Engineering-Wise - System Thinking
... and the need for advanced training.
The cable industry has learned that changes in one part of a complex telecommunications network can often have an impact on another part, and sometimes the entire system. As the industry continues to deploy numerous interconnected and integrated services, this relationship will intensify. Thus, the possibility is escalating that an engineer who is tasked with maintaining a single service on the network may inadvertently affect the performance of other services.
Examples of the need for system thinking in our cable infrastructures are easily discovered. For example, the long loop adaptive level control (ALC) issue outlined in three previous CED articles by Lamar West describes an end-to-end system adaptation loop using DOCSIS on the upstream, and that loop can be thwarted by field technicians introducing pads to solve an upstream RF level issue, or even just by adding a splitter in the headend to accommodate another CMTS blade. This topic in particular was included in the SCTE’s Live Learning event from the NCTA show last month as part of a renewed emphasis on upstream best practices and capacity management.
System thinking is at the forefront of challenges in deploying adaptive bit rate (ABR) technologies for IP video. The algorithms are designed to prevent video from freezing by adapting the bit rate of the IP video stream to the channel conditions, which vary in particular due to unpredictable wireless links in the access networks or the customer premise.
But when the wireless link conditions improve, the algorithms seek to increase the bit rate to enhance video quality and/or the overall quality of experience (QoE) for the customer. If large numbers of such streams suddenly increase their bit rates, one can imagine the impact on the overall network capacity just for IP video. Add to this the potential for an increase in ABR video traffic to happen when a sudden burst of interactive traffic is simultaneously added to several of those video streams, and one can readily see that the more flexible and variable the cable network is in each individual service, or even stream, the wider the range of global network traffic conditions that must be covered in engineering the overall network.
Dynamic adaptation in video, as opposed to ‘chunking’ – switching on previously encoded streams at different bit rates – also leads to an interesting phenomenon: As the ABR algorithms seek to drive the bit rate down, the CPU load on the video appliance that is performing the video processing or adaptation can reach a point where the CPU load is maxed out, and it can result in network errors in the output. Thus, the need to drive efficiency in video transport may bump up against resource limitations or, at a minimum, may drive up the power consumption of appliances in the headend, thereby raising energy costs of the facility. How many video engineers are currently concerning themselves with the energy costs of video processing equipment? Probably not many, but it is certainly another example of where system thinking – and even facility thinking – is increasingly needed in the cable industry.
The need for this type of “system thinking” has been expressed at an increasing frequency by both cable operators and content providers in cable, and the SCTE has responded with training courses that cover advanced RF and optical engineering of the access network, as well as an upcoming end-to-end comprehensive cable training series designed for electrical engineers and computer scientists that are new to the cable industry. The electrical engineers must learn the IP networking basics of the most current cable services, while the computer scientists must learn the fundamentals of RF systems in order to understand the access networks they are controlling via IP network management tools.
Engineers must also understand and be trained in how to characterize modern digital video artifacts, especially as more IP video gets deployed. How many video engineers realize that network artifacts (packet errors in video transport streams) can be subsequently re-encoded into the video via transcoding, grooming or statistical multiplexing devices, with the result that the packets look perfectly fine to a machine algorithm, but to a human, the video errors are just as pronounced? Or that some packet errors in video transport can easily be corrected by the decoder so that humans won’t even notice that an error was present? These are system concepts, in this case including the human in the loop, that must be considered if we’re to achieve the required QoE at cost-effective price points in operating our networks.
But system thinking can’t stop there. The industry must consider the entire plethora of services being delivered over the unified network, and further must consider cost of operations for delivery of those services. Energy costs must be brought to light so that if there are options for cable engineers to consider that have different energy requirements or flexibility, they can be weighed on that basis, as well as the more typical telecommunications functionalities.
System thinking is not just needed to ensure that our cable workforce is well trained; it’s critical to optimizing our networks and driving out cost so that the industry remains more competitive.