Global Quality

Description

Global Quality (GQ) is a standardized, objective perceptual quality metric specified in 3GPP TS 26.935. It provides a single numerical score that predicts the overall quality of a conversational speech or video service as perceived by the end-user. Unlike metrics that measure quality in isolated network sections (e.g., packet loss in the radio link), GQ aims to model the cumulative effect of all quality impairments from one user's mouth to the other user's ear (or camera to display). It is an instrumental, computational metric, meaning it is calculated algorithmically from measurable network and terminal parameters rather than through subjective human testing for each call.

The calculation of GQ is based on the E-model methodology, extended and adapted for full conversational scenarios. It integrates a wide range of input parameters that contribute to quality degradation. These include codec-specific impairments (e.g., from AMR, EVS), delay impairments (mouth-to-ear delay, including codec processing, network jitter, and playout buffer delays), equipment impairments (from terminals like handsets), and various packet-layer impairments prevalent in IP networks. The latter encompasses packet loss (both random and bursty), packet delay variation (jitter), and the effects of error concealment algorithms used by the codec when packets are lost or late.

The output is a scalar value, often mapped to a Mean Opinion Score (MOS) scale (e.g., from 1 to 5), providing an intuitive estimate of user satisfaction. The GQ model is designed to be applicable for planning purposes, where network engineers can input expected parameter ranges to predict service quality, and for monitoring/assurance, where actual measured parameters from a live network can be fed into the model to assess the quality of ongoing calls. Its end-to-end nature makes it a valuable tool for service providers to guarantee a consistent quality level across heterogeneous network domains (e.g., radio access, core, transit networks).

Purpose & Motivation

GQ was created to address the challenge of managing and predicting quality in increasingly complex, packet-switched, and multi-vendor communication networks. Prior to its standardization, quality assessment was often fragmented. Network operators might monitor radio link quality, core network latency, and packet loss separately, but lacked a unified metric that could accurately predict the final user experience for a conversational service. This made end-to-end service level agreement (SLA) management and troubleshooting difficult.

With the transition from circuit-switched voice (where quality was relatively stable) to Voice over IP (VoIP) and later Voice over LTE (VoLTE) and Voice over NR (VoNR), new impairment factors like variable delay, jitter, and packet loss became dominant. The purpose of GQ, introduced in Rel-8 alongside the early development of IMS-based voice, was to provide a standardized, objective tool that could model these combined effects. It allows operators to plan networks (dimensioning codec selection, delay budgets, loss targets) to achieve a desired quality level and to monitor live networks to identify if the actual quality deviates from predictions, prompting investigation into which network segment is causing the degradation.

Key Features

• End-to-end quality prediction integrating all conversational service impairments
• Based on the extended E-model methodology for computational efficiency
• Considers codec impairment, delay, equipment quality, and packet loss effects
• Outputs a scalar value typically mapped to a Mean Opinion Score (MOS) scale
• Applicable for both network planning/design and live service monitoring
• Standardized algorithm ensures consistent measurement across different vendors and networks

Evolution Across Releases

Defining Specifications