Diversity Routing Convex Optimization Jitter Tradeoff
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current network technologies face challenges in maintaining low latency and jitter, especially in dynamic environments with time-sensitive applications like video streaming and tele-surgery, as they struggle to accurately track and adapt to rapid network state changes, leading to unpredictable delays and increased jitter.
Innovation Solution
A computer-implemented method and system that uses convex optimization to distribute traffic across multiple routes, accounting for correlated delays and jitter, to achieve lower end-to-end jitter by trading slightly higher average delay, inspired by portfolio selection techniques to minimize variance for a desired expected delay.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If packets are transmitted through a single network path, then the transmission is simple and direct, but the delay jitter increases due to dynamic network conditions and resource competition
Solution Approach 1:
The patent segments the traffic flow into multiple parallel paths between source and destination. Instead of using a single routing path, the system divides the packet stream across multiple diverse routes, where each path experiences independent network conditions. This segmentation allows the system to exploit statistical multiplexing gains and reduce overall delay jitter by averaging out the variations across paths.
Solution Approach 2:
The patent transitions from one-dimensional single-path routing to multi-dimensional parallel path routing. By adding the dimension of path diversity, the system creates a routing space where packets can be distributed across multiple routes simultaneously. This dimensional expansion enables the system to navigate around congestion and dynamic network conditions that would affect a single path.
2Reliability
If multiple routes are used for traffic distribution, then the end-to-end jitter is reduced, but the average delay increases slightly
Solution Approach 1:
The patent changes the routing parameter from single-path to multi-path distribution, fundamentally altering how traffic is routed. By adjusting the traffic distribution parameters across multiple paths and using optimization techniques, the system achieves a parameter configuration that minimizes jitter while keeping average delay acceptable. The convex optimization framework allows finding the optimal balance between these competing parameters.
Solution Approach 2:
The patent employs prior cushioning by pre-computing optimal traffic distribution parameters across multiple paths before actual traffic transmission. The convex optimization is performed in advance to determine the best routing parameters that will minimize jitter. This preparatory optimization cushions against the inevitable variations in network conditions by having pre-established optimal distribution strategies.
3Reliability
If network buffers are increased to combat jitter, then the delay jitter is reduced, but the cost increases and excessive delay may occur
Solution Approach 1:
The patent introduces multiple routing paths as an intermediary mechanism between the packet stream and the network core. Instead of relying solely on buffering at queues to manage jitter, the system uses path diversity as an intermediary that distributes packets across different network routes. This intermediary approach reduces the burden on network buffers and prevents excessive delay that would result from large buffer sizes.
Solution Approach 2:
The patent substitutes the mechanical approach of increasing physical buffer capacity with a computational approach of optimizing traffic distribution across multiple paths. Instead of adding more buffering infrastructure (mechanical solution), the system uses convex optimization algorithms to dynamically allocate traffic. This substitution replaces hardware expansion with intelligent software-based routing decisions.
Data Source
AI summary
Systems and methods reduce delivery delay jitter in a delivery network. A processor identifies a plurality of routes between an originating node and a destination node. Each route has a respective mean delivery delay time and a respective delivery delay jitter. The processor solves a convex optimization problem for a plurality of values of delivery delay, thereby yielding a plurality of solutions. Each solution represents a corresponding allocation of traffic among the plurality of routes. Each allocation of traffic has a corresponding mean delivery delay time and a corresponding mean delivery delay jitter. The processor selects, from the plurality of solutions, a selected solution, which has a mean delivery delay jitter less than the delivery delay jitter of any route of the plurality of routes. Traffic is automatically distributed over the plurality of routes according to the allocation of traffic that corresponds to the selected solution.