HOME > research > Service quality assessment/management > Introduction to research themes

Service quality assessment/management

Models to evaluate the quality experienced by users

To provide comfortable communication services on an economical network, quality assessment models for quantitatively evaluating the service quality experienced by users are essential. Here, targeting communication services that use audiovisual media, we are trying to establish methodologies for quantitatively and efficiently evaluating the user's quality of experience (QoE), and methodologies for QoE design and management.

The mission of this research is to contribute to maintaining and improving the service quality of the NTT Group by using these media quality assessment, design and management models. Another important mission is to provide common concepts measures, and methodologies to communications carriers and the vendors of communication equipment through international standardization activities.

There are currently three research activities relating to IPTV which is one of the most important services over NGN.

[Research themes]

Research theme 1

Objective media-layer video assessment models

Objective media-layer video assessment models estimate subjective quality by using video media signals (pixel signals) to quantify the degradation of quality. NTT's work on media-layer objective assessment models includes the development of a full reference (FR) model that uses the pixel signals of the original video and degraded video. A promising application of this technique is in-service management for purposes such as optimizing codec parameters, ascertaining the true service quality, and automatically performing pre-delivery content quality checks relating to coding, transcoding and the like. (see Fig. 1)

By applying these models to the assessment and management of quality in video delivery services, we can expect the following benefits. (1) It will be possible to improve customer satisfaction (CS) by promptly isolating and dealing with claims from clients. (2) By monitoring the clients' perception of quality, it is possible to quickly reduce the spread of quality degradation effects to clients. (3) Currently, the quality of content is checked visually before it is delivered. Automating this process makes it possible to reduce the workload of service providers.

In August 2008, NTT Network Technology Laboratories contributed to ITU-T recommendation J.247 regarding the standardization of objective assessment methods and models for video delivery services using terminals such as PCs and PDAs. In the future, we plan to continue with international standardization by expanding this model to HDTV content.

Fig. 1: Application concept

Fig. 1: Application concept

Research theme 2

Objective packet layer quality assessment models

Packet-layer quality assessment models are used for estimating quality of experience (QoE),which is what end-user's perceive, using transmitted packet headers, such as Internet protocol (IP), user datagram protocol (UDP), real-time transport protocol (RTP), transport stream (TS), and packetized elementary stream (PES) headers that exclude media-related bitstream information, with low computational load. These models first extract quality parameters (e.g., bit rate and packet loss) affecting media quality from the transmitted packet headers. Then, they output the QoE using media quality characteristics between quality parameters and media quality. (see Fig. 2).

These models can be incorporated into the client terminal, such as a home gateway or set-top box, because these model's computational load is very low. In addition, they can be used for QoE monitoring even when media-related information is encrypted to protect the copyright of media content. Although application scenarios of a bitstream-layer model introduced in Research theme 3 are similar to packet-layer models, the use of these models should be determined based on the computational power of the client terminal and the encryption of bitstream information.

Fig. 2: Packet-layer quality assessment model

Fig. 2: Packet-layer quality assessment model