Evaluating ORB and SIFT With Neural Network as Alternatives to CNN for Traffic Classification in SDN Environments
Software-Defined Networking (SDN) centralizes control and separates the control plane from the data plane, enhancing flexibility and efficiency but also increasing the attack surface. Consequently, effective traffic classification is crucial for identifying cyber threats within SDN infrastructures....
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10937120/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850046892441862144 |
|---|---|
| author | Hicham Yzzogh Hafssa Benaboud |
| author_facet | Hicham Yzzogh Hafssa Benaboud |
| author_sort | Hicham Yzzogh |
| collection | DOAJ |
| description | Software-Defined Networking (SDN) centralizes control and separates the control plane from the data plane, enhancing flexibility and efficiency but also increasing the attack surface. Consequently, effective traffic classification is crucial for identifying cyber threats within SDN infrastructures. To address this need, we explore the potential of Convolutional Neural Network (CNN), which is known for its superior performance in image classification tasks. To leverage CNN in this context, we propose converting traffic flow data into images using the flow-to-bar chart conversion method. We also employ two well-known methods for generating images from tabular data: the Image Generator for Tabular Data (IGTD) based on Euclidean distance and the IGTD based on Manhattan distance. As an alternative to CNN, we use ORB (Oriented FAST and Rotated BRIEF) or SIFT (Scale-Invariant Feature Transform) for feature extraction, combined with neural network for classification. This approach reduces computational overhead while maintaining high accuracy. Our approach, which incorporates flow-to-bar chart conversion along with ORB and a neural network (NN), achieved 97.14% accuracy on the InSDN dataset and 99.86% on the CIC-DDoS2019 dataset, demonstrating its effectiveness in traffic classification. |
| format | Article |
| id | doaj-art-8dfafc56cb47464ca6efdc03b48b081a |
| institution | DOAJ |
| issn | 2169-3536 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj-art-8dfafc56cb47464ca6efdc03b48b081a2025-08-20T02:54:22ZengIEEEIEEE Access2169-35362025-01-0113514845149810.1109/ACCESS.2025.355344910937120Evaluating ORB and SIFT With Neural Network as Alternatives to CNN for Traffic Classification in SDN EnvironmentsHicham Yzzogh0https://orcid.org/0009-0009-8532-7371Hafssa Benaboud1https://orcid.org/0000-0002-5027-8217IPSS, Faculty of Sciences, Mohammed V. University in Rabat, Rabat, MoroccoIPSS, Faculty of Sciences, Mohammed V. University in Rabat, Rabat, MoroccoSoftware-Defined Networking (SDN) centralizes control and separates the control plane from the data plane, enhancing flexibility and efficiency but also increasing the attack surface. Consequently, effective traffic classification is crucial for identifying cyber threats within SDN infrastructures. To address this need, we explore the potential of Convolutional Neural Network (CNN), which is known for its superior performance in image classification tasks. To leverage CNN in this context, we propose converting traffic flow data into images using the flow-to-bar chart conversion method. We also employ two well-known methods for generating images from tabular data: the Image Generator for Tabular Data (IGTD) based on Euclidean distance and the IGTD based on Manhattan distance. As an alternative to CNN, we use ORB (Oriented FAST and Rotated BRIEF) or SIFT (Scale-Invariant Feature Transform) for feature extraction, combined with neural network for classification. This approach reduces computational overhead while maintaining high accuracy. Our approach, which incorporates flow-to-bar chart conversion along with ORB and a neural network (NN), achieved 97.14% accuracy on the InSDN dataset and 99.86% on the CIC-DDoS2019 dataset, demonstrating its effectiveness in traffic classification.https://ieeexplore.ieee.org/document/10937120/Software-defined networkingconvolutional neural networkscale-invariant feature transformoriented fast and rotatedsecuritytraffic classification |
| spellingShingle | Hicham Yzzogh Hafssa Benaboud Evaluating ORB and SIFT With Neural Network as Alternatives to CNN for Traffic Classification in SDN Environments IEEE Access Software-defined networking convolutional neural network scale-invariant feature transform oriented fast and rotated security traffic classification |
| title | Evaluating ORB and SIFT With Neural Network as Alternatives to CNN for Traffic Classification in SDN Environments |
| title_full | Evaluating ORB and SIFT With Neural Network as Alternatives to CNN for Traffic Classification in SDN Environments |
| title_fullStr | Evaluating ORB and SIFT With Neural Network as Alternatives to CNN for Traffic Classification in SDN Environments |
| title_full_unstemmed | Evaluating ORB and SIFT With Neural Network as Alternatives to CNN for Traffic Classification in SDN Environments |
| title_short | Evaluating ORB and SIFT With Neural Network as Alternatives to CNN for Traffic Classification in SDN Environments |
| title_sort | evaluating orb and sift with neural network as alternatives to cnn for traffic classification in sdn environments |
| topic | Software-defined networking convolutional neural network scale-invariant feature transform oriented fast and rotated security traffic classification |
| url | https://ieeexplore.ieee.org/document/10937120/ |
| work_keys_str_mv | AT hichamyzzogh evaluatingorbandsiftwithneuralnetworkasalternativestocnnfortrafficclassificationinsdnenvironments AT hafssabenaboud evaluatingorbandsiftwithneuralnetworkasalternativestocnnfortrafficclassificationinsdnenvironments |