Dynamic QoS Management via Performance Model Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current network management systems face challenges in providing scalable and adaptive Quality of Service (QoS) guarantees, particularly in Differentiated Services (DiffServ) frameworks, where traditional methods struggle to dynamically adjust to changing network conditions and user demands, leading to potential performance breaches and inefficient resource allocation.
Innovation Solution
The method involves creating and managing performance models based on historical data, using cluster analysis to identify prototype vectors and confidence intervals for transmission parameters like jitter, loss, and delay, which are then assigned to routes and used to accept or reject communication sessions, allowing for real-time adaptation of QoS guarantees and dynamic resource allocation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional DiffServ frameworks are used to provide QoS guarantees, then service classes can be defined with pre-agreed SLAs, but the system cannot dynamically adapt to changing network conditions and user demands
Solution Approach 1:
The patent implements dynamic QoS management by continuously monitoring network performance parameters (delay, jitter, loss) and adapting service class assignments in real-time. The system transitions from static pre-agreed SLAs to dynamic performance-based service differentiation, where performance models are updated based on current network conditions and traffic patterns.
Solution Approach 2:
The system employs self-service mechanisms through automated performance monitoring, model generation, and service class assignment. The network automatically identifies performance characteristics, generates appropriate performance models, and assigns traffic to suitable service classes without manual intervention, enabling adaptive QoS management.
2Measurement precision
If performance models are generated and assigned to routes based on historical data and cluster analysis, then precise and adaptive QoS management is achieved, but computational complexity and data processing requirements increase
Solution Approach 1:
The patent extracts key performance characteristics from historical network data by identifying and isolating dominant performance patterns through cluster analysis. Instead of processing all raw performance data, the system extracts representative performance models that capture essential QoS characteristics, reducing computational complexity while maintaining measurement precision.
Solution Approach 2:
The system transforms raw performance parameters (delay, jitter, loss measurements) into derived performance model parameters through statistical analysis and cluster analysis. This parameter transformation converts complex time-varying performance data into simplified performance models that can be efficiently used for service class assignment and QoS management decisions.
3Productivity
If the system dynamically assigns performance models to routes and adjusts service classes in real-time, then resource utilization improves, but the speed of decision-making and session admission may be affected
Solution Approach 1:
The system performs preliminary actions by pre-generating performance models from historical data and pre-establishing performance thresholds for different service classes. When new traffic sessions arrive, the system can quickly match incoming traffic against pre-computed performance models and thresholds, enabling fast admission decisions without real-time complex calculations, thus maintaining both high resource utilization and fast decision-making speed.
Data Source
Figure 1
AI summary
The present invention provides a method of operating a communications network such that the classes of service offered by a network operator will depend upon the underlying conditions in the network. A number of performance models, each of which is described by two vectors, is determined from historical network data. These performance models can be assigned to routes through the communications network, such that a request for a session can be made in accordance with the offered performance model for that route.