviations (Predictive Analytics)
• Allows engineers to pinpoint errors faster using historical and real-time data

历史图

• 查看随时间变化的运行状况评分、异常情况、统计信息等
  确定趋势

视频内容质量

• 沉默的检测
• 剪音频
• 空白的图片
• 静态图像
• 编码质量
• 音频水平

感知视频质量:

• ePSNR、eVMAF和eSSIM

智能警报

• 实时错误检测和通知
• 可根据您的用例定制警报阈值
• 基于ml的运行状况分数警报

基于ai的智能协议

• 紫溪.AI协议提供实时动态ML速率控制

IDP会带来什么好处?

IDP提供了直接和持久的提高能见度, 生产力, will reduce operational cost and save time – while ultimately improving the quality of service delivered to consumers. 有了IDP，广播工程师和媒体组织可以:

1. 消除数据过载，专注于重要的事情: 的 IDP’s intelligent alerts and health scores help sift through and aggregate data trends so that teams can focus where they need to without the distraction of false alarms.
2. 有预见性和前瞻性: 在系统出现错误后修复问题既昂贵又有压力. Predictive and proactive maintenance is conversely orders of magnitude less expensive and simple.
3. 降低成本，改善RCA: Engineers and broadcasters need to be able to identify the root causes of instability and failure faster and more easily.
4. 提高生产率:而不是花费时间和金钱从海量的数据中筛选, IDP能有效地识别问题.
5. 获得更大的洞察力和可见度: 反应性的或 break-fix approach suffers from lack of oversight and is subject to mass variability and volatility. 的 IDP keeps ahead of errors and issues before they become problems with continuous visibility.

借助高级分析的力量, 机器学习, 还有人工智能, the Intelligent Data Platform provides the transparency and control needed to guarantee reliable broadcast operation. To learn more about how the IDP can help perfect live video delivery over any IP network, 请访问我们的网站: 紫溪智能数据平台.

关于紫溪

紫溪提供基于云的本地软件定义视频 Platform that enables reliable broadcast-quality video delivery over any IP网络、任何协议、任何云提供商和任何边缘设备. 的 company offers technologies for broadcasters, enterprises, over-the-top 全球的视频提供商和移动服务提供商. 的 对于媒体公司来说，SDVP使其更容易、更经济地获取百家乐软件， 管理, localize, and distribute live events and 24/7 live linear channels in broadcast QoS, securely and at scale, using any form of IP network or Hybrid IP environments. 十多年来，紫溪使能网络 (ZEN)已发展到250多家OEM和服务提供商，并提供良好的服务 over 700 customers representing most of the top media brands around the world with 20,每天超过000个频道. www.紫溪.com