[1] Each resulting traffic class can be treated differently in order to differentiate the service implied for the data generator or consumer.
[4] A comprehensive comparison of various network traffic classifiers, which depend on Deep Packet Inspection (PACE, OpenDPI, 4 different configurations of L7-filter, NDPI, Libprotoident, and Cisco NBAR), is shown in the Independent Comparison of Popular DPI Tools for Traffic Classification.
A form to achieve this is by using traffic descriptors from connection traces in the radio interface to perform the classification.
It has been generally proven that using methods based on neural networks, vector support machines, statistics, and the nearest neighbors are a great way to do this traffic classification, but in some specific cases some methods are better than others, for example: neural networks work better when the whole observation set is taken into account.
This is traffic that the ISP deems isn't sensitive to quality of service metrics (jitter, packet loss, latency).
The applications use ICMP and regular HTTP traffic to discover servers and download directories of available files.
The encrypted BitTorrent protocol does for example rely on obfuscation and randomized packet sizes in order to avoid identification.