SBIR Phase I: A Hybrid Neural Net Enterprise System Simulation

This Small Business Innovation Research Phase 1 project is designed to address the problem of managing and simulating extended enterprise systems as specified in The Extended Enterprise/C.2. As networks and enterprise systems have grown in complexity, it has become increasingly difficult to determine the most cost effective way to deploy hardware to maximize system availability and network performance. Poor resource planning decreases the effectiveness of information technology organizations, resulting in increased labor costs and decreased company wide productivity. New tools are required to increase productivity. This proposal is for research to produce a modeling and forecasting software system that use combination of neural net and rules based algorithms to simulate enterprise systems. Individual components of the enterprise system will be modeled with appropriate neural nets trained by constructive techniques. The enterprise system will be modeled by connecting these component neural nets with data pipe rule-based software. Such software would be able to forecast strains on the system and identify bottlenecks in the network. The completed modeling and forecasting software will become a cornerstone technology for a tool that will diagnose, troubleshoot and predict IT issues. The project will be developed in Linux with C, C++ or FORTRAN, as appropriate. This research could help in making the management of enterprise systems and networks more efficient therefore making them more affordable to teaching and service institutions. The result of this endeavor will assist enterprise systems and network managers to manage the tasks in faster and more economically manner.