Beginning Anomaly Detection Using

 Real-World Use Cases of Anomaly Detection

We will look at several industry verticals and businesses, and how anomaly detection can 

be used.

Figure 8-2.  Daily count of cars crossing

Chapter 8   praCtiCal Use Cases of anomaly DeteCtion

---

300

 Telecom

In the telecom sector, some of the use cases for anomaly detection are to detect roaming 

abuse, revenue fraud, and service disruptions. So how do we detect roaming abuse in the 

telecom sector? By looking at the location of the cellular devices, we can categorize the 

kind of behavior of the cellular device at any particular moment as normal or abnormal. 

This helps us detect cellular device usage at that period of time. By looking at all of 

the other information we know in general about roaming activity, we can also detect 

how this cellular device is being used and whether any roaming abuse is taking place. 

Figure 

8-3

 shows how roaming works for your phone as you travel around the world.

Service disruption is another very high impact use case for anomaly detection. 

Cellular devices are connected to cellular networks via towers, which are all over the 

place. Your cell phone connects to the nearest tower in order to participate in the cellular 

network. In case of events involving large crowds such as a concert or a football game, 

the cellular towers that typically perform quite well get heavily overloaded, causing 

serious service disruptions and very bad customer experience for the duration of the 

overload. Figure 

8-4

 shows a service disruption of phone service in the northwestern 

United States.