Intelligent Automation for 5G Network Design

By Dr. Konstantinos Stavropoulos, Product Management

It used to be simpler. Yes, mobile network design – and the related activities of planning and optimization – used to be simpler, especially in the early days. But year upon year, various developments, including the advent of 5G, have made network design increasingly more complex.

Complexity in mobile networks has often been associated with their ‘multi’ nature. For years, networks have been multi-technology, multi-band, multi-layer, multi-domain, multi-vendor, etc. 5G has taken this ‘multi’ notion to another level. Multi-X networks now need to support a multitude of use cases and address the needs of multiple (demanding) users, based on the right mix of (yes!) multiple design options. And this is far from simple.

Increased complexity has exacerbated a number of network design-related challenges:

• Design duration – engineers need time to come up with good-quality network designs, but human resources are limited, and planning/optimization timelines are tighter than ever before
• Design variability – the designs produced by different engineers should be of similar good quality (and not require extensive reworking), but this depends on each engineer’s skills and experience
• Design applicability – network designs must consider real-world (including ROI) criteria, but most engineers are only familiar with engineering metrics (and mainly for RAN)
• Design fitness – design decisions should also anticipate potential future needs, but forecasting uncertainty and diversity of options make this an intricate and risky endeavor

Automation + Intelligence

In these terms, mobile network operators have been looking into ways to address complexity. The ability to partially or fully automate network design tasks, workflows, and processes has been of particular interest. Operators who have embraced network modeling and simulation as their network design foundation have also been interested in a key component required for automated decision making: an intelligent – network design – engine.

It is by adding such ‘intelligence’ to network modeling and simulation that operators can make the most of their digital network ‘twin’ (‘replica’) in the 5G era. When engineers use such a powerful basis for network design, they can produce better and more consistent designs, typically in less time compared with the ‘traditional’ manual-only approach. These designs can incorporate a diversity of real-world (business as well as engineering) criteria, to reduce the need for design reworking as well as the risk of suboptimal performance when designs are implemented in the live network. Engineers are also able to consider a variety of alternative (network-related) hypotheses, including low-likelihood but high-impact scenarios.

Intelligent automation – 5G case studies

Importantly, this intelligent automation approach has been proven in practice. Let’s have a quick look at how some leading operators worldwide have been using (Aircom’s) intelligent automation solutions to address specific challenges related to network design in the era of 5G.

1. Focus on speed

An innovative EMEA network operator emphasized the need for speed (and quality) when planning large network areas using an automated approach. Producing good-quality results as quickly as possible was essential for this operator to embrace intelligent automation in their 4G/5G network design.

2. Focus on consistency

For a leading network operator in EMEA, their new-site and updated-network configuration process had to consider multi-technology (2G/3G/4G/5G) cell constraints – including power limits – concurrently. The automated process has helped the operator minimize manual errors, which can be costly as they may delay network deployment and affect customer experience.

3. Focus on real-world relevance

Automated network design may be based on simulation and (mainly) engineering metrics, but it is also expected to make the most of an operator’s real-life network infrastructure. A Tier-1 network operator in N. America has used intelligent automation based on representative network modeling (including, for example, crowdsourced user data) to create 5G plans that address both engineering and business requirements when implemented.

4. Focus on alternative (What-If) scenarios

For a network operator group OpCo, a multi-year and multi-band rollout plan needed to consider various assumptions/metrics as part of a 5G license application. In this case, the ability to create and assess multiple year-dependent (gradual) network plans – with regards to different criteria, which included signal strength as well as capacity – was a key benefit of automated network design.

Intelligent automation – AI/ML

Would artificial intelligence (AI) and machine learning (ML) be relevant to 5G network design? Definitely. AI/ML is expected to drive learning and more dynamic solutions to help resolve complex and evolving network design-related challenges. Where is AI/ML most likely to benefit engineers? AI/ML-assisted input data use (e.g. for network model tuning/update), objective setting (e.g. target specification), and analysis/optimization methodology (e.g. to be scenario-specific) are only some indicative areas of interest highlighted by operators and considered in research/development projects.

Of course, ‘intelligence’ in network design is not an entirely new concept, and AI/ML should not become a ‘feature obsession’. In other words, AI/ML may not necessarily be a must-have or best-fit feature for every design related challenge. In addition, the ‘intelligence’ and ‘automation’ discussion goes beyond features, tools and solutions.

Intelligent automation – Mindset

There are technical obstacles to overcome for any organization that wants to introduce and fully embrace automation – especially intelligent automation. But in many cases, what seems to be the most significant hurdle is not solution related. It is much more about the organizational mindset. It is about understanding why the skeptics and critics may not yet be convinced. It is about putting in place the right pieces (including processes, and not just capabilities) to help teams make the most of ‘intelligence’ and ‘automation’.

Similar to other mobile network-related activities, intelligent automation has improved and will further enhance network design in the 5G era. However, without the right mindset, success will be unnecessarily compromised. And that would hardly be a smart way of using intelligent automation for 5G network design.

How Aircom Helps

Over the past 20 years, Aircom has delivered the best network planning tools for every major mobile technology. We understand what it takes.

Aircom’s ASSET delivers all the radio planning capabilities you need to design the best 5G network possible. 5G NR modeling with advanced propagation models, complex antenna arrays and full multi-technology 3D coverage and capacity simulations are all supported. Its extensive radio planning capabilities are designed to cope with the inevitable uncertainties surrounding new 5G technologies as they are developed.

ASSET isn’t just for 5G. It is one tool that does many things, including the simultaneous planning of wireless networks across GSM, UMTS, LTE, 5G and all other supported technologies.

You can learn more about ASSET’s capabilities across all technologies here, and be sure to check out the rest of Aircom’s three-part series on 5G network design: