Identifying Skype Traffic by Random Forest

Despite of the great popularity, little is known about Skype network attributed to proprietary protocol. End-to-end encryption disables the traditional traffic detection methods. We presented genetic algorithm based Random Forest algorithm to identify Skype traffic using only transport layer statistics. Experimental results show that the proposed approach can immune to the encryption of the payload and be capable of detecting Skype traffic with accuracy over 95% while low computational complexity is required.