Dataset supporting the University of Southampton Doctoral Thesis "An investigation into passive information gathering for uncooperative ad-hoc mesh network assessment"

Dataset supporting the University of Southampton Doctoral Thesis "An investigation into passive information gathering for uncooperative ad-hoc mesh network assessment"

Data contains results from multiple simulation runs. The dataset is in 2 parts, the first being the data that supports Phase 2: Routing Algorithm Reconnaissance in Ad-Hoc Mesh Networks.

The first experiment contains results for 6 machine learning techniques: Support Vector Machine, Random Forest, Convolutional Neural Network, Bernoulli Naive Bayes, Gaussian Naive Bayes and Deep Forest. For each of these various combination of network packet fields sizes were used. These were ‘Subtype’, ‘Header Length’ , ‘Frame Length’ and UDP length and concerns 2 or 3 classes of routing algorithms.

The second experiment contains results for 3 machine learning techniques: Support Vector Machine,Random Forest and Convolutional Neural Network. It uses the same network packet fields.

The third experiment is a repeat of the second experiment but with the number of nodes being incrementally reduced. The other dataset supports Phase 3: Predicting Node Importance In Temporal Dynamic Networks. The data for this phase contains information for link prediction techniques and the accuracy of the prediction vs ground truth. This is from 700 simulation runs of the mesh network with each of the 3 mobility models.

The data is presented as excel files and is accessible via CC BY license.