A Framework for the Fast Evaluation of the Capability-Based Connectivity Robustness of a Collaborative Information Network
MetadataShow full item record
The research objective of this thesis is to develop a method to measure the capability-based connectivity robustness of a Collaborative Information Network (CIN) against link failures by using existing topological connectivity robustness measures. A capability-based network modeling process was developed to transform the problem of measuring capability-based connectivity robustness into the problem of measuring the (structural) connectivity robustness between several critical node pairs. The connectivity robustness of a node pair can be directly quantified by the average number of link failures until its disconnection happens, which can be estimated using the effective resistance between the node pair. The estimation method proposed is fast and scalable. The estimation error stabilizes as network node number increases. Centrality analyses for both existing and non-existing network entities were performed in terms of their importance to the capability-based connectivity robustness of a CIN. The proposed centrality measures of network entities are based on the Moore-Penrose Pseudoinverse of a network Laplacian, which was also used to calculate the effective resistance value between a node pair. In addition, a framework for the fast evaluation of the capability-based connectivity robustness of a CIN was constructed and demonstrated, followed by an alternative topology design generation process. Assigning substitution nodes, which is a dynamic link failure copying mechanism, can help strengthen connectivity. In this thesis, it was also demonstrated how the proposed evaluation framework can be used to quantify the effects of having substitution nodes on the capability-based connectivity robustness of a CIN. Finally, the effects of the capability-based connectivity robustness of a CIN on the required information processing capacity of each network node were also explored.