Global Sources
EE Times-Asia
Stay in touch with EE Times Asia
EE Times-Asia > Manufacturing/Packaging

Measuring IPTV QoS performance

Posted: 16 Aug 2006 ?? ?Print Version ?Bookmark and Share

Keywords:Yves Cognet? QoSmetrics? communications? spotlight? video over ip?

Quality-of-experience (QoE)!or QoS as experienced by the end-user!is the only way to understand what makes a customer unhappy and how QoS relates to network impairments. It's the only way a carrier can understand how to improve its service.

A suitable method is one that can pinpoint a user's problem from the living room. Somehow, a service provider has to be able to look directly into the home for problems. Thus, the ideal STB in a living room should be able to report to a service provider problems experienced in the household.

Most researches on video quality assessment have been focused on a dual-ended model. The dual-ended model requires a comparison with the impaired to the video source. Impaired video is the result of different and complex processes, such as video digitalization and compression, streaming, network transport and decompression.

This dual-ended model fits the needs of most engineers working in a development lab, where the source video and the impaired video can be brought in. This model is impractical for settings with the source and impaired video miles away from each other. Moreover, it doesn't scale well.

MPQM model
V-factor patent-pending technology, designed specifically for iptv, leverages Moving Picture Quality Metrics (MPQM) research that several labs have developed over the last 10 years.

MPQM is a single-ended model that assesses video quality in real-time. The MPQM model doesn't require any well-known source and scores the video quality the same way the E-model assesses VoIP quality. MPQM is based on the human vision system and considers how the degradation of the source through the whole process!compression, transmission, decompression!affects image quality as perceived by the eyes and the brain, not just by inferences made from core network impairments. Most artifacts that downgrade video quality come from the compression itself and from networks impairments such as jitter and packet loss.

V-factor is an implementation of the MPQM model. V-factor provides the video quality score, and some extra information needed for monitoring and diagnosing the root of problems. Information provided by V-factor includes the well-known transport stream key parameters (KPI) defined in ETSI TR 101290 and parameters related to the network layer's performance as defined by ITU Y1540/1541 or IETF RFC2330.

Video quality
For MPQM to work, it should be fed with several parameters!packet loss rate probability, the image's entropy (quantity of information carried by an image), network jitter, network losses and the program clock reference (PCR) jitter.

The packet loss probability is deducted from the network and streaming impairments, and the depth of the buffer that will be used for the image decompression. The image entropy considers several parameters, such as the nature of the video frame that will be affected or are affected by jitter or loss (the loss of I, B and P frames doesn't have the same impact on video quality as perceived by an end-user).

It also takes into account the relative sizes of frames and the size of the group of pictures (GOP). The loss of an I frame belonging to a video stream that has been compressed with a GOP equal to 1 doesn't have the same impact if this video stream has been compressed with a GOP of 12.

The GOP describes the sequences of I, B and P frames. The partial or total loss of an I frame makes the reconstitution of the next related B and P frames very difficult. From an information point of view, an I frame has a greater entropy than a B or a P frame.

With such a model, an end-to-end monitoring system can be easily deployed for either unicast video streams such as video-on-demand or multicast of video streams such as IPTV. By deploying non-intrusive hardware probes at different locations on the network, an IPTV service provider can easily monitor the performance of its services across the whole chain. Even the video streamed by a content provider can be monitored over an asynchronous serial interface, for example. By adding a V-factor agent into an STB, supporting users experiencing problems in the home can be easily diagnosed and fixed.

V-factor into STB
V-factor can provide self-defined KPIs and service-level statement monitoring for IPTV service providers. Moreover, it can provide advanced alarm systems that give real-time information for the video stream (image freezing, V-factor below a defined threshold) and the audio stream.

For the MPQM model to work, it needs to be fed with typical parameters, including the size of A/V buffers. The size of these buffers will help the model compute how much jitter the STB can handle for A/V parts of the streams. It is also important to tell the model whether or not some packet loss concealment techniques such as FEC are being implemented.

The model then needs to access some real-time data. The real-time data are related to the IP layer, the MPEG-2 TS layer or the real-time transport protocol (RTP) layer when RTP protocol is being used instead. There are different ways for collecting real-time data from the STB. A V-factor library is forthcoming with straightforward APIs that can be viewed as a wrapper (an application will need to forward all IP packets to the API) or as a pass-through module sitting between the network driver and the upper layers.

Depending on the software architecture of the STB, the V-factor library can also be fed directly from statistics delivered by the TV core or even collected through local adaptation. A lot of flexibility is provided to match the different requirements of most STB vendors.

- Yves Cognet
Chairperson, QoSmetrics

Article Comments - Measuring IPTV QoS performance
*? You can enter [0] more charecters.
*Verify code:


Visit Asia Webinars to learn about the latest in technology and get practical design tips.

Back to Top