Abstract
This paper intends to introduce an implementation of a novel Self-Organization Map (SOM) in Host-based Stepping Stone Detection (SSD). Previous works have introduced Artificial Intelligence (AI) approaches such as Artificial Neural Network (ANN), however we found that the approaches are complex due to the requirement of variable to be known and tested to detect a stepping stone. SOM provides unsupervised capability in learning process. This feature helps to decrease the complexity of the AI approach. Moreover, this paper uses packet arrival time instead of Round Trip Time (RTT), which in turn reduces CPU usage as well as improves network load balancing. Through a series of real-time experiment, we show that our novel SOM approach is able to detect the stepping stone by only looking into the number of involved connection chain. In addition, the usage of SOM in Network-based SSD had been proven can detect the stepping stone in our previous research paper.