User Tracking Based on Behavioral Fingerprints

The pervasiveness of wireless communications networks is advancing particularly in metropolitan areas. Broadband computer networks as IEEE 802.11 are seriously competing with cellular network technologies such as UMTS and HSDPA. Unfortunately, this increased mobility comes with privacy and security related issues. We are currently in the process of identifying possible attacks on the privacy of wireless network users, since the development of effective countermeasures is only possible with a thorough understanding of such attacks.

One serious threat we are discussing here, is the tracking of users in metropolitan networks by means of determining their physical location. Any individual user can be identified either by the devices she is using or by the behavior she is displaying. Suitable features range from single identifiers such as IP or MAC addresses to complex conglomerates of different values that provide valuable information due to their combination.

This article focuses on the extraction and analysis of features that are valuable for fingerprinting by employing

Activation Patterns

, a concept based on artificial intelligence and machine learning techniques. The concept is applied to email header data, since this allows for an effective illustration of the employed techniques. Furthermore, due to the human understandable data, we can easily evaluate the effectiveness of the concept before we start to analyze more complex data-sets.