Randomized algorithms for stability and robustness analysis of high-speed communication networks.

Area of Science:

• Computer Science
• Network Engineering
• Algorithm Analysis

Background:

• High-speed communication networks face stability challenges due to congestion and delays.
• Utility-maximizing flow controllers are crucial for individual user satisfaction.
• Existing models often lack robustness analysis for complex network dynamics.

Purpose of the Study:

• To develop and apply randomized algorithms for analyzing the stability of high-speed communication networks.
• To investigate congestion and delay-based flow controllers for utility-maximizing sources.
• To provide a comprehensive stability analysis for various network configurations.

Main Methods:

• Introduction of a nonlinear algorithm for source flow controllers using aggregate network feedback.
• Linearization of the nonlinear model and robustness analysis for symmetric cases.
• Development of randomized algorithms (Monte Carlo, quasi-Monte Carlo) for nonsymmetric network scenarios.

Main Results:

• Derivation of closed-form conditions for stability and robustness in symmetric network cases.
• Exact computation of parameter ranges guaranteeing network stability.
• Probabilistic stability estimates for general nonsymmetric networks with random topologies.

Conclusions:

• The study provides a complete analysis of congestion control algorithms for internet-style networks.
• Randomized algorithms offer effective solutions for stability analysis in complex, nonsymmetric network conditions.
• The findings are applicable to networks with single bottleneck nodes and general random topologies.