Revolutionising NOC Operations for the Telecom Industry

About the customer

The client is a market leader in telecommunications organizations. They provide an innovative range of digital and data services. They operate in and around 12 countries and 10,000+ employee size with a mission to transform technology by connecting humans rather than isolate them. The strategy that drives their business growth is based on these pillars: Economic and social development through digitization, working towards sutainable initiatives.  


The realm of digital has become essential to our daily lives. As the IT industry advances at a breathtaking pace, the need to revolutionise networking as it were is at its peak. We transcended beyond clunky, traditional networks to an age of customer focused solutions, and Network Operation Centres (NOC) have been at the centre of this transformation. 
To put it simply, the NOC is the central nervous system of the mobile network infrastructure. It is not only the first line of defense against potential threats, but also ensures seamless data inflow and outflow. However, the idea that this is all driven by machines is a misconception, for if the NOC is the brain of the entire operation, then NOC operators are its heart. 
“The pace of evolution in IT, their experience of working in environments with an absence of fully functioning network capabilities”

Business Need  

Simplifying NOC Operations and network states through automated, deep insight driven AI-ML systems 
Network Operation Centres are hubs of bustling activity when it comes to monitoring transactions, traffic, and user patterns. The NOC operator acts as a liaison between the customer and the network infrastructure to solve any issues related to the above. They are an invaluable resource in the technology puzzle, but are also amongst the most overburdened, owing to inefficient process management, lack of agile working environments and pre existing redundancies.
The traditional NOC system involves multiple dashboard screens parallelly in use, with each individual request catered separately. This involves a massive investment of time and effort for the operator, often coupled with slow delivery and issue resolution.
Challenges Faced
Handling large volumes of data and integrating them together :
Ensuring its accuracy and quality
Securing data through privacy regulations:
Striking a balance between data accessibility and security
Upgrading legacy systems to modern data analytics platform:
Seamless data flow and interoperability between legacy and modern systems can be complex
Low latency in data analysis and dashboard updates:
Real-time Data processing to promptly respond to network issues and outages 


Multitasking like a Pro 

NOC engineers may find themselves juggling hundreds of requests at the same time. These ticketing requests are often unrelated to each other, and would need immediate resolution. Faults may range from server issues, software bugs or physical defects at the site of the network tower. Thus, the requirement for an automated, agile and intuitive NOC system is increasingly apparent in a digital first era.

The key requirements of a modernized NOC setup include :  

Auto suggestions for ticket key metrics

For reduced human error and analysis of assigned values 

Automated alarm correlation and RCA

For faster alarm detection

Digitally managed KPIs and inventory insights

For a holistic view of network corresponding alarms 
For reduced human error and analysis of assigned values 
For faster alarm detection
For a holistic view of network corresponding alarms 
NOC operators working in a centralized location struggle to keep track of time, incident management and service resolution 24x7, and this is where the need for an end-to-end dissolution model comes in. Here is where the Design thinking approach can make a world of difference. A human centric approach to solving these complex problems not only enables deeper customer engagement but also allows for creative, critical solutions that go beyond just usability.  
This is why AI & ML techniques come in handy - to reduce not just the number of tickets created per problem, but also dispatching them on time. Rule mining, problem classification techniques and time series predictions have been able to successfully decrease the time to dispatch for requests. 
Keeping this in mind, the iauro team has developed a state of the art application to digitize network operations effectively. With microservices at its core, this application will be used by NOC operators and decision makers to primarily deal with identity management and email servers effectively. In this particular case, the approaches for each problem are tentative and would be finalized post the data analysis for alarms raised, incident management and KPI related requirements.  
Common challenges and network problems from an NOC operator’s perspective, the role of AI and ML in enabling agile, multitask-able environments

AI-ML Approaches top level view

ML Ops pipelines for continuous training of models

App and Data Architecture What efficiency is made of  
Well-built mobile and web apps are defined by the easy accessibility and stability of their individual components. Thus, creating this skeleton in a manner that allows for both cost and resource efficiency becomes imperative at every step of the way.  
In this case, the frontend of the application consists of DAAIP applications and component libraries. The DAAIP unit consists of the following modules: 
Authentication Module
Responsible for user authentication and session management 
Alarm Module
Responsible for showcasing alarm details, correlation and summaries
Ticket Module
Responsible for providing network ticket summary in a specific timespan
KPIs Module
Responsible for providing KPI by Vendor-Region-Site-Cells 
AI Studio Module
Responsible for workspace management, data preprocessing and training configuration  
The application’s backend is microservices based, and runs on a combination of LDAP platform authentication mechanisms and reporting services. This provides NOC operators a real time view of current issues in the network, with cohesive email updates and notifications along with an audit trail. A microservices enabled environment tackles complexity by decomposing applications into smaller, more manageable clusters that are easier to maintain. It also enables independent development and delivery as per the ease of the developer, with more accurate error detection, readable programs and reusable codes. 

Application Toolset

Data Architecture : Data Store & Pipeline

Solution Architecture

The application is divided into three major components
MongoDB, MySQL, Redis. A cumulation of these three ensures secure information storage, proper reporting of microservice metadata and seamless communication between various tools and services 
Data Engineering
This drives the correlation of alarms, resolutions, time series predictions and anomaly identification. Network outages based on voice call drop, weather conditions and data speed can thus be preemptively predicted
Based on an angular framework, this is created using the Google maps library.
Experience of how app development is evolving nowadays, and what developers can do better to ensure a seamless frontend and backend 

Data Model : The foundation that holds it all together 

In any modernized application, the main role of a data model is to support the development of information systems by providing the system with the correct definition and format of data, such that there is no lag between its inflow and outflow.  

Layer 1 Deployment Architecture - MVP

ML Ops pipelines for continuous training of models

The app functions on five data processing pipelines including the following
KPI Data Processing Pipeline
Responsible for time series prediction of KPI future value and anomaly prediction as per the ML model. It also decides the backend processing for the same 
Ticket Data Processing Pipeline
Responsible for finding out RCA, resolution classification, group classification and problem category classification 
Alarms Data Processing Pipeline
Responsible for pushing alarm data to kafka through the app, finalise on alarms schema, design alarm dependencies and display the results in an alarm dashboard
CRQ Data Processing Pipeline
Responsible for reading CRQ Data and corresponding the same to input CRQ records 
Inventory Data Processing Pipeline
Responsible for aligning input inventory data table with lookup tables in Redis, design data processing, and implementing data processing models 
The importance of having different data libraries in place, their experience in making this solution what it is and tips for developers to keep in mind for similar future solutions

Tech Stack : Building seamlessness within pre existing processes  

When creating web applications, developers have a plethora of tech tools, solutioning mechanisms and libraries to choose from. The final decision making needs to rest upon a thorough understanding of the tool, the ways in which they interact with the environment, specific use cases and direct impact areas.   

Open source Big Data Toolset

The frontend interacts with the backend using REST APIs and HTTP service communication. The backend is designed in a node.js framework, with code level security and continuous code scan processes for catching vulnerabilities at regular intervals. 
Simplifying the decision making process of picking and choosing tech stacks as per the requirements of the customer, general thumb rules for what works and what doesn’t

End Impact

The end result of this activity was a highly efficient singular dashboard for NOC operators, which allowed them to access information related to requests, tickets and status updates visible in real time on a single screen. This not only saves the effort on the operator’s part, but also frees up their time. An easily manageable application infrastructure built on a microservices architecture makes for unparalleled customer experiences. With a promise of end to end delivery, this solution is a holistic approach towards ensuring not just customer satisfaction but customer delight - it has the potential to bring in new business, attract more consumers and ensure seamless delivery throughout.  

Read More