With the massive increase in OTT viewing, how will traditional broadcasting using RF transmissions continue? OTT has the benefits of being bi-directional and available to any hand-held device, but are we relying too much on the internet as a method of live streaming?
One of the challenges for OTT is that the internet was designed to be a transactional system. That is, a client requests data from a server, if it receives it all is fine, if not, then the client simply requests the information again. This method works well in an asynchronous environment but has some more limiting challenges for synchronous delivery as requests to resend data can cause undesirable latency.
Although the decisions regarding flying electron beam scanning made in the 1930s were correct for their time, it’s fair to say, that if we could redesign television and start again, we would not be using the serial transmission method we are all so familiar with. Consequently, broadcast television continues to be a synchronous delivery format and with OTT, we’re trying to make an asynchronous system work with a synchronous one.
Broadcasting is quite inefficient as it delivers all channels to all viewers simultaneously regardless of whether the viewer wants to watch the channel or not. In fact, with hundreds of channels now being broadcast to televisions through cable and RF, and viewers only being able to watch one channel at a time, we can see how inefficient traditional broadcasting really is.
In Kanban lean manufacturing terminology, OTT is a pull method and broadcasting is a push method. This means that OTT viewers only receive the programs they want to watch, or at least a small subset of streams. Consequently, the OTT system is much more efficient than traditional RF broadcast systems.
One of the key advantages that IP and COTS have brought to the broadcast community is it has allowed us to think more efficiently in terms of our system design. This allows us to build scalable systems that can respond to demand when required and not continually have to build systems that are designed for peak demand. Clearly, designing a system for continuous peak demand is incredibly inefficient as our expensive resource spends a great deal of time doing nothing.
However, dynamic systems are not easy to design and although I think they are the utopian solution to television, there are some anomalies regarding peak demand that are an inconvenient truth. For example, dynamic systems are based on statistical analysis of resource. If all viewers decide to watch a stream at the same time and my peak demand shoots through the roof, I may find that some people receive a poor quality of service until new resource can be added quickly.
Predicting peak demand is one area where machine learning can help enormously, and a great deal of research is going into this area.
Like everything in life, engineering is about finding a compromise and as intelligent OTT systems develop then the quality of experience for viewers will tend to the excellent service-levels we see from traditional RF broadcasting. I believe, as broadcast television still plays a critical role in our lives traditional RF broadcasting will continue to provide a valuable service. It’s not that this will go away, I just expect OTT to offer more.