Session Admission Control Using Predictive Tunnel Performance Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current communication networks face challenges in efficiently managing session admissions due to limitations in predicting network performance and resource availability, leading to potential adverse impacts on existing sessions and quality of service.
Innovation Solution
A method involving the creation of a datastore with historical parameter values for tunnels, using predictive models to assess expected performance and impact on existing sessions, and evaluating historical performance to determine if session admissions will have no adverse effects, with a network gatekeeper configured to make informed decisions based on these analyses.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional session admission control is used based on current network state, then admission decisions are made quickly, but the accuracy of performance prediction is poor leading to potential QoS degradation
Solution Approach 1:
The system performs preliminary actions by collecting historical performance data and building predictive models before actual session admission decisions are needed. The datastore is pre-populated with performance metrics from multiple time intervals, and predictive models are pre-computed so that when a session admission request arrives, the system can quickly query pre-analyzed data rather than performing complex real-time predictions
Solution Approach 2:
The system dynamically adapts the prediction process by using multiple time intervals of historical data to capture changing network conditions. The predictive models are updated over time as new historical data becomes available, allowing the system to adapt to evolving network patterns while maintaining quick decision-making capability through efficient data structures and query mechanisms
2Measurement precision
If historical performance data for multiple time intervals is collected and predictive models are built, then performance prediction accuracy is improved, but the complexity of the admission control system increases
Solution Approach 1:
The system creates simplified copies of complex network performance data by storing aggregated metrics in a structured datastore that captures essential patterns from historical data. Instead of maintaining and processing complete raw datasets, the system uses representative samples and summary statistics that preserve predictive power while reducing complexity
Solution Approach 2:
The system transforms complex multi-dimensional historical performance data into simplified predictive parameters and models that capture the essential behavior patterns. By changing the representation from raw data to processed metrics and predictive functions, the system maintains high prediction accuracy while reducing the complexity of data storage and processing requirements
3Productivity
If multiple tunnels are evaluated for session admission, then resource allocation optimization is improved, but the computational overhead and processing time increase
Solution Approach 1:
The system performs preliminary evaluation of tunnel performance characteristics and pre-ranks tunnels based on their historical performance and current capacity. This preliminary sorting and pre-analysis allows the admission control to quickly evaluate tunnels in order of likelihood to succeed, avoiding exhaustive evaluation of all tunnels and reducing processing time while still optimizing resource allocation
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A session admission process is provided which identifies the weakest link in a route between a first node and a second node and determines if the route is able to cope if the session is admitted. The suitability of a link is determined on the basis of: historical link performance; the predicted future performance of the link; and the predicted future demands on the link from other sessions supported by that link.