Spotlight on Tech

Fast transformation of performance monitoring, fault management, data management using any cloud (public or private)

By
Geoff Hollingworth
Chief Marketing Officer
Rakuten Symphony
January 11, 2023
4
minute read

Rakuten Mobile reimagined how a mobile network could be operated. Completely transformational outcomes followed, courtesy of a supporting software platform from Rakuten Symphony. Now, results of the partnership between Rakuten Mobile and Rakuten Symphony are available for any mobile operator in the world.

One point of entry gaining momentum for MNOs eager to take advantage of software-based outcomes is transformation of performance monitoring and fault management from legacy siloed solutions to new data lake powered operations. This powers new operational capabilities augmented by artificial intelligence and machine analytics. It also enables Root Cause Analysis (RCA) and recommended Next Best Actions (NBA) to be available in minutes rather than hours.

Performance Monitor, along with Fault Monitor are two of the components of the Symops solution provide by Rakuten Symphony. While Performance Monitor provides users with almost real-time insights around the health and performance of the network based on collection of raw data, the Fault Monitor application helps telecom operators identify the root cause of network issues by simplifying the process of network alert management. Read more about some of the revolutionary apps under the Symops solution here.

Rakuten Symphony’s software is cloud-native and can run on any cloud infrastructure, public or private. A common request we get is to trial whether outcomes similar to those achieved by Rakuten Mobile are truly possible in existing brownfield networks based on relevant network operator data, insights and KPIs. We can implement these trials on any regional cloud presence that meets regulatory, security and privacy needs.

Following is a template of a cloud native engagement, installing our applications and supporting Symworld platform on a Google Cloud Platform (GCP) instance. We can achieve the below within a four-week period.

Confirm The Use Cases

We begin by determining which insights were most important to discover and which KPIs should be surfaced in the resulting performance monitoring dashboard. During this process, we share all of our existing use cases and KPIs, which can be used without any modification. Typically, the recommended trial scope encompasses validation of less than 70 use cases.

Data Sharing

After determining use cases to validate, we move on to understanding what operator data can be shared and where it should be shared (e.g., specific cloud provider or region, etc.).  Typically, the volume of data shared is somewhere between 50-100 GB of text and CSV files. This volume of data represents less than one day’s operations data. The size of data represents the size of the network involved and the number of hours and days being monitored. For trial, the fastest way to transfer the data is via secure FTP and sharing sites, which also limits integration complexity. Based on the trial results, decisions based on insights generated can integrate into streaming data for real-time analysis and decisions regarding hosting cloud and location. The Symworld platform abstraction ensures cloud portability across any cloud provider. It is important to validate the data does not contain any Personally Identifiable Information (PII) and the dataset is approved to share from a legal perspective.

Data Ingestion

All operational datasets are different, depending on the vendor landscape and products generating and storing the source data. We write parsers to ingest the bespoke data, ensure quality and completeness, and store it in our normalized data storage format. This removes the legacy dataset silos and allows cross domain analytics and machine learning. As data management becomes more sophisticated, these datasets can be combined with real-time public datasets to create a 360-degree view of network experience, combining customer experience (e.g., sentiment analysis of social media) with current understanding of network fault monitoring and performance monitoring.

Data Analysis

Machine learning models are re-used, modified or created to deliver the desired results for the use cases and KPI measurement and presentation. This is where the real value of the data can be identified, realized and released.

Dashboard Presentation

Presentation of the data results to different audiences is key to realizing the true value of insights generated from the data. Network operations need to understand RCA and Next Best Action(s) as quickly as possible to resolve any developing outage issues. As data understanding increases, pre-emptive resolution can be possible. As confidence increases, full autonomy resolution can be realized. At this stage, it is important that there is complete trust between the machine learning, AI operation and the operational staff that are held accountable for all outcomes. Customer care officials can understand if customer complaints are related to known outages or existing problems. If no problems exist, they can signal that potential problems in experience exist at the micro level, but are currently not being seen at the macro network level.

Summary

It is possible to rapidly test Rakuten Mobile learning models using an operator’s own data and preferred public or private cloud. We recommend starting with fault monitoring and performance monitoring datasets, which can be executed across AWS, GCP or Azure.

Within four weeks, we can ingest the data, deliver the wanted use cases and share existing Rakuten Mobile use cases and KPI presentations.

Based on the trial results, shared learnings can be initiated. There are three dimensions of discussion:

  1. Fact-based discussions on cloud implementations
  2. Additional discussions on additional use cases
  3. Inclusion of additional datasets to increase the power of the insights and possible operational improvements.

The journey of Rakuten Mobile has been one of combining datasets from all telecom business dimensions. For example, the addition of maintenance work scheduling datasets can also be included in real-time analysis of customer support. If a customer has an issue, the agent can immediately be informed of potential maintenance side effects and the customer can be managed appropriately. The datasets can come from network data, customer data and social network data.

For more information on the total Symworld platform, see here.

Subscribe to our Community newsletter for regular updates.

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
Symworld™
Future of Telecom
Software Centric
How can Symphony help?
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
By clicking “Accept All Cookies”, you agree to the storing of cookies on your device to enhance site navigation, analyze site usage, and assist in our marketing efforts. View our Privacy Notice for more information.