Saturday, October 10, 2009

IP backbone QOS

The ultimate objective of IP QOS is reliable, high-quality voice service, the kind that users expect from the PSTN.

It is hard to achieve the same level of QoS as in PSTN. The main QoS issues are speech quality, service availability and usability. Voice requires lower delay, jitter and packet loss where as Ordinary Data transfer can be delayed without affecting much to the client’s requirement.

To withstand to such needs a minimum level of QoS mechanism must be maintained.

Quality of Service Parameters

Delay : The amount of time taken by a packet to reach from the source to the destination.

Issues with Delay

Echo

Talk Overlap

Types of Delay in a VoIP Call:

Processing Delay

Network Delay

Jitter

Jitter is the variation in the time between packets arriving, caused by network congestion or route changes.

Removing jitter requires collecting packets and holding them long enough to allow the slowest packets to arrive in time to be played in the correct sequence which causes additional delay.

Packet Loss

Packet Loss is losing packets along the data path which further degrades the VoIP Applications.

Voice packets are time-sensitive unlike Data packets. Therefore, retransmission is not a solution to this problem.

Latency

When the packet is being sent, there is a “latent” time till the computer that sent the packet waits for a confirmation that the packet has been received.

Latency causes packets to be lost. If a packet does not arrive in time to be replayed at the receiving end, the packet is dropped.

Latency does not distort the voice signal but delay can be very annoying, making normal conversation difficult for the speakers. The parties may start to talk at the same time or interrupt each other. As a result, the conversational quality is perceived as being poor.

Solutions for QoS Issues

For Voice communications over IP to become acceptable to the user, the delay needs to be less than a threshold value.

To ensure good quality of service, we can use Echo Cancellation and Packet Prioritization.

Use of service quality models that gives an estimate of perceptual quality rating using the networking parameters. Mean Opinion Score (MOS) is one of the quality rating on a scale of 1 (bad) to 5 (excellent)

Perceptual Assessment Model

PESQ is a model for perceptual evaluation of speech quality.

One novel feature of PESQ is the identification of transmission delays.

First PESQ adjusts the degraded version to be time aligned. Then it assesses the distortion between original and degraded sample.

Constant delays are not considered in the calculation of the MOS value, but delay variations change the rating of the speech quality. One should note that PESQ can only be applied for distortions which have been known before its development.

In PESQ the original and the degraded signals are mapped onto an internal representation using a perceptual model. The difference in this representation is used by a cognitive model to predict the perceived speech quality of the degraded signal. This perceived listening quality is expressed in terms of a mean opinion score (MOS), an average quality score over a large set of subjects.

Perceptual Evaluation Of Speech Quality Model

E-Model

“Mouth to ear” transmission quality measurement

Produces an “R” factor typically in the range 50 (bad) -95 (good)

R factor can be related to MOS score, Terminate Early (TME) etc.

ITU G.107/ G.108 and ETSI ETR250

R = Ro - Is - Id - Ie + A


Ro=Base R value

- Noise level

Is=Impairments that

occur simultaneously

with speech

- received speech level

- sidetone level

- quantization noise

Id=Impairments that

are delayed with

respect to speech

- talker echo

- listener echo

- round trip delay

Ie=Equipment Impairment

Factor

- CODEC

- multiplexing effects

A=Advantage factor


How to Calculate MOS?

For the codec used pick a corresponding default R value. The R-values for the most popular codec has been used. These are R=93 for the G.711 codec, R=80 for the G.729a codec and R=86 for iLBC codec.

From the Rider streaming test we can calculate the Jitter and Packet Loss.

From the Rider response time test, we can measure the network latency between the control and remote locations.

Add 10 ms if you are using G.729a and 5 ms for iLBC codec to account for computation time.

Add step no. 2, 3 and 4 to calculate the effective latency. (Latency plus jitter plus computation time.)

Adjust the R value down based on effective latency. Deduct 5 for a delay of 150 ms, 20 for a delay of 250 ms, 30 for a delay of 350 ms.

From the packet loss test in step no. 2, deduct the R value from consecutive packet losses using the table given below.


More explanation from ITU-T Spech coding and MOS related here

0 Comment :