Two optimization approaches for parameter determination of PNNI complex node model

The PNNI (private network-network interface) specification was recommended by The ATM Forum for path selection and connection management in ATM networks. It is a routing algorithm specially designed for the ATM networks. We propose two optimization approaches for the parameter determination of the PNNI complex node model. One optimal objective is the least square approximation and the other one is the max deviation minimization. The generalized inverse matrix and linear programming are used to find the solutions. We also propose a practical and heuristic criterion for setting up bypasses. The numerical results show the efficiency of our approaches.