FEATURE Smart factories and AI
Feature sponsored by
Automating your way to AI, and why manual approaches fail
Automating your way to AI, and why manual approaches fail
By Aleksi Helakari, Head of Technical Office, EMEA, Spirent T
he telecommunications sector wants to start using AI. Although there’s a will, for many the way is less clear.
Companies are contending with stubborn obstacles that prevent them from even starting to implement AI/ ML solutions. Namely, the complexity of such technical stacks, fragmented data collection and companies’ ingrained reliance on manual processes. The historical backload of technologies and manual practices is colliding head on with the need to innovate and survive in business.
AI starts with automation AI adoption may transform telecoms but doing so will need more than putting new technology on top of old – it requires a wholesale restructuring of some of the most fundamental practices and operations.
Data collection is perhaps the most fundamental. One of AI’s most critical needs is large amounts of data-rich, high-quality data that can be used to train models. For automation in telecommunications, for example, a crucial first step on the path to AI is automating data collection at the most basic levels. That’s impossible for the many whose data collection practices are still not only manual but siloed and entirely unintegratable between the various departments.
30 June 2024 | Automation
Siloed data collection Telecom operators can be highly siloed in multiple areas. Departments will generally have their own budgets, tools, practices and purposes. Each of their technology stacks will be highly complex and, as a result, have their own metrics that may not integrate. Similarly, those departments – The problem gets bigger when we take the example of larger companies with more complex structures operating in multiple countries with a new galaxy of departments and command structures for each region. Furthermore, some of the processes that these telecoms organisations use are still done manually. This is a problem on several levels. Manual data collection is organisation automate. This can create problems down the line if perpetuated. If, say, an operator wants to create a
a design and then send it to the testing lab. Manually testing that design will produce some data which will often then have to be integrated manually, opening up the process to various inaccuracies. That inaccurate data will most likely pass into the next stage of development, thus AI capabilities on top of these shaky
To make matters worse, manual testing can’t be done with nearly the speed or frequency required to test products and services expected of an operator in the near future. They’ll need to test, for example, 5G assets dozens or hundreds of times a day in order to accurately emulate real-life working conditions for the 5G technologies they’re developing. From this point of view, operators deal with a lot of redundancy. They use processes and technologies that have built up over time and, although once useful, stay in place because they do the job well enough. Yet, we’re hitting an impasse. Some operators are trapped within their own legacy technologies and outdated practices which stop them from innovating any further.
Where to start Operators are now hindered by their fragmented data and the complexity of their technical stacks that keeps data collection siloed and processes manual. New data collection processes will have to be put in place at the most fundamental levels of the organisation. That could mean ripping out older technology stacks and processes to harmonise the toolings that organisations work with. It will also mean fundamentally of the organisation share and collect that data. However it happens, automating data road to AI/ML innovation in telecoms.
automationmagazine.co.uk
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46
