International Conference on Computer and Knowledge Engineering

Home / 14th International Conference on Computer and Knowledge Engineering

Optimizing Text-Based Protocol Clustering in Reverse Engineering with Auto-Encoders and Fine-Tuned Parameters

Authors :

Shiva Mahmoudzadeh¹ Mohaddese Nemati² Mehdi Teimouri³

1- University of Tehran 2- University of Tehran 3- University of Tehran

Keywords :

Protocol reverse engineering،clustering،text-based protocols،Auto-encoder،K-means،DBSCAN

Abstract :

As documentation for unknown protocols is often confidential, reverse engineering is required for analysis. This multi-step process is time-consuming, leading researchers to automate it with tools. A key aspect of this process is classifying similar messages into categories. Traditional clustering methods in deep learning and machine learning are used for this purpose. This research aims to evaluate and test clustering methods such as K-Means, DBSCAN, Agglomerative Clustering (AC), and Affinity Propagation (AP) on text-based protocols. To improve cluster homogeneity, parameter tuning methods, including auto-encoder feature extraction, are utilized. Experiments reveal that combining Auto-encoder with K-Means (78% homogeneity) and Auto-encoder with DBSCAN (86% homogeneity) are the most effective for clustering text-based protocols. These methods can automatically classify various message types.