Anomaly Analytics: It’s Role in 5G and Assurance


According to the latest trends and reports published by industry forums, by mid-2021 5G adoption would have reached around 250 million active subscribers from this year’s 80 million accounting for an increase of almost 200%. Though this increase is substantial, in the larger scheme of things, 250 million subscribers only account for 3% of the total mobile subscription globally. Furthermore, revenue predictions for 5G also state that 5G technology will not boost revenues but give a telecommunication service provider far more flexibility in the business models that can be deployed within the larger eco-system.

This fact also means from an assurance perspective; the telecommunication service provider will need to invest in far more hardware and the associated services to onboard 5G services into the revenue assurance practice. While this may not be a significant investment initially considering the low adoption of 5G. However, this trend takes a significant turn as we go into the year 2022 to 2025, wherein the 5G service is expected to account for 30% adoption, thereby exponentially increasing your data volume, the hardware footprint and the OpEx required to perform daily revenue assurance activities for 5G services. At the end of the day with falling ARPUs, the discrepancies quantified will not justify the investment made towards onboarding 5G services into the traditional revenue assurance practice.

You will need to look for a newer approach that provides a better cost vs benefit.

Pressure on ROI from RA investments Traditionally revenue assurance involves the process of taking two or more datasets between the switch and billing and reconciling them to identify usage that was not billed due to issues in the data sent downstream. During 2G services, this volume for a tier 1 operator was anywhere between 100 million to 500 million xDRs, which required to be reconciled daily. However, with 3G and 4G service, this volume exponentially increased due to the larger mobile service adoptions and the associated services being provided. Today for a tier 1 operator with 100 million subscribers, the revenue assurance tool is required to reconcile anywhere from 5 billion to 10 billion xDRs daily. With ARPUs falling year on year, there is a significant pressure on the revenue assurance practice to justify the return on the investment made by the organization to plug leakages, a goal that will become significantly harder to reach.

5G will only carry forward this trend at an even faster pace.

Role of Anomaly Analytics

As an example, let us look at a hypothetical revenue assurance setup. For an operator with 100 million subscribers, you will have anywhere from 10 billion to 5 billion xDRs generated daily. This data is then either stored in your data warehouse or your big data lake. Here the revenue assurance practice will require data to be processed individually, loaded in the tool, reconciled daily, and retained for 30 days with the goal being to identify revenue leakages. This setup would require at least about 600 CPUs, 500 TB of storage, and 8 TB of RAM, the hardware easily being a million-plus dollars on its own, excluding the cost of license, deployment, support, and manpower.

Here we employ an Anomaly Analytics methodology, which significantly simplifies the TCO and the complexity of reaching the end goal. Taking the same hypothetical revenue assurance setup as above. The data stored in the data warehouse or big data lake can be aggregated accordingly to specific dimensions such as type of subscribers, cell site, rate plans, type of services, and measured against various metrics. In this hypothetical approach, only the aggregated data is processed using anomaly detection to monitor millions of KPIs, identify discrepancies quicker and correlate with other associated datasets. The approach here significantly uses lesser hardware and provides near real-time insights to only KPIs wherein the solution derives as a genuine discrepancy. More importantly, we achieve the same goal of identifying discrepancies.

Traditional Reconciliation vs Anomaly Analytics Approach

The point of concern between the traditional reconciliation and the anomaly analytics approach is that with the latter, the missing data identified by the prior is not readily available. In the anomaly analytics approach, one must dive into the data in the data warehouse or the data lake to identify the granular data. So we arrive at the question if it is worth investing is an anomaly analytics approach even if one does not have ready access to the granular data?

As mentioned earlier, as a trend,we see a drop in ARPU; regardless of the service, the value of a single unit of measure, in this case, xDR, is trending downwards. As of last year, the ARPU dropped by 2% in the developed market (America and Europe) and by 18% in the developing market (South East Asia). This translates to the fact that if one identifies say 10,000 xDRs missing at a value of $1000, this value drops by a percentage every year due to the growing service usage and the drop in prices. Ultimately the question of going with an anomaly analytics approach for revenue assurance reconciliation is a question of “cost vs benefit.”


With newer 5G services, we see  a significant shift in our approach to revenue assurance and especially reconciliations. The term “single source reconciliation” has taken birth due to anomaly detection allowing the business to analyze the data and obtain insights on service, quality, billing, and service orchestration from a single source of data processed over a short period. While we are in the early stages of this approach, we are certainly expecting anomaly analytics to take the initial steps within a revenue assurance function and grow to become the mainstay within a couple of years.

Want to know how Anomaly Analytics can play a role in 5G Assurance?

Schedule a demo!

5 2 votes
Article Rating
Notify of
Inline Feedbacks
View all comments
Get started with Subex
Request Demo Contact Us