Towards Automatic Traffic Classification

Lai, Zhaohong; Galis, Alex; Rio, Miguel; Todd, Chris

doi:10.1109/icns.2007.120

Cited by 6 publications

(4 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our work combines these research results and takes this further, where MGtoolV2 collects a diverse set of traffic features and these features are transformed into XML and MSC models. Lai et al [15] describe the flow concept in modeling network activities. In our work, the MCS model concretizes the modeling of network traffic flow and Eq.…”

Section: B Comparison With Current State-of-the-artmentioning

confidence: 99%

“…A clustering method [10], Hidden Markov Models [9], or a statistical approach [15], [16] are good approaches for indicating the network traffic behavior. However, when models for describing traffic sequences are required, these approaches are not enough.…”

Section: B Comparison With Current State-of-the-artmentioning

confidence: 99%

“…When this is combined with the XML models describing details of single packets, it provides an efficient way to model and to describe the network traffic behavior. When compared to works in [9], [15], and [16] the models created using XML and MSC notations are more usable. The XML and MSC notations are well-known and the models can be deployed using simple parser technologies.…”

Section: B Comparison With Current State-of-the-artmentioning

confidence: 99%

“…The most common features that are collected from network traffic are the source IP address, destination IP address, source port number, destination port number, and used protocol (such as TCP, UDP (User Datagram Protocol), or ICMP (Internet Control Message Protocol)) [11], [13], [14]. These and packet size information are important for collecting the network traffic flow information [15]. Extracting only the relevant features lowers the costs and the resource consumption [16].…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Automatic Model Creation to Support Network Monitoring

Maatta

Räty

2014

IEEE Access

View full text Add to dashboard Cite

The large variety of network traffic sets many challenges in modeling the essential aspects of network traffic flows. Analyzing and collecting features for the model creation process from the network traffic traces is a time-consuming and error-prone task. Automating these procedures are a challenge. The research problem discussed in this paper concentrates on the analysis and collection of features from the network traffic traces for the model development process, by automating the analysis and collection. The proposed system of this paper, called MGtoolV2, supports the model development process through the automation of collection and analysis in the actual model creation procedures. The model development process aims to enhance the development of a model by reducing the development cost and time. The proposed tool automatically creates large sets of models according to the network traffic traces and minimizes the errors of manual modeling. The experiments conducted with MGtoolV2 indicate that the tool is able to create the models from the traffic traces cost effectively. MGtoolV2 is able to unify similarities between packets, to create very detailed models describing specific information, and to raise the abstraction level of the created models. The research is based on the constructive method of the related publications and technologies, and the results are established from the testing, validation, and analysis of the implemented MGtoolV2.

show abstract

Section: B Comparison With Current State-of-the-artmentioning

confidence: 99%