11 Times Faster, 9 Times More Cost-Effective: The Real Business Value of Automating Operations with TuplOS®

Introduction

The current era of digital transformation has forced telecom operators to re-examine their businesses and revise their strategies to deliver services in a smarter, cost-efficient way. 

When building an automation use case, multiple factors need to be considered as they will have a huge impact on the total cost and time of the project. Apart from the license costs of a given solution, the most important factors are the flexibility and scalability of the platform, the resources needed to develop and deploy a solution to production, and the time required to put the solution in the market.

In this article we will cover the business benefits of the TuplOS MLOps platform, based on real data of a delivery for a Tier 1 Telecom operator in Europe (in this case the network size was around 2,800 sites and 20,000 cells).

In this particular case, we are presenting the benefits of using Tupl’s Network Advisor framework, a solution based on the TuplOS platform, for the automation of existing use cases and the creation of new use cases, comparing it with both other approaches in the market and the legacy way of working with no automated processes.

Automated solution for existing use cases: Tupl automation vs No Automation

During the above mentioned work, three use cases were delivered, automating a total of 45,108 engineering hours a year, amounting to over 2,000,000 € a year in engineering time cost.

The three Use Cases delivered clearly show the benefits of Tupl Network Advisor versus using a pure legacy approach with no automation.

a.     First Use case: LTE User Throughput management.

LTE User Throughput is one of the key aspects that Telco operators have to manage to ensure a good customer experience, and they amount to a big number of technical issues or tickets created by Customer Care operations teams. In the real case we are discussing, Network Advisor was able to analyze 6191 in just two months, which manually would have taken 3096 hours. 

Assuming a specialized labor cost of 400.00 € / day, the total cost of all the issues resolution is 955,050.00€/year.

b.     Second use case: VoLTE Drop Cause management.

Voice over LTE is another of the key features that Telcos need to manage daily for a successful customer experience. Several issues can affect the normal performance or VoLTE, such as network congestion, anomalies in Inter-RAT mobility, or transport anomalies. Being able to automatically identify the root cause of such technical issues and implement a solution provides Telcos with the ability to respond to these issues much faster. In the particular example we are describing, Network Advisor was able to solve 6,562 tickets.

c.      Third use case: LTE Uplink Interference management.

In the case of LTE Uplink interference, it is key to determine what type of interference is affecting the network performance, whether it is narrow or wide-band, permanent or intermittent, and dependent or independent of traffic. The ability to run this assessment with ML automation is key to manage LTE uplink interference in a quick and accurate way. 

In this case, 1,756 tickets were analyzed automatically with TuplOS in just two months, which means 40 issues per day. Besides the cell-day problems root cause identification, tickets are created for a better management and control of the issues.

 All the above calculations do not include the time spent by engineers in aggregation and reporting, which is estimated to take 2 hours per day on engineering time. 

Beyond the automation of the root cause analysis, it is key to consider what we call the “Fix Implementation Delay", defined as the elapsed time between a network issue occurs and the moment a proper solution is implemented to fix it.

In the traditional manual approach, daily counters and KPIs are assessed every day to assess performance, and different top offender lists are calculated for each relevant KPI, showing the most degraded network elements (mostly radio cells) from the previous day. In this approach there is up to 33 hours delay (24h + 9h), since the top offenders are analyzed from 9am the following day.

In the Network advisor approach, the top offender lists updated every hour. This ensures that the degradation analysis includes cells that got recently degraded, minimizing the delay between the issue detection and the implementation of the suggested action.

Implementing a New Use Case: TuplOS-based Network Advisor vs. Third-party Software

The work developed for the Tier 1 Operator in Europe included developing a new use case for 4G UL Interference Management. We compared the process and costs to develop this new use case with Network Advisor versus acquiring a new third-party solution to address it. Network Advisor, based on the TuplOS MLOps platform, allows the flexibility to create new use cases as its approach is exactly the opposite to the traditional black-box approach where there is no flexibility by the user.

As a result of this process, we can affirm that: 

1. Provided Network Advisor is already deployed, developing additional new use cases with is 11 times faster in terms of time-to-market when compared to acquiring a new Third-Party solution. 

Time-to-market is defined as the time passed since the use case was needed and scoped until that solution was in production running live. This includes some market research and shortlisting of potential solutions, the internal procurement process, and the delivery time of the solution delivery project, which includes solution deployment, integration, and validation. 

On the other hand, if we develop a new use case based on Tupl Network Advisor already in place, these phases are reduced to scope definition, use case implementation (KPIs, features, ML model) and use case validation, which can bring the time required to 3 weeks instead of over 6 months.

This massive reduction of use case delivery is what makes Network Advisor a true enabler of Agile processes and organizations. 

2. Developing a new use case with Tupl Network advisor is 9 times less costly

In terms of costs and taking into account that acquiring a third-party solution includes all the operational expenses related to the internal efforts of acquiring the Use Case (procurement, project management during deployment, etc...), we can say that the overall cost of deploying a new use case with TuplOS Network Advisor instead of a third-party tool can be 9 times smaller, even without including the actual cost of the third-party license. 

From the resources’ standpoint, when the TUPL framework has already been developed and a new Use Case must be implemented, the only professional resources needed to develop the use case are data engineers and RF engineers. In the case of using a new third-party solution, additional resources are required, such as Procurement, Project Management and a Software architect, multiplying the cost and complexity of the project. 

3. Flexibility of adapting to requirements changes.

Lastly, a key advantage of developing a new use case with Network Advisor already implemented is the flexibility of the solution: changes can be done in a few weeks instead of several months. Network advisor is the opposite of a black box it is an open framework to develop and adapt solutions to the ongoing and evolving needs in an “Agile” way. Thus, any engineer, direct contractor or Tupl consultant can easily adapt an already ongoing solution to the new upcoming needs in just one or few weeks. 

Conclusion

Network Advisor enables automated network performance analysis and issue resolution, with an open architecture approach that allows any subject matter expert to develop use cases with the flexibility of adjusting to evolving needs and requirements changes. 

It is based on TuplOS, a MLOps platform that is based on a pragmatic approach to ML. It is conceived to facilitate the creation of complex automation utilities with a fast turn-around time, all the while delivering a quality product that is stable, scalable, and easy to maintain.

Network Advisor exemplifies the power of TuplOS  to develop ML applications in an agile and flexible way. It can deliver lower costs and the fastest time-to-market than any other tool in the market, because it is based on an open architecture that allows any SME to use their domain expertise to develop an application that adapts to requirements needs, unlike other ML platforms or as seen in this case, other network management operations tools.