ML Network Assurance Service for RF Anomaly Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Complex computer networks face challenges in identifying and addressing radio-frequency (RF) related anomalies, such as roaming failures and throughput issues, which impact user experience due to the difficulty in pinpointing and mitigating persistent RF problems without disrupting backend systems.
Innovation Solution
A machine learning-based network assurance service that classifies wireless network anomalies as radio-related or backend-related, uses machine learning models to assess RF issues, and initiates changes by moving clients from problematic access points to new or existing access points strategically placed to alleviate these issues.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional network monitoring methods are used to track network metrics, then basic network health can be assessed, but RF-related anomalies cannot be effectively identified and mitigated
Solution Approach 1:
The patent introduces machine learning models as intermediary components between raw network data and anomaly detection. These models process captured network information and client feedback data to identify RF-related anomalies that traditional monitoring cannot detect, effectively bridging the gap between basic metric tracking and sophisticated anomaly identification
Solution Approach 2:
The patent replaces traditional mechanical/network monitoring approaches with machine learning-based detection systems. Instead of relying on conventional threshold-based monitoring, the system uses trained models to analyze patterns in network metrics and client feedback, substituting automated intelligent analysis for traditional manual or rule-based detection methods
2Reliability
If access points are added to mitigate RF issues, then user experience improves, but network complexity increases
Solution Approach 1:
The patent implements preliminary action by using machine learning models to predict RF anomalies and identify optimal access point placement locations before deploying new hardware. The system analyzes historical data and client feedback to determine where access points should be strategically positioned to prevent problems rather than reactively adding them throughout the network
Solution Approach 2:
The patent applies local quality by strategically placing access points in specific locations identified by machine learning analysis rather than uniform distribution. The system determines precise locations where RF anomalies occur and positions access points to target those specific problem areas, ensuring each access point serves a localized function optimized for its particular environment
3Measurement precision
If machine learning models are used to classify anomalies, then RF-related issues are accurately identified, but processing time and computational resources increase
Solution Approach 1:
The patent applies partial action by having machine learning models focus specifically on classifying RF-related anomalies rather than analyzing all possible network issues. The system processes only the subset of data relevant to RF problems identified through client feedback and specific network metrics, reducing overall computational burden while maintaining high accuracy for the target anomaly type
Solution Approach 2:
The patent implements preliminary action through offline training of machine learning models using historical network data. Once trained, the models can quickly classify new anomalies in real-time operations. The computationally intensive training phase is performed in advance, allowing for rapid inference during actual network monitoring without significant processing delays
Data Source
AI summary
In one embodiment, a network assurance service that monitors a wireless network identifies a set of wireless network anomalies detected in the wireless network that are associated with a set of one or more network measurements. The network assurance service classifies the set of wireless anomalies as radio-related or backend-related. The network assurance service, when the set of wireless anomalies are classified as radio-related, determines that the wireless anomalies are recurring for a particular wireless access point in the network. The network assurance service initiates a change to the wireless network in part to move clients in the wireless network from the particular wireless access point to another wireless access point in the network.