Wi-Fi is all around us and can be very useful when on the move. Many places such as coffee shops, libraries and restaurants offer free open Wi-Fi for their customers to connect to. There are however many threats against open Wi-Fi, such as eaves dropping, man in the middle, evil-twin and more. There are ways to mitigate these threats, but another solution is to instead of connecting to the open Wi-Fi, clients can connect to mobile cellular networks. Using machine learning to analyze data can reveal how common these attacks are and whether or not it’s worth connecting to public Wi-Fi.

A pilot survey revealed that the people who responded to the survey generally trust mobile networks more than public Wi-Fi, and would avoid the latter when possible. The most common places where people use public Wi-Fi were airports and train stations. This could be caused by mobile data being higher outside people’s home countries and trains often traveling outside of cellular range.

Using machine learning models to detect public network traffic anomalies is becoming more important when unethical users are finding more ways to trick the general user. With these kinds of models, they can be trained to understand common network attacks and their patterns. By testing a machine learning model with a public dataset, this thesis shows how well it can perform and how a well trained machine learning model can identify these kinds of patterns.