Friday, July 19, 2024

Optimizing Network Operations During Times of Change

Datamation content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More.

As the impact of COVID-19 has shifted many aspects of our personal and professional lives, one such area of change is in how we use the internet. With many people staying home, the result has been usage of mobile usage at an all-time high, increased rates of uploading, and massive surges in video conferencing.

As many individuals and companies have shifted their internet consumption habits, this has meant that telecommunications companies have had to rapidly adapt to these changes. 

With many countries depending on this digital infrastructure to keep their economies going, this has also led to a much greater reliance on network performance.

With these changing usage and behavioral patterns, many network incidents that could be reliably predicted in previous years are now much harder to prevent. Along with these challenges, however, many telecom operators have increased pressure to grow revenue and profitability.

In this article, we’ll review a few of the specific challenges faced by telecom companies during these times of change. We’ll then look at how AI and machine learning are being used to monitor and improve network performance.

The Challenges of Managing Network Performance and Usage Patterns During Times of Change

Before the global pandemic, broadband consumption patterns around the world were quite predictable — the majority of people were at work or school during  the day and usage would predictably drop. Then in the evening, internet usage would increase.

Now, with parents and children both at home for most of the day, everything from social media to video conferencing usage has skyrocketed. As a result, without this predictable downtime when network incidents could be resolved, many network providers have been scrambling to keep up with demand. As a result, minimizing network downtime and continuous forecasting of network traffic to cater to changing usage patterns, especially as many countries who have emerged from the first wave of lockdowns are now potentially facing a resurgence and potentially a second period of quarantine.

Shifts in Uplink Traffic Leading to Network Issues

In particular, one common theme prior to the global pandemic was that most people’s internet consumption primarily consisted of downlink usage, including downloading web pages, video streaming, and so on.

Now with more and more people working from home, many people’s rates of uploading have drastically increased. Whether it’s for remote learning, video conferencing, or uploading to social media—networks simply weren’t designed to handle this much uplink traffic.

This means that while the companies need to monitor for faults in the network, they simultaneously need to be able to forecast future demand and re-engineer usage capabilities.

Similarly, now that many telco employees need to work from home, they also need to manage the network at a distance. As their existing tools weren’t built to do this, telco companies need to upgrade their tools to be able to detect issues autonomously.

In short, they need to utilize their resources more efficiently and take advantage of autonomous network monitoring. This is where AI and machine learning come into the picture.

AI & Machine Learning for Network Monitoring

Many operators are now monitoring and optimizing their networks with the use of big data, machine learning, and artificial intelligence.

Although we’re currently in the early days of AI adoption in the telecommunications industry, the fact that communications networks are so complex and so data-rich makes the potential for disruption increasingly important.

In order to take advantage of these emergent technologies, two core applications of AI and ML for communications network should be considered, including:

●  Anomaly Detection: As mentioned, with behavioral and data consumption patterns shifting dramatically, detecting anomalies with static thresholds simply isn’t viable. Instead, a branch of machine learning called unsupervised learning can be used to learn each individual metrics normal behavior on its own. As this normal behavior constantly shifts the anomaly thresholds also automatically shift, resulting in increased granularity and fewer false positives.

●   Demand Forecasting: With the increased reliance on network performance, accurate demand forecasting has become more important than ever. Similar to anomaly detection, AI and machine learning can take in 100% of the data in order to forecast user demand so the appropriate network provisions can be provisioned in time.

Now that we’ve discussed the application of AI for communications networks, let’s review a real-world use case and see how telcos are already taking advantage of their data.

Use Case: AI for Fixed Broadband Access Networks

Many telcos have transformed themselves into fixed and mobile service providers, which can include a complex mix of technologies, such as:

●      Fiber to the premise, node, or curb

●      Digital Subscriber Loop (DSL)

●      Hybrid Fibre Coaxial (HFC)

●      WiFi

●      Satellite broadband

Each of these technologies will experience subscriber uplink and downlink incidents such as throughput drops, packet loss, and many other KPIs, which means that each needs to be monitored in real-time. In the example below you can see an anomaly was detected in downstream throughput drops in a HFC network.

Through the correlation engine that accompanies each anomaly, the AI-based monitoring solution was able to link the incident to the following incidents:

●      A drop in uplink throughput

●      A spike in upstream Codeword Errors (CER)

●      A drop in upstream Signal-to-Noise Ratio (SNR)


Based on these incidents, the telco was able to be proactive about the anomalies and could notify customers in the region about the service degradation. What’s more, by having the exact correlated anomalies that caused the incidents, their technical team was able to resolve the issue much faster than previous cases.

Summary: AI for Telecom

As the global economy has become increasingly complex and connected, the reliance on network performance has never been greater. From shifting consumption patterns to surges in uplink traffic — telecom operators have had to adapt their networks rapidly, and in many cases remotely.

For these reasons, many telcos are starting to embrace AI and machine learning for network monitoring. In particular, two of the main applications of AI for communications networks are anomaly detection and demand forecasting.

Although we’re still in the early days of AI adoption, these global changes make it more apparent than ever that these emergent technologies present an opportunity to improve performance, increase efficiency, and stay ahead of the competition.

Author Bio:

Vikram Pulakhandam is Anodot’s Solutions Director for AIPAC, leading pre-sales technical engagements across the region. Prior to joining Anodot, he led the Big Data Engineering team for an Australia telco operator. He has more than 20 years of mobile telco experience with a variety of vendors across the globe

Subscribe to Data Insider

Learn the latest news and best practices about data science, big data analytics, artificial intelligence, data security, and more.

Similar articles

Get the Free Newsletter!

Subscribe to Data Insider for top news, trends & analysis

Latest Articles