Network Capacity Planning Using ML for Control Plane Bottlenecks

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current network capacity planning tools largely ignore the control plane, leading to inefficiencies and performance degradation due to the complexity of modeling control plane resource utilization, especially in wireless networks where signaling traffic characteristics differ from normal data traffic.

Innovation Solution

The implementation of a comprehensive network planning system that incorporates both data plane and control plane utilization using machine learning algorithms, specifically selecting key features to accurately predict control plane resource utilization and determine when new capacity is needed.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If conventional network capacity planning tools focus only on data plane resource utilization, then data plane resources can be efficiently allocated, but control plane bottlenecks occur leading to network performance degradation

Engineering Contradiction:

Improvedata plane resource allocation efficiencyVSAvoidnetwork performance

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The patent combines data plane and control plane resource utilization modeling into a unified network capacity planning framework. The machine learning model simultaneously processes both data plane traffic characteristics and control plane signaling patterns to generate comprehensive capacity recommendations, preventing control plane bottlenecks while maintaining data plane efficiency.

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The patent introduces machine learning algorithms as an intermediary that bridges the gap between conventional data plane-focused planning tools and the need for control plane awareness. The ML model translates complex control plane signaling patterns into actionable capacity planning insights without requiring fundamental changes to existing planning tool architectures.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Device complexity

If control plane resource utilization is modeled using conventional methods, then modeling complexity is reduced, but accuracy of control plane capacity planning deteriorates due to unique signaling traffic characteristics

Engineering Contradiction:

Improvemodeling complexityVSAvoidcontrol plane resource utilization accuracy

Core Design Contradiction:

Device complexityVSMeasurement precision

Solution Approach 1:

The patent replaces conventional analytical modeling methods with machine learning-based modeling for control plane resource utilization. The ML approach automatically learns complex patterns in signaling traffic without requiring explicit mathematical formulations, achieving high accuracy while managing complexity through automated feature engineering and model selection.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent transforms control plane modeling by changing from traditional parameters (based on assumptions about signaling patterns) to data-driven parameters extracted directly from network measurements. The ML model uses input features such as signaling message rates, message types, and temporal patterns to accurately predict control plane resource utilization under varying network conditions.

Inventive Principle:

Principle #35Parameter changes

3Measurement precision

If machine learning algorithms are used to model both data plane and control plane utilization, then network capacity planning accuracy improves, but system complexity increases

Engineering Contradiction:

Improvenetwork capacity planning accuracyVSAvoidsystem complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the network capacity planning problem into distinct data plane and control plane components, each with specialized feature sets and modeling approaches. This segmentation allows the system to handle complexity in a modular fashion, processing different traffic types through dedicated analysis pipelines before integrating results for comprehensive capacity recommendations.

Inventive Principle:

Principle #1Segmentation

Data Source

PatentUS20230284082A1Network capacity planning considering the control plane bottleneck

Publication Date: 2023.09.07 AT&T INTELLECTUAL PROPERTY I L P

AI summary

Aspects of the subject disclosure may include, for example, obtaining first information indicative of data plane utilization, wherein the data plane is associated with a wireless communications network; obtaining second information indicative of control plane utilization, wherein the control plane is associated with the wireless communications network; applying the first information and the second information to one or more machine learning algorithms; and generating via the one or more machine learning algorithms one or more outputs, wherein the one or more outputs indicates whether one or more network resources should be added to improve the data plane utilization. Other embodiments are disclosed.