Nokia AVA Anomaly Detection in Telecom
Find hidden patterns to improve telco customer service
As data continues to grow exponentially and move to the cloud, Communications Service Providers (CSPs) struggle to keep up. Rising network complexity, rapidly changing user behaviour and the growing threat of cyber-attacks can all lead to errors and failures.
Although communications networks are stable and reliable, there will always be a certain number of ‘normal’ failure events – but what about the telco network errors or anomalies that come and go quickly? By their nature, these are hard to detect and analyze, yet pinning them down is vital, as they could lead to more significant failures in your network.
How to perform network anomaly detection?
You’ll want to find the most important hidden patterns and connections in the mass of telco network data, patterns that might point to anomalies. Finding these patterns shows what is really happening in the network, allowing you to take action to increase efficiencies, improve telco security and provide a better customer experience.
Reduce anomaly detection time and cost to resolve issues
Telecom anomalies can be classed as high, medium or low severity.
The costs of medium and low severity anomalies outweigh the cost of high severity anomalies.
This is because lower severity anomalies are frequently harder to detect and last longer than high severity anomalies, and thus have a greater impact on true costs over time.
Advanced anomaly detection, using real-time machine learning detection of all anomalies, can save nearly USD40 million per annum compared to conventional techniques.
This means that anomaly detection can be the difference between business-as-usual and ongoing glitches that cost millions
Anomaly detection in telecom
Many CSPs recognize this need and are seeking to improve the way they detect anomalous telco network events. A survey by Heavy Reading suggests that about half of CSPs expect AI-based telco network automation to save costs by automatically detecting and predicting issues and congestion before they cause an outage or network performance issues (Source: Heavy Reading global CSP survey, Aug 2021).
The telco anomaly detection opportunity in numbers
Telecom anomaly detection and diagnosis to the rescue
It is difficult to detect anomalous events manually – it is vital to have a system based on artificial intelligence/machine learning to read the data and tell you what the problem is and what actions you need to take.
This is where Anomaly Detection and Diagnosis comes in. Part of Nokia AVA,Anomaly Detection and Diagnosis is a Software-as-a-Service (SaaS) that reports on and ranks the top anomalous conditions in the telco network and links them to the most likely root causes. The result is that you can address network issues before they affect your customers.
The logical sequence of Detection, Diagnosis, Explanation helps you move from reactive to proactive telco network operations. The Nokia AVA solution is multi-vendor and supports a variety of use-cases, including network analysis, security, IoT and many more.
Towards a better telco network
Anomaly Detection and Diagnosis uses a series of Bell Labs algorithms to analyze telco network performance data, categorize network behavior and identify previously unknown anomalies.
Using machine learning, Nokia AVA automatically identifies issues such as call set-up failures.
Anomaly detection benefits for your telco network:
- Over 70% new failure scenarios detected
- Vendor and data agnostic
- Detection of new telco network issues not captured by today’s tools
- Efficient problem fixing through a catalogue of root causes
- Automated monitoring and actions reduce time to correct problems
Case study: trusted by Vodafone
Vodafone’s European networks were facing a number of challenges. These included: Data segregated in silos across the organization; a variety of tools that were not harmonized; undetected network anomalies that were affecting customer experience.
Nokia’s Anomaly Detection solution, deployed on Google's public cloud, helped discover, identify and remedy problems in network and services.
Vodafone’s use of anomaly detection saw several major benefits:
- Increased telco customer experience
- From hours to near real-time to detect anomalies, unrecognizable by humans
- Nokia and Google allowed Vodafone to adopt a ‘develop once, deploy many times’ model, which has resulted in a 60–70% reduction in effort.
- Reduce the time and cost to find network performance anomalies and hidden underlying issues: 25-30% faster RCA
Nokia AVA Anomaly Detection is now being rolled out across Vodafone’s pan-European network.
Related products and services
AVA- Fixed Network Analytics
Detect and remediate network, service and device issues proactively before customers are impacted
AVA - Customer experience analytics
Power your customer service management and improve your customer experience
AVA - Mobile Network Insights
Automatically correlate 5G network quality
AVA - Network Data Analytics Function
Create new revenue opportunities and monetize valuable use cases leveraging AI/ML and deep network insights from data analytics.
AVA - Traffica
Real-time network traffic analytics for monitoring and troubleshooting
AVA Open Analytics framework
Our Open Analytics framework helps CSPs accelerate AI projects
AVA - AI for Network Performance
Augment human intelligence to improve decision making & increase network efficiency
AVA AI for video services
Use AI to uncover hidden patterns to enhance experience
AVA – Energy efficiency
How AI reduces the CO2 emissions and network energy costs of telco networks