Crowdsourcing data for 5G network planning

What to consider when using crowdsourcing data for 5G network planning

David Nathan
Nov. 12 2020

With the increasing adoption of mobile services in recent years and big shifts to give more power to consumers and social media, crowdsourcing has grown in popularity.

The benefits of crowdsourcing to mobile network operators (MNOs) as part of their wider strategy are significant. It’s all about deriving actionable insights from vast pools of data, so that you can make better decisions.

The results are pretty straightforward, and the benefits are clear, but we’ll list them.

Four key benefits of using crowdsourcing data for 5G

  • Improve your efficiency;
  • Reduce your operating costs;
  • Improve user experience for your customers; and
  • Increase profits.

Sounds highly desirable, doesn’t it? But it takes some fine-tuning. To run a successful crowdsourcing operation, one that feeds into your wider network management in a way that will significantly improve your planning and implementation, there are many things you should consider.

This blog discusses the main factors.

But first, let’s start with a quick overview of what crowdsourcing actually is.

What is crowdsourcing?

Crowdsourcing is the practice of obtaining needed services, ideas, opinions or content by soliciting contributions from a large group of people, especially from the online community, rather than from traditional employees or suppliers.

Companies can also outsource and distribute labor to a large group of people, allowing them to select the best result not from a single provider but from a sea of talents performing micro-tasks.

Wherever crowdsourcing is used, the first benefit is to reduce costs, and thanks to the massive amount of data collected, it also increases visibility and possible options for better decision-making. Results can also be delivered much more quickly than when using traditional or conventional methods.

Let’s explore this further.

Crowdsourcing and mobile services

Within the realm of mobile services, crowdsourcing has been used with various objectives in mind, such as:

  • Improving popular services and apps used by mobile subscribers;
  • Providing operators with detailed insights on the quality its network delivers; and
  • Helping them to understand, manage and improve network quality as perceived by their subscribers.

By way of example, consider Waze, a mobile app that shows the value of leveraging dedicated crowd data. Waze uses crowdsourced information from mobile users to measure drivers’ speeds, to report traffic jams in real-time, predict traffic jams and automatically give directions for the best road to take.

For mobile networks, crowdsourcing methodology leverages a crowd of participants – the mobile subscribers – to gather network measurements, either manually or automatically through mobile apps, or directly from the network using call traces.

Collecting massive amounts of geolocated sample measurements gives valuable insights for network operators, and such data can be leveraged to improve some key operational activities, such as:

  • Network planning and deployment;
  • Network optimization; and
  • Network troubleshooting.

However, despite great advantages relating to costs and the ability to discover information that wouldn't have been revealed through a more traditional approach, crowdsourcing also results in limited control over the how the data is collected.

Low-quality results, which are always a risk, can impact negatively on decision-making processes that use this information.

So let’s review the challenges that MNOs face when considering crowdsourced data.

The main challenges of using crowdsourcing data for 5G

The main challenges of dealing with crowdsourced data for 5G boil down to three key questions:

1. Is the data valid?

Data is usually collected directly by mobile operators through their own apps, as well as by third parties via dedicated apps with embedded measurement components. Reported at regular periods, this data often goes through a number of post-processing steps to improve its validity.

If you don’t know which metrics have been collected, how they’ve been measured or how they’ve been processed, it’s easy to get the wrong impression of the network-quality picture, especially when using data coming from different sources.

So strive for transparency when sourcing this data.

2. Is the data reliable?

Unreliable data can still apply to some use cases, if associated with enough volume.

Given that it’s dependent on unknown end-devices, collecting network performance measurements from a crowd is one of the most challenging issue in using crowdsourced data.

Information on battery status, CPU load or the number of applications running simultaneously with these measurements might be unknown but these impact on measured network quality indicators. Also, distinguishing between indoor and outdoor calls isn’t straightforward from the perspective of longitude and latitude information, and could lead to wrongly interpretating the perceived network experience.

So tread carefully.

3. Is the data sample representative of the entire population?

Make sure you understand all of the biases that are introduced by the data-collection approach, such as:

  • Whether it’s passive versus active testing;
  • What the data collection trigger is, so whether it’s a background ‘agent’ or user action; and
  • What data collection service and devices are used.

With a comprehensive set of answers to these questions, you can weigh up whether and how to use this data to refine your planning.

Let’s look at network planning in more detail.

Crowdsourcing for network planning use cases

Crowdsourcing data for 5G can bring significant value during network planning activities, helping you to:

  • Understand your current network performance;
  • Highlight areas that needs investment to improve coverage and/or capacity; and
  • Plan for network developments, such as 5G, small cells or even new indoor deployments.

More specifically, if properly collected and interpreted, crowdsourced data can be used to get highly accurate traffic modeling to understand how network demand changes depending on location and time.

Using traffic maps as realistic as these, operators not only get a clear picture of their current network behavior, but they can also predict how it will perform in the future by using such traffic information within their radio prediction processes.

Most radio network design use cases rely on accurate modeling of traffic demand. To optimize network CAPEX, it’s important to get accurate traffic modeling in planning activities that take real traffic demand into account. And an accurate traffic picture allows:

  • Optimal site/cell placement;
  • Optimized antenna configuration and antenna model selection;
  • Massive MIMO deployment prioritization; and
  • Better 3D hotspots coverage, enabling lower interference and higher subscriber throughputs.

Crowdsourcing for automation: use cases and questions to ask

With MNOs moving towards automating elements of their planning processes, this is a subject for a blog post in its own right. But let’s round off with a quick look at some of the use cases that involve crowdsourcing.

Site prioritization

Ask yourself:

  • What is the network experience-related value of adding a site in a particular area?
  • What is the return of investment of adding sites in a particular area?

Key benefits of crowdsourcing:

  • Allows for site deployment prioritization to address network coverage and/or capacity issues detected from the crowdsourced data;
  • Provides an estimate of the network quality before and after site densification; and
  • Assesses the ROI of deploying new sites.

Capacity planning

Ask yourself:

  • When will the 4G layer run out of capacity?
  • What is the network quality value of introducing a 5G layer?
  • What is the network quality value of purchasing additional spectrum?

Key benefits of crowdsourcing:

  • Input on traffic density estimation and forecasts;
  • Offers an estimate of the network score before and after 5G deployment, based on simulated network KPIs; and
  • Estimates the congestion factor – cell loads, throughputs, packet losses and the like – both for peak hours and periods of lower demand

The service launch

Ask yourself:

  • What is the ROI of a new service deployment in any given area?
  • How easily can we deploy in-building solutions?

Key benefits of crowdsourcing:

  • Input into traffic density estimation and forecasts at minimum level of network quality; and
  • Identifies buildings in which service quality would be unacceptable without deploying in-building solutions

In conclusion

Crowdsourced data is becoming an increasingly vital part of the mix for network planning.

Of course, there are intrinsic limitations of this measurement methodology; it complements rather than replaces other means of deriving realistic estimates of network performance.

But as it matures, it’s showing great potential for improving network engineering activities and reducing operational costs. Add in the possibilities of automation, and it’s clear you shouldn’t miss out.

When looking for solutions to support your own network planning, make sure you choose one that not only handles crowdsourced data but can present its insights in a meaningful and digestible way for more effective planning.

Written By

Explore our coverage map to see where US companies are using Infovista planning and testing tools

View map