Nested Optimization for Network Topology and Traffic Engineering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current technologies face challenges in optimizing network topology design and configuration to meet growing traffic demands and failure scenarios, often leading to over-provisioning and inefficiencies in cloud and remote computing systems.
Innovation Solution
A nested optimization system that combines traffic engineering and network topology constraints using convex optimization techniques to determine minimal link capacities, enabling dynamic reconfiguration and automated scaling of network topologies in response to real-time changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional traffic engineering and network topology design are performed separately, then each can be optimized independently, but the overall network capacity utilization is suboptimal and leads to over-provisioning
Solution Approach 1:
The patent combines traffic engineering and network topology design into a unified nested optimization framework. The outer optimization loop determines network topology configuration while the inner optimization loop determines traffic engineering parameters, allowing both to be optimized simultaneously rather than independently. This integration enables the system to achieve optimal capacity utilization without over-provisioning, as the topology and traffic parameters are co-optimized to work together efficiently.
Solution Approach 2:
The patent implements a nested optimization structure where the inner optimization problem (traffic engineering) is embedded within the outer optimization problem (topology design). The inner optimization determines optimal traffic routing and bandwidth allocation for a given topology, while the outer optimization selects the optimal topology configuration based on the results from the inner optimization. This nested approach allows hierarchical optimization where inner-loop solutions inform outer-loop decisions, achieving global optimality.
2Adaptability or versatility
If manual network configuration and updates are performed, then changes can be carefully reviewed and approved, but the system cannot respond quickly to changing traffic demands and failure scenarios
Solution Approach 1:
The patent pre-computes and stores optimal network configurations for various traffic demand scenarios and failure modes using the nested optimization framework. When a change in traffic demand or failure scenario occurs, the system can quickly retrieve or slightly adjust pre-computed configurations rather than performing full optimization from scratch. This preliminary computation enables rapid response to changing conditions while maintaining optimality.
Solution Approach 2:
The patent implements dynamic network configuration that automatically adapts to changing traffic demands and failure scenarios. The nested optimization system continuously monitors network conditions and dynamically adjusts topology and traffic engineering parameters in response to real-time changes. This dynamic approach replaces static manual configuration with automated adaptive control, enabling the network to respond quickly to changing conditions while maintaining optimal performance.
3Ease of manufacture
If iterative non-convex or greedy optimization processes are used, then the problem can be solved with simpler algorithms, but the solutions may be suboptimal and require over-provisioning
Solution Approach 1:
The patent transforms the original non-convex optimization problem into a convex optimization problem by changing the parameterization and formulation of the optimization variables. Specifically, the patent uses convex relaxation techniques and reformulates the topology design and traffic engineering parameters in a way that makes the overall optimization problem convex. This allows the use of efficient convex optimization algorithms that guarantee global optimality rather than getting trapped in local optima, while maintaining computational tractability.
Data Source
AI summary
Improved network optimization and scaling are provided by combining optimization of traffic engineering and network topologies. Sets of minimal link capacities are determined based on the constraints of network topology, network demand, and failure modes. The optimization problem is reframed to allow for simultaneous optimization across all failure modes, reducing or eliminating overprovisioning in link capacity allocation by utilizing convex optimization solvers and techniques.