Iterative Max-Min Fairness Network Resource Allocation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for allocating network resources among multiple traffic demands are inefficient, scaling poorly with network size, lacking control and understanding for operators, and are inherently centralized and unstable.
Innovation Solution
A method and network system that utilize upward max-min fairness (UMMF) principles, where traffic demands are managed through routing circuitry to increase flow values on certain paths while decreasing others, with routers generating authorization messages to optimize resource allocation, and an iterative algorithm (Iterative Exhaustive WaterFill) to determine fair splits across paths.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If the standard iterative solution of multiple linear programs is used to find max-min fairness allocation, then the allocation fairness is improved, but the computational complexity and scaling performance deteriorate
Solution Approach 1:
The patent segments the complex max-min fairness problem into multiple simpler sub-problems by introducing a sequence of linear programs with increasing complexity. Each linear program optimizes a specific aspect of the allocation, building upon previous solutions. This segmentation allows the system to achieve fair allocation without solving the entire complex problem at once, thereby reducing computational burden while maintaining fairness guarantees.
2Ease of operation
If the centralized iterative solution is used, then the allocation control is improved, but the system stability and distributed implementability deteriorate
Solution Approach 1:
The patent divides the centralized control function into distributed components by assigning different linear programs to different network entities. Each entity independently executes its assigned linear program based on local information and feedback, eliminating the need for a single centralized controller. This segmentation improves system stability by removing single-point failures while maintaining coordinated control through the iterative optimization process.
Solution Approach 2:
The patent implements feedback mechanisms where each linear program's output feeds into subsequent programs, and network state information flows back to adjust optimization parameters. This feedback loop allows the distributed system to maintain coordination and converge to the fair allocation solution without requiring centralized decision-making, thereby improving both stability and distributed implementability while preserving control.
3Measurement precision
If the iterative solution is used, then the allocation accuracy is improved, but the convergence time and computational resources deteriorate
Solution Approach 1:
The patent applies preliminary action by pre-ordering the sequence of linear programs and pre-defining their objective functions and constraints. This preparation allows the iterative process to converge faster because each subsequent program builds on the optimized foundation of previous programs rather than starting from scratch. The preliminary structuring of the optimization sequence reduces redundant computations and accelerates convergence while maintaining allocation accuracy.
Data Source
AI summary
Systems and methods are provided for allocating resources of a network among a plurality of traffic demands to optimize fairness and network utilization. Methods based on flow-increase dynamics converge toward an upward max-min fair (UMMF) allocation, in which the value of each traffic demand cannot be increased, along any of its paths, even if larger traffic demands are removed from the network. An efficient iterative algorithm that converges to a UMMF solution is also provided. The described methods and systems can be implemented efficiently, distributively, and asynchronously.