Cover Story |
the need to monitor them has become even
more crucial . To run multiple complex
workflows — and manage multiple formats
and variants — fast and reliable diagnostics ,
analytics , and quality monitoring at each stage
of processing is essential .
Plume : ML and AI play an important role
in T & M . Our research shows that QoE scores
have remained consistently high in smart
homes leveraging AI-powered adaptive WiFi
systems , even increasing by 1.2 % from the
second half of 2021 to the second half of 2022 .
This makes clear how these technologies
support service providers to monitor and
optimise the home network and help them
deliver a first-class customer experience . ML
and AI , moreover , can provide insights down
to the application category level , empowering
service providers to take proactive steps to set
priorities for their installed base .
Torque Network Systems : I think the jury
“ Striking a balance between managing costs and delivering top-notch services is the central challenge .” - WorldCast Systems
is still out on where ML and AI fit into T & M .
The telecom industry has been working on AIbased
network monitoring tools and help desk
assistants for years . In those cases , the network
monitoring tools can make a difference given
the massive number of devices , network points
and transmission links . However , broadcast
networks are in order of magnitude more
simple . So , will AI make sense there ? I ’ m not
so sure . One possibility is for the analysis
of large quantities of end user data to distil
viewing trends and preferences . Those kind
of big data ( and Big Brother ?) insights can , in
turn , drive recommendation engines as well
as which ads are displayed to which viewers .
Certainly one of the more creepy applications
of AI .
VeEX : Machine learning and artificial
intelligence can certainly play a key role in
proactive maintenance / troubleshooting
( identifying issues before they become
problems ), test results analysis , interpretation ,
and diagnostics ( e . g ., fibre optics , RF ,
advanced coherent optical modulations , etc .).
However , the key barrier can be the lack of
real-life training data , since the cause and
effects ( data interpretation , diagnostics , and
“ Providers should prioritise assessing the quality that viewers receive on real devices .” – Witbe
|
suggested fix ) are not necessarily documented and filed with the raw test results . Even if historical correlated data exists , in whatever format , service providers consider them confidential . A change needs to be introduced to centralised test data repositories , to encourage end users to document the diagnostics and final outcomes along with the original test results , as well as making them available for AI training purposes . Today , the only thing T & M vendors can do is train their intelligent systems with labgenerated samples , limiting the efficacy of the solution . Hopefully , we can influence the required changes going forward . In the end , it is becoming even more important to assist the test equipment operators , who we can no longer assume are technicians , experienced , trained or knowledgeable of the technology they are hired to test . T & M solutions must become smarter and easier to use . Witbe : We ’ ve seen the role that automation plays in testing and monitoring , powering through demanding tasks that would be prohibitively time-consuming for human testers . AI and ML will likely play a similar role , taking over cumbersome and repetitive functions and freeing up team members to focus on tasks requiring meticulous attention to detail . In fact , AI may revolutionise automated testing in the same way that automation revolutionised manual testing , significantly streamlining the development process . With global teams being expected to deliver flawless performance across dozens of different devices — each requiring individual attention to maintain service consistency — this increase in automation would be especially valuable .
What are the challenges for ensuring QoE for budget-conscious service providers / subscribers with content being consumed over a range of
|
networks and devices ?
Bridge Technologies : Who isn ’ t budget conscious these days ? The core part of the question is ‘ over a range of networks and devices ’. That adds so many layers of complexity and places for things to go wrong . Each one needs a very specific approach to how it is understood and monitored – which traditionally will have meant expensive equipment tailored to each network / delivery type . But more important than the fact that it was costly was the fact that it was chaotic ; requiring the engineer to learn a range of systems , keep eyes on a whole host of disparate interfaces , and make space for a host of individual screens .
Avoiding this chaos has always been at the core of the Bridge mission – again , returning to the idea of making the complex simple . Our VB330 facilitates monitoring of pretty much every signal format across IP multicast , video OTT / ABR streaming , video-on-demand unicast , Ethernet trunk micro bursts , PCAP recording , L2TP unpacking and monitoring and general traffic protocol inspection . It therefore significantly condenses the amount of expensive monitoring equipment a broadcaster needs , by putting it all in one tool .
Moreover , it incorporates a huge number of QoE metrics to ensure seamless delivery to the end user including MOS score , ETSI TR 101 290 with Gold TS reference , IP jitter , packet loss , OTT profile alignment , SRT and closed caption verifications . Most importantly , it allows for an engineer to monitor this from a single interface – accessible from one screen , anywhere in the world – and indeed , to join up other additional platformspecific probes , such as the VB252 for digital terrestrial transmissions and the VB272 for satellite – combining them all through our VBC and displaying them at-a-glance through our Remote Data Wall . In this way , it puts all
|