Implementation, Experiments and Improvement of Optimal Trust System Placement in Smart Grid SCADA Networks

ABSTRACT: The aim of this thesis is to investigate the optimal Trust System Placement (TSP) method for smart grid Supervisory Control And Data Acquisition (SCADA) networks. At present, as SCADA networks are connected to the internet the scope of cyber-security concerns becomes much wider. Trust Systems (TSs) are deployed to provide the cyber-security of SCADA networks. TS are used to detect and block malicious activities. Optimal TSP problem is used to minimize the cost and maximize the security by selecting minimum number of TSs. Segmentation is the main part of the optimal TSP problem. Segmentation is used to divide the SCADA graph to small segments and ideal segmentation problem uniforms the size of the segment. Linear Programming Problem (LPP) is used to assign the TSs to some nodes among the bordering nodes and its constraint is that all inter-segment links are connected to at least one trust node. The experiments are conducted on five IEEE test system topologies. The IEEE test system is categorized into small and large networks. The obtained results show that by increasing the quantity of segments the required quantity of TSs increases and small networks are more balanced than the large networks in size of segments. In optimal TSP problem TS number per segments are not uniform over the segments. It may deteriorate the security of the segments and more delays happened to the inter-segment links. We propose optimal TSP uniformity problem to maximize the security and minimize the operational expenditure by distributing TSs over the segments. After executing the optimal TSP uniformity on IEEE BUS 14, the distribution of TSs over the segments is improved by 100% with 3 segments and 5 segments and it is improved by 81% with 4 segments and it is improved by 35% with 6 segments. Keywords: Smart grid Supervisory Control and Data Acquisition (SCADA) network, cyber-security, trust node, Minimum Spanning Tree (MST), Linear Programming Problem (LPP), segment, bordering node, uniformity problem