Cellular Voice Quality Control Using Speech-Silence Packet Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for assessing and improving voice quality in mobile communication networks are inadequate, as they either rely on simulations that do not reflect real-world conditions or costly packet inspection, leading to inefficient resource allocation and incorrect quality evaluations due to the lack of separation between speech and silence frames, particularly in congested networks.
Innovation Solution
A method and system that utilize a machine-learning model to analyze voice packet data, distinguishing between speech and silence frames, and dynamically adjust base-station configurations to improve perceived voice quality by modifying transmission parameters based on real-time metrics, including packet loss and delay variance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If packet-based voice service is provided using QCI1 bearer in LTE/MCS networks, then voice service can be carried over the network, but packet loss and delay increase under heavy network load due to lack of separation between speech and silence frames
Solution Approach 1:
The patent segments the voice packet stream into speech frames and silence indication frames (SID frames) by analyzing packet lengths and patterns. This segmentation allows the system to distinguish between actual speech content and silence periods, enabling selective resource allocation that improves both voice quality reliability and network resource efficiency under heavy load conditions
2Measurement precision
If existing voice quality assessment methods are used, then voice quality can be measured, but the methods are either based on simulations that do not reflect real networks or require expensive packet inspection solutions
Solution Approach 1:
The patent enables the network itself to perform voice quality assessment by analyzing packet length patterns and SID frame presence in the existing voice traffic. This self-service approach eliminates the need for external probe systems or complex packet inspection equipment, providing accurate real-network measurements through standard network processing functions
3Measurement precision
If QCI1 bearer packet loss and delay metrics are measured without separating SID frames, then voice quality can be evaluated, but incorrect evaluation results in excessive allocation of expensive radio network resources
Solution Approach 1:
The patent extracts and removes SID frames from the voice packet analysis, treating them separately from speech frames. By excluding SID frames from packet loss and delay calculations, the system obtains accurate metrics that reflect only actual speech quality, preventing unnecessary resource allocation to correct silence period issues
4Measurement precision
If deep packet inspection is deployed to identify voice codec type, then accurate voice quality assessment is possible, but the cost and complexity of the system increases
Solution Approach 1:
The patent uses simple, lightweight packet length pattern recognition instead of expensive deep packet inspection. By analyzing the lengths of received packets and their patterns over time, the system can identify voice codec types and detect SID frames using minimal processing resources, avoiding the need for complex inspection infrastructure
Data Source
AI summary
Voice data packets are analyzed to determine the audio codec mode and one or more data metrics associated with the voice data packets in a cellular voice call that includes a cellular mobile device in cellular communication with a base station. The audio codec mode and the data metric(s) is/are provided to a trained machine-learning model to determine a predicted mean-opinion score for the voice call. One or more network parameters are updated on the base station for the cellular user equipment device to improve the data metric(s) and the perceived quality of the cellular voice call.