Neural networks for shortest path computation and routing in computer networks

Area of Science:

• Computer Science
• Network Engineering
• Artificial Intelligence

Background:

• Optimum routing in packet-switched networks aims to minimize network-wide average time delay.
• Real-time shortest path computations are crucial for effective routing algorithms.
• Existing neural network models have limitations in efficiency and adaptability.

Purpose of the Study:

• To propose an efficient neural network shortest path algorithm for real-time optimum routing.
• To improve upon existing Hopfield models for network routing applications.
• To enable adaptive routing in response to dynamic network conditions.

Main Methods:

• Development of a novel neural network architecture for shortest path calculations.
• Detailed discussion of the design principles of the proposed neural network.
• Validation of computational power through computer simulations.

Main Results:

• The proposed neural network efficiently computes shortest paths in real time.
• The model demonstrates adaptability to changes in link costs.
• The network effectively handles dynamic network topology adjustments.

Conclusions:

• The developed neural network provides a real-time, adaptive solution for optimum routing in packet-switched networks.
• This approach enhances the efficiency and responsiveness of network routing algorithms.
• The model represents a significant improvement over previous neural network routing methods.