Application Telemetry SDK for Cognitive Network QoE
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing network systems fail to accurately predict the quality of experience (QoE) of online applications due to reliance on service level agreement (SLA) thresholds, which do not account for complex impairments or user and application behaviors, and lack a mechanism to unify the application and network layers for feedback.
Innovation Solution
A software development kit (SDK) enables bidirectional communication between applications and cognitive networks, allowing applications to provide telemetry data to a cognitive network service that uses machine learning models to predict QoE and recommend configuration changes for optimizing network performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If SLA thresholds are used as a proxy for QoE, then network control is simplified, but accurate prediction of true QoE is lost
Solution Approach 1:
The patent implements a feedback mechanism where applications provide telemetry data about their actual performance and user experience to the network. This feedback loop allows the network to learn from real QoE measurements and adjust its control decisions, resolving the contradiction by maintaining simplicity while improving accuracy through continuous learning from actual user experience data.
Solution Approach 2:
The patent replaces the mechanical/protocol-based SLA threshold system with a machine learning-based prediction system. Instead of relying on rigid SLA thresholds, the system uses ML models that process application telemetry and network data to predict true QoE, achieving both simplicity and accuracy through intelligent automation.
2Device complexity
If strict layering of OSI model is maintained, then network architecture is simplified, but application-layer feedback to network is eliminated
Solution Approach 1:
The patent introduces an intermediary mechanism (application telemetry interface) that enables information exchange between the application layer and network layer without violating OSI layering. This intermediary allows applications to send performance feedback to the network while maintaining architectural simplicity and layer boundaries.
Solution Approach 2:
The patent creates a universal telemetry interface that serves multiple functions: it collects application performance data, transmits it to the network, and enables the network to make informed control decisions. This multi-functional interface maintains architectural simplicity while enabling rich information flow between layers.
3Measurement precision
If application telemetry collection is added, then true QoE prediction is improved, but system complexity increases
Solution Approach 1:
The patent implements self-service by having applications automatically collect and report their own telemetry data about performance and user experience. This eliminates the need for complex external monitoring systems, as applications themselves provide the necessary feedback, improving QoE prediction without significantly increasing overall system complexity.
Solution Approach 2:
The patent changes the approach from static SLA threshold parameters to dynamic application telemetry parameters. By collecting actual runtime parameters from applications (such as user interactions, performance metrics), the system achieves accurate QoE prediction through parameter transformation rather than complexity increase.
Data Source
AI summary
In one embodiment, a device provides a software development kit that includes a set of functions for inclusion in an application developed using the software development kit to communicate with a cognitive network service in a network. The cognitive network service receives application telemetry data from the application sent via the set of functions from the software development kit. The cognitive network service uses the application telemetry data from the application and network telemetry from the network as input to a prediction model to predict a quality of experience metric for the application. The cognitive network service provides, based on the quality of experience metric predicted by the prediction model, a configuration change recommendation to the application via the set of functions.