Operational Analysis of Survivable Networks

Ong Keng Soon Alvin

Abstract

In networked systems such as those found in communication, transportation and computers, the effectiveness and survivability of the system depends on the combinations of nodes and links that have been rendered inoperative. Disruptions to communication networks could result in serious consequences, especially in military contexts. Enormous advantages would be gained if the communications networks of opposing forces were disrupted. This project aims to give the preliminary results on the cost efficient way to destroy the specified node and/or link of the network using fictitious sample data sets.

Firstly, a general non-linear model was formulated to represent the most cost efficient way to destroy the nodes and/or links of a communication network. Decision variables were identified and accompanying constraints were added to complete the model. The model computes the effective success rate of destroying the node/link and calculates the minimum amount of resources required to achieve this.

After formulation, spreadsheet optimisation method, using Microsoft Excel, was used for quantitative analysis. The model was reconstructed onto a spreadsheet and an add-in tool, Solver, was used to calculate the best estimates for an optimal solution.

Next, a GUI (Graphical User Interface) environment was then developed using Visual Basic to interface the spreadsheet model to the users with no prior knowledge of operations research. The users only need to input the required parameters for the spreadsheet program and the answer for the best estimate would be returned.

The return value (cost of resources) is a close estimate but not necessarily the best. This is due to the fact that for the non-linear model, solver will take the best current feasible solution as the optimal when no further improvements can be made. This, however, may only be a locally optimal solution. Further investigations could be done to improve the chances of obtaining the true value.