A novel method to detect adversaries using MSOM algorithm’s longitudinal conjecture model in SCADA network

Abstract

-In a neural network, for classifying any high dimensional data, SOM (Self Organizing
Maps) is highly preferred. Self-Organizing Maps (SOM) are unsupervised neural networks that
cluster high dimensional data and transform complex inputs into easily understandable
inputs. It works on data sets and used to find previously unknown patterns. Patterns help in
categorizing elements for finding an association between them. They can detect the
adversaries and defects in the data. Neural networks are considered as one of the prolific
unsupervised learning methods which are a fast, and powerful technique that can be used to
solve many real-world problems. These networks are widely used for data representation. An
unsupervised algorithm has to understand the patterns in the data and then further process the
desired output. This research proposes a modified neural network algorithm called MUSOM
(mutated Self Organizing Maps). This algorithm has two major conjectures based on attacker’s
response on incoming packets from SCADA sensor. It has two components 1.nodes and
2.neurons. These nodes and neurons are inter connected with each other. One is used to
accomplish fault tolerance, error correction and dimensionality reduction concerning the
unknown anomalies. The next one is to find the outlying anomalies in the network