Network automation: ‘Everything comes back to data quality’
A data-first approach to operational excellence
As operators consider any type of artificial intelligence- or machine learning-based network automation projects, “Everything comes back to data,” according to Damion Rose, senior product manager of mobile signaling and data analytics at BICS. “Using machine learning starts with data.”
Rose recently participated in an RCR Wireless News webinar during which he shared his perspective on network automation and taking a data-first approach to developing operational excellence.
In order for AI and ML systems to learn or demonstrate intelligence, those systems need access to data — and the higher quality the data, the higher quality the outcome. Rose said that effectively translating data collection into ML and AI efforts further rests on organizationally making investments in data engineers and scientists that can work with other internal stakeholders to turn data-based ML and AI systems into positive, customer-centric outcomes.
“A lot of what’s being said about AI and machine learning is very… future-oriented,” he pointed out. “It’s good sometimes to take that step back and look at what’s happening in the real world and how we can apply this. The first step is actually understanding that it has to be a data-first approach.”
In terms of making data science an asset to an operator’s overall business model, he said this is determined by asking whether data science capabilities — classification, pattern recognition and prediction — increase market relevance, increase differentiation and overall create a stronger value proposition. From there, operators need to focus efforts on high-value, customer-centric use cases that can accelerate sales cycles and increase contract acquisition. The next step is to do this quickly and at scale which would ideally result in higher margins and positioning as a value-creating partner.
“When we talk about automation there are several things that mobile operators in particular are hoping to achieve,” Rose said. “They range from increasing capex efficiency to generating ROI on the network resources that they have deployed, to reducing operational expenses which can be a significant part of overall business expenses. All of this complexity makes the job of being a mobile operator and driving efficiency through automation a complex challenge. But, there are a number of reasons why automation be- comes interesting,” he said, calling out crucial operational steps taken by service providers, including network design, load balancing, understanding network coverage, capacity optimization, and cyber security-related functions. I think everyone would understand and can see our way forward to automating these [use cases]by enabling software-defined networks to make decisions on a day-to-day basis. Any adventure into applying data science should start from the use cases and the data behind that use case. Everything comes back to data quality.”
Rose suggested service providers looking to leverage data science, AI and ML consider the following recommendations:
- Have a clear view of the target and define your level of accuracy
- Make data available as soon as possible
- Build a multidisciplinary team where data scientists, engineers and others can effectively collaborate
- Understand that machine learning techniques have prerequisites
- Remember that AI and ML projects are research-heavy, starting over may be necessary, failure isn’t forbidden, and scalability is key.