Link Behavior Prediction for Dynamic Path Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current computer network technologies lack the ability to predict future link instability and dynamically adjust paths to prevent network traffic disruptions, leading to potential service delays and unnecessary network operations.
Innovation Solution
A method and system that uses a machine learning system, such as a deep learning system, to predict future link metrics and generate dynamic thresholds, allowing for the exclusion of anomalous links from path computation and adjustment of link weights to preemptively reroute traffic around potentially unstable links.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional reactive path selection is used, then network operations are simpler, but network stability and reliability deteriorate due to inability to predict link instability
Solution Approach 1:
The system performs preliminary actions by collecting historical link metric data and training machine learning models in advance to predict future link stability. This allows the network to proactively identify potentially unstable links before actual failures occur, enabling preemptive path adjustments that improve reliability without requiring complex real-time intervention mechanisms.
Solution Approach 2:
The patent replaces traditional mechanical/reactive path selection mechanisms with an intelligent system based on machine learning models. These models analyze historical data patterns and generate predictions about future link behavior, substituting simple threshold-based or static routing algorithms with adaptive, data-driven decision-making that enhances reliability while managing complexity through automation.
2Measurement precision
If machine learning prediction system is implemented, then link instability prediction capability is improved, but computational complexity and processing requirements increase
Solution Approach 1:
The system segments the complex prediction task by implementing separate machine learning models for different link metrics (e.g., packet loss, latency, bandwidth). Each model is trained on specific historical data patterns and can be independently optimized and updated. This modular approach improves prediction accuracy for individual metrics while managing overall computational complexity through distributed, specialized processing.
Solution Approach 2:
The system applies partial action by implementing machine learning predictions selectively for critical links or during periods of high network instability, rather than uniformly across all network operations. This allows the system to achieve high prediction accuracy where needed while reducing computational overhead during normal operating conditions, effectively balancing precision requirements with processing complexity.
3Loss of time
If dynamic path adjustment is performed, then service delays are reduced, but network operation complexity increases
Solution Approach 1:
The system implements continuous feedback loops where link metric data is collected, predicted using machine learning models, and used to dynamically adjust paths. The system monitors actual link performance and compares it with predictions, using this feedback to refine models and adjust paths in real-time. This automated feedback mechanism reduces service delays by rapidly responding to link instability while managing operational complexity through closed-loop control rather than manual intervention.
Solution Approach 2:
The network system performs self-service by automatically detecting link instability patterns, predicting future failures, and rerouting traffic without human intervention. The machine learning models continuously learn from historical data and autonomously make path selection decisions, reducing service delays through rapid automated response while simplifying operations by eliminating the need for manual path management in dynamic conditions.
Data Source
AI summary
Techniques are described for predicting future behavior of links in a network and generating dynamic thresholds for link metrics for use in path selection. In one example, a computing system receives historical values of a link metric for links of a network. The computing system executes a machine learning system which processes the historical values of the link metric to generate: (1) a predicted future value of the link metric for each link; and (2) a threshold for the link metric indicating whether the predicted future value for each link is anomalous. The computing system computes a path based on the predicted future values of the link metric and the threshold for the link metric. The computing system provisions the computed path, thereby enabling a network device to forward network traffic along the computed path.