Video over IP Measurement techniques

Video over IP quality measurement is important for the growth of video on the internet.

Quality measurement algorithms are divided into two main types: Subjective and objective measurements. Subjective measurements results are not preformed by monitoring devices only by humens Therefore this discussion about monitoring devices algorithms will focus on objective video quality monitoring methods

The objective quality measurements types are further segmented into several algorithm types:
1 Payload based
2 Codec Aware Packet based
3 Codec independent packet based
4 Network parameters

Payload based methods requires a comparison of the video to the source. Those methods are used for lab measurements in which the source stream is available.
The most common Payload Based video quality evaluation methods are PSNR and ITU-T J.144.
In a monitoring (in-service) testing scenario, the source stream is not available; Hence payload based measurements are not used in monitoring equipment.

Monitoring equipment must utilize zero refrence algorithms which could be one or more of the three methods network monitoring, packet based codec aware and packet based codec independent methods.
For regular data oriented communication networks, the raw packet loss and jitter parameters suffice.
In media oriented networks, raw network parameters are not enough and there is a need to measure the effect of the network impairments on the perceived media quality.

The two quality monitoring methods employed in media oriented packet networks are the codec aware and codec unaware packet based algorithms.
The codec unaware algorithms like MDI provide a quality index based only on network parameters without considering the codec type. This type of measurement will provide the same quality index for all codecs, although each codec degrades differently in the presence of network impairments.

The most used methods for media quality in IP networks today is the codec aware packet based method which considers the codec type when grading the quality based on network parameters. For example, while MDI will provide one media quality index for any given packet loss, codec aware methods will consider the codec type.
degradation curve described in the graph above and provide the quality score based on the codec type.

The drawing above presents a rough taxonomy of the monitoring algorithms reviewed and their uses.

Web 2.0 Video Distribution

It often seems like there isn't anything new in the field of Video Content distribution, as it has been controlled and dominated for a long time by the Peer-to-Peer networks:

• Most of us already have P2P software client for downloading files and Video content, such as eMule and Bittorrent
• Over 80% of internet traffic today is related to file sharing and transferring, mostly of Video files.

So what can be new?
The advancement and innovation in content distribution fields, contradict one another:

• Content P2P network filtering and blocking
• Acceleration and caching of content traffic

P2P Filtering
Many companies today have reached the conclusion that P2P trafficking will fully utilized their enterprise communication capabilities that is why products that block or limit P2P traffic have been developed. The P2P filtering products are designed for the ISP and the Enterprise markets.
Among the companies that offer P2P filtering solutions are: Ellcoya, Sandvine and Allot (Israeli Company).

In the attaché Diagram (by Snadvine) you can see the influence of the products in minimizing and lowering the bandwidth of P2P networks.

Acceleration and Caching of content

File download Acceleration
In the field of Video content acceleration we can see a great success for Limelight Networks who provide caching services.
Limelight had done many deals with User Generated Content (UGC) distributors, a recent one which was announced this month was with UGC leader Metacafe.
Limelight service assists content distributors that wish to accelerate file download time and so improve the users quality of experience. The UGC files are usually not streamed but simply downloaded over HTTP. By utilizing "Progressive Download" technology, the flash player is able to play those files before they are completely downloaded. This technology saves money to content distributors since streaming technology costs much more than simple download technology.
These services still do not solve Video Streaming issues.

Video Streaming acceleration
The standard architecture of video streaming is a Client-Server architecture in which one server provides a limited number of concurrent clients. The disadvantages of this type of solution are the high price of servers and the need to increase the number of servers as the number of users grows.
There are two possible directions in which P2P could evolve in order to provide better Quality-of-Experience (QoE) to the user.

• Accelerating current P2P file download – a file will be downloaded in a matter of minutes instead of in a few days. This is fairly easy and requires only changing the download queuing mechanism.
• Video streaming – Creating a streaming support mechanism in P2P network which will enable streaming and not just file download. This streaming method could later replace the traditional Client-Server streaming architecture.

The use of P2P networks for Streaming is not a new concept; vTrails has developed a streaming technology based on P2P networks for radio stations. vTrails solution never picked up, probably due too bad timing.

Today however, the P2P market is bubbling. The players today are:
Bittorrent – the wild and common P2P protocol that was used in the past for the purposes of file sharing (partly illegal) is trying to change its image and is signing distribution deals with large media companies. In addition it seems that BitTorrent is creating a new paid service that accelerates the downloading time.
Joost – Joost is a new venture by the creators of Skype and Kaaza that creates new opportunities for media consumption over P2P networks. More updates on Joost will be found in following articles.
Xunlei – a Chinese P2P network with 100 million users. Does Google have planes to use it for the purpose of multimedia streaming?
OverSi – an Israeli start-up company who provides acceleration and caching solutions for P2P networks. The serious downloader's among us have already identified the advantage in hooking up with Barak internet provider that uses OverSi caching technology to accelerate and shorten downloading time for there perscribers and to save on expensive bandwidth.
OverSi caching solutions enable to accelerate performances for all P2P networks and allow streaming as well.

Analysts predict that these solutions will be popular for Video on Demand (VOD) services of IPTV providers, and will allow acceleration of downloading movies and content in a separate payment.

In following articles we will continue to review the field of personalization and advertisement in Video, an up and coming field that is attracting large companies such as Google.

Web 2.0 Billing trends

It seems that there are three major pricing alternatives for services:

• Per interaction charge – this type of pricing is common in the telecom market (PSTN or mobile), companies like Amdocs offer complex billing solutions which log every phone call and feeds it to complex personalized billing options.


• Flat rate – this type of billing is common for basic continues services and subscriptions like cable television


• Advertisement based service – this type of services generate revenues by exposing the users to advertisement.

With the proliferation of Web 2.0 philosophies, non-interaction based pricing and billing schemes are gaining popularity among service providers:
VoIP service providers like Vonage provide fixed monthly rates
In a recent press release, Skype announced a new yearly flat pricing of 30$ per year on landlines calls.

As communication services pricing declines we will view a trend of using more and more fixed and advertisement based revenue generation methods and less per-interaction methods.

In multimedia billing, the same revenue generation methods exist although the trend is not a clear trend. The most common revenue source for content owners till recent years was selling a complete album or a video.
The revenue generation models described previously also exist for media distribution:
Per interaction (per song) – This model is used by companies like Apple and their iTune music store which sells songs for under 1$.
Fixed price service – This model is used by Real- Networks' Rhapsody music service which enables subscribers to consume multimedia at a monthly flat rate.
Advertisement – services like Pandora uses advertisement and referral plans to iTunes and Amazon to generate revenues
According to IceRocket comparison(on the right side graph) , iTune service is much more popular than Rhapsody.

The sector that will interest many entrepreneurs is mobile billing for both VoIP and media content.
For Mobile VoIP, the trend is clear; pricing will be reduced to air time by aggressive competitors of current mobile companies.
Content billing is more complex: Content owners would like to see revenues but in order to provide it, mobile companies will be required to implement complex content management, billing and DRM systems.
The current challenge for DRM, billing and content management vendors is to create converged low cost systems that could enable new Web 2.0 oriented business models.