流媒体服务“新常态”的基本要素
分析. 带宽. 压缩. 这三个词 这对流媒体服务的成功至关重要, 在当今世界, 它们从未如此重要. Around the globe, spikes in streaming TV viewing continue to be exponential. 事实上，87%的美国人.S. 消费者表示，他们在COVID-19大流行期间消费了更多的内容, 根据全球网络指数报告. 然而，随着居家限制不断放松， 流媒体电视收视率仍高于大流行前的水平. 观众对内容的持续需求, the time is right for video service providers to leverage the ABCs (analytics, 带宽和压缩)，而不仅仅是短期需求, 但从长远来看, 持续增长.
值得庆幸的是, both technology 和 expertise are now more available than ever before to help address such hurdles, particularly related to visibility 和 control capabilities that help video service providers more efficiently 和 effectively serve their subscribers. 此外, 端到端的专业知识可以无缝地支持分析, 带宽, 压缩需要帮助维持和扩大用户基础.
带宽
当我们考虑到我们在3月份看到的近乎一夜的观看高峰时, 我们可能首先想到的是带宽需求. Such concerns are not new for our industry (think of 生活 sports viewing over streaming, 例如), 但自2019冠状病毒病以来，它们已经进化甚至成长. The p和emic has brought to light questions that we've been talking about for some time, 但可能还没有完全回答, 例如:
- 我们怎样才能适当地扩大规模?
- 先发制人的规模需要考虑什么? 地理位置? 预定的事件? 内容的受欢迎程度?
- How can service providers leverage artificial intelligence (AI) 和 analytics to help scale in advance?
- How can we alleviate the pressure of more dem和ing video applications on the networks?
Closely tied to 带宽 is the amount of bitrate required to de生活r the high video quality viewers have come to expect. 在COVID-19大流行的早期, 带宽 was such a concern that many providers quickly reduced video service bitrates by percentages in the double digits. 在 加拿大 和 欧洲, 例如, Netflix reduced network traffic by 25% in an effort to sustain good quality in late March. 5月初，有 报告 比特率实际上还不到标准的50%. It's been reported that these network reductions have caused noticeable lowered image quality, 比如模糊和像素化.
Netflix并不孤单——许多视频服务都是如此, 如果不是全部, are still trying to find the most effective route to address the 带宽 challenge. Negatively impacting quality of experience (QoE) for viewers is simply not an option—they'll just go to another service that promises better video quality. 幸运的是, tools such as artificial intelligence (AI) 和 machine learning (ML) are making strong inroads 和 seeing positive results, particularly when it comes to compression techniques for maintaining the best quality at the lowest bitrate.
压缩
Next-generation video specifications can make video compression more efficient, 但也需要大量的编码器复杂性. 通过以一种聪明的方式使用ML等工具, we can find a balance between 带宽-efficient video streams 和 encoder complexity. AI 和 ML allow us to not only compress whole or partial frames individually, but also determine the areas within each frame that will matter least to viewers 和 compress them more than the others. 通过利用AI和ML作为编解码器优化的工具, matching the high sensitivity to image quality 和 abrupt changes that are detected by the human eye becomes more intuitive 和 helps to significantly improve adaptive bitrate encoding at the 生活 program or event level.
Content aware encoding (CAE) by scene can further help reduce the need for more 带宽, 再加上更智能的质量指标, 使我们更接近恒定的感知质量. Moving so quickly across so many geographies 和 at such scale leads to an explosion in the computational requirements of encoders. This means that the need for faster innovations 和 more efficient codecs have become immediate. Streaming services looking to exp和 the geographies they serve can benefit greatly by tapping into the capabilities of CAE 和 working with a technology partner who underst和s the ins 和 outs of the various codecs, implementation approaches 和 scalability requirements to ensure high QoE while the expansion is underway.
When considering QoE, it's important to think of content de生活ry networks (CDNs). 是否支持视频点播(VOD), 生活, 或电视内容, cdn使缓存过程更接近查看者, 这最终有助于改善观众体验, 即使是跨设备. 对于视频服务提供商也是如此, cdn提供了几个操作优势, as they de生活r more intelligent traffic routing 和 load-balancing 和 offer better isolation 和 faster diagnosis of network issues.
分析
cdn捕获发送给最终用户的每段视频的日志, 通常每个流每隔几秒钟. 强大的分析系统, 以及CDN的分布式特性, 允许服务提供者快速识别模式异常, 或者隔离产生错误或拥塞的区域. 可以确定问题是否由独特的设备类型引起, 设备故障, 地理区域, 一种特殊类型的流量, 甚至单个用户的特定ABR格式或内容标题. By rapidly processing analytics 和 generating appropriate alerts the provider can address issues quickly before they worsen—or last for a significant duration of a time-sensitive event.
此外, 同时在服务器上进行高级CDN分析, network 和 application level can facilitate efficient 和 stable de生活ry, video habit insights derived from AI 和 ML can assist with customer retention by providing valuable insights at the subscriber level—such as predicting future actions based on past ones.
思科可视化网络指数发现 到2022年，IP视频流量将占所有视频流量的82%. 更甚于以往, finding the right technologies 和 expertise to help avoid network congestion must be a priority for video service providers. The uniqueness of consumption by viewer/household ensures that a "one size fits all" solution will not work when consumers dem和 a premium QoE. 分析用户在寻找什么, tackling the 带宽 challenge without sacrificing quality 和 leveraging advanced technologies to improve compression efficiency are all key components to the equation that can result in both customer retention 和 acquisition.
为意外做好计划从来都不容易, when video service providers are assured they have the right technology 和 expertise to quickly 和 accurately de生活r more control 和 visibility, 他们的订户是最终的赢家.
相关文章
Verizon的达伦·莱普克, fuboTV的Geir Magnusson报道, 和 Reality Software's Nadine Krefetz discuss the present 和 future of OTT in this clip from 流媒体 East Connect 2020.
Aug 31 2020
SVOD占收入的最大份额——120亿美元，占58%
8月10日2020