Technical White Paper for QoS on SE2900 V300R002

HUAWEI SE2900 Session Border Controller V300R002

Technical White Paper for QoS

Issue 01

Date 2016-01-15

Huawei Technologies Co., Ltd.

Issue 01 (2016-01-15) Huawei Confidential i

Copyright © Huawei Technologies Co., Ltd. 2016. All rights reserved.

No part of this document may be reproduced or transmitted in any form or by any means without prior

written consent of Huawei Technologies Co., Ltd.

Trademarks and Permissions

and other Huawei trademarks are trademarks of Huawei Technologies Co., Ltd.

All other trademarks and trade names mentioned in this document are the property of their respective

holders.

Notice

The purchased products, services and features are stipulated by the contract made between Huawei and

the customer. All or part of the products, services and features described in this document may not be

within the purchase scope or the usage scope. Unless otherwise specified in the contract, all statements,

information, and recommendations in this document are provided "AS IS" without warranties, guarantees or

representations of any kind, either express or implied.

The information in this document is subject to change without notice. Every effort has been made in the

preparation of this document to ensure accuracy of the contents, but all statements, information, and

recommendations in this document do not constitute a warranty of any kind, express or implied.

Huawei Technologies Co., Ltd.

Address: Huawei Industrial Base

Bantian, Longgang

Shenzhen 518129

People's Republic of China

Website: http://www.huawei.com

Email: [email protected]

http://www.huawei.com/

mailto:[email protected]

HUAWEI SE2900 Session Border Controller


Issue 01 (2016-01-15) Huawei Confidential ii

About This Document

Overview This document is not for the communication with customers and therefore is not a

commitment to customers. The MO is responsible for the preparation of the materials used in

the communication with customers. The solutions described in this document may not be

implemented in the released version. For details about the schedule and implementation,

contact the SBC Product Management Department of Core Network Product Line.

Intended Audience This document is intended for the following engineers who participate in the life cycle of the

SE2900:

Marketing engineers

Technical sales engineers

Product management engineers



Issue 01 (2016-01-15) Huawei Confidential iii

Contents

About This Document ....................................................................................................................... ii

1 Overview .......................................................................................................................................... 1

1.1 Introduction .................................................................................................................................................................. 1

1.2 Value ............................................................................................................................................................................. 2

1.3 Service Models ............................................................................................................................................................. 2

2 Key Technologies ............................................................................................................................. 4

2.1 Anti-Packet Loss and Anti-Jitter Technologies ............................................................................................................. 4

2.1.1 Packet Loss Compensation ........................................................................................................................................ 4

2.1.2 RFC2198 Redundancy ............................................................................................................................................... 4

2.1.3 De-Jitter Buffer .......................................................................................................................................................... 5

2.2 Bandwidth Control Technologies.................................................................................................................................. 6

2.2.1 Policy Control (Rx) .................................................................................................................................................... 6

2.2.2 Call Admission Control ............................................................................................................................................. 6

2.2.3 Committed Access Rate ............................................................................................................................................. 7

2.2.4 Voice Activity Detection/Comfort Noise Generation ................................................................................................. 8

2.3 Forwarding Priority Control ......................................................................................................................................... 8

2.3.1 Differentiated Services Code Point ............................................................................................................................ 8

2.4 QoS Monitoring ............................................................................................................................................................ 9

2.4.1 Voice Quality Reporting ............................................................................................................................................ 9



Issue 01 (2016-01-15) Huawei Confidential iv

Technical White Paper for QoS on SE2900

V300R002

Key words: QoS

Abstract:

List of abbreviations:

Abbreviations Full spelling

QoS Quality of Service

CAR Committed Access Rate

MOS Mean Opinion Score


Technical White Paper for QoS 1 Overview

Issue 01 (2016-01-15) Huawei Confidential 1

1 Overview

1.1 Introduction

Quality of Service (QoS) refers to the capability of the service provider to meet needs of its

customers. For network services, QoS involves concepts such as the bandwidth, latency, jitter,

and packet loss rate. QoS can be used to manage or avoid network congestions and reduce the

packet loss rate by providing differentiated quality for various types of network services, such

as audio, video, and data services.

Major QoS indicators are as follows:

Bandwidth/throughput

The maximum transmission rate between two network nodes in byte/s or bit/s is termed

as the throughput.

Packet loss rate

The percentage of lost packets within a certain period of time during network

transmission is termed as the packet loss rate. In normal cases, the packet loss rate is 0.

When network congestion occurs, packets can be discarded as required.

Latency

The duration from the time point when a packet is sent from a node to the time point

when the packet is received at another node is termed as the latency. For a network

device, the latency refers to the duration between the arrival of the first bit of a data

stream to the departure of the last bit of the data stream.

Jitter

The variation in packet delay is termed as the jitter. The latencies for different types of

data packets within the same connection are different. For example, the first packet

within a connection takes 10 ms to be transmitted from a network node to another, and

the second packet within the connection takes 15 ms. The jitter between the two

transmissions is 5 ms.

QoS can be used to implement the following functions:

Categorization

Categorization refers to the capability to distinguish the data packets of each application

from those of other applications. With the categorization function, the SE2900 implements differentiated management on the data packets of various applications.




Tagging

A data packet must be tagged after being identified so that it can be processed by

switches or routers with a proper priority.

Priority marking

Data packets of different services must be allocated with different priorities so that

important services can be ensured when the network is overloaded or congested. All the

service traffic must be identified by the devices on the backbone network so that

different services are properly processed based on their priorities.

1.2 Value

QoS can guarantee end users to obtain the expected service experiences. For example, the

maximum downlink rate is 1 Mbit/s when end users download files from the Internet; the

maximum downlink rate is 512 Kbit/s when end users use P2P applications; VoIP services

with acceptable voice quality are available when end users browse web pages on the Internet;

Smooth video streaming is available when end users send and receive email messages.

QoS can help carriers gain the following benefits:

Guaranteeing service quality

QoS enables carriers to accurately manage bandwidth resources and provide standard

services for end users.

Earning interest on differentiated services

QoS enables carriers to provide differentiated services for different types of users.

Specifically, carriers can categorize users into different groups, allocates different

bandwidths and priorities to users in these groups, and charges them accordingly.

Ensuring user experiences to improve brand trust

QoS enables carriers to ensure user experiences by preferentially allocating bandwidths

for applications, such as VoIP and media streaming, that are sensitive to network latency

and jitters.

Avoiding network congestion to improve user satisfaction

QoS enables carriers to control the bandwidth allocated to specific services, such as P2P,

that consume much more bandwidths than other services, avoiding network congestion.

1.3 Service Models

Common QoS service models are as follows:

Best-Effort service

This service model applies to the scenario where only one type of service is available. In

this case, no guarantee is provided for the bandwidth, latency, jitter, and packet loss rate.

This service model is the most common one on the Internet and applies to the most data

applications, such as FTP, WWW, and email.

Integrated service (IntServ)

This service model is a comprehensive model that provides E2E QoS assurance using

RSVP. Before sending packets, UEs initiate requests for bandwidths and latencies. Each

involved node allocates resources to the UEs based on the resource usage on the node

and monitors the status of each passing data stream. This service model is difficult to




deploy and expand. In addition, the storage and processing capabilities of involved

routers in this service model may be exhausted as network traffic surges.

Differentiated service (DiffServ)

This service model usually applies to the scenario where multiple services are available

and helps distribute packets of various priorities into corresponding service streams. In

this service model, each node involved in packet forwarding implements a per-hop

behavior (PHB) for the packets of each service. This service model is easy to deploy and

expand. However, this service model cannot provide E2E QoS assurance due to the lack

of signaling communication between the network and end system.


Technical White Paper for QoS 2 Key Technologies


2 Key Technologies

2.1 Anti-Packet Loss and Anti-Jitter Technologies

2.1.1 Packet Loss Compensation

Because of network congestion, buffer overflow, and transmission errors, packet loss is

common in connectionless IP networks. For audio services, packets are transmitted in

temporal order. If certain packets are lost, letting them be lost is better than retransmitting

them. For the most audio compression algorithms in IP telephony technologies, frames are the

minimal unit in network transmission. Continuous packet loss affects the quality of the

decoded voice on the receiving end. The packet loss compensation (PLC) algorithm is used

for the SE2900 to prevent the impact of media packet loss. Based on the dependencies of the

context voice information, the lost frames are regenerated during decoding, which guarantees

the quality of the received voice.

Figure 2-1 shows the PLC mechanism.

Figure 2-1 Mechanism of the PLC algorithm

2.1.2 RFC2198 Redundancy

RFC2198 redundancy enables soft terminals to access the SE2900 or interwork with other

UEs. Specifically, RFC2198 redundant transmission of media packets is implemented

between the SE2900 and soft terminals to compensate for packet loss on the IP network

because of the complex Internet environment and severe packet loss, ensuring call quality on

soft terminals.

Figure 2-2 shows the mechanism of RFC2198 redundant transmission.




Figure 2-2 Mechanism of RFC2198 redundant transmission

Fra

me

1

Packet 1

IP access

network

Fra

me

2

Fra

me

1

Fra

me

2

Fra

me

3

Fra

me

3

Fra

me

4

Packet 2Packet 3Packet 4

Fra

me

1

Packet 1

Fra

me

2

Fra

me

3

Fra

me

3

Fra

me

4

Packet 3Packet 4

IP core

network

Fra

me

1

Packet 1

Fra

me

1

Fra

me

2

Fra

me

3

Packet 2Packet 3Packet 4

In RFC2198 redundancy, each packet carries certain frames in the previous packet and certain

new frames. Once a packet is lost, the receiving end still receives the new frames carried in

the packet because those frames are also carried in the subsequent packet.

2.1.3 De-Jitter Buffer

Jitters are inevitable when audio packets pass the IP bearer network. However, audio packets

must be transmitted in a specific sequence because they are sensitive to temporal order. In this

case, a jitter buffer with sound performance is a must. To reduce the impacts caused by jitters

on IP networks, the receiving end can postpone the play of received audio packets by caching

them in the buffer. The delay is determined by the capacity of the jitter buffer. The larger

capacity the jitter buffer has, the more packets the SE2900 receives before playing, which

means that fewer media packets are lost. However, a longer duration for audio packet to be

cached in the jitter buffer causes a longer latency for the E2E call. The size of a jitter buffer is

determined based on requirements. Overflow and redundancy usually cause packet loss,

packet transmission delay, and jitters. The SE2900 is able to adjust the jitter buffer size in real

time based on network situations when audio transcoding is implemented.

Figure 2-3 shows the de-jitter buffer (DJB) mechanism.

Figure 2-3 De-jitter buffer mechanism




2.2 Bandwidth Control Technologies

2.2.1 Policy Control (Rx)

Policy control (Rx) enables the SE2900 to interwork with the policy and charging rules

function (PCRF) over the Diameter-based Rx interface. The SE2900 extracts the session

information about a UE, such as the signaling address, media address, and media bandwidth,

from SIP messages, and sends such information to the PCRF over the Rx interface. The PCRF

uses the session information to instruct the policy and charging enforcement function (PCEF)

to implement PCC so that the bearer resources are under control and bandwidth is guaranteed,

ensuring service quality and improving user experience.

Figure 2-4 shows the mechanism of policy control (Rx).

Figure 2-4 Policy control (Rx) mechanism

SE2900

(P-CSCF) UE PGW

RxPCRF

Gx

2.2.2 Call Admission Control

Call admission control (CAC) is a function which enables the SE2900 to restrict the resources

available for registrations, calls, and subscriptions, based on locally configured CAC policies.

This function enhances the security of the SE2900 and core servers and guarantees QoS for

registration and call services.

Figure 2-5 shows the CAC mechanism.




Figure 2-5 CAC mechanism

Access

network A

Access

network B

Core

network A

SE2900

(A-SBC)

Core

network B

SE2900

(I-SBC)

Implements the control over

the resources used by users

on the entire access network.

Implements control over the

resources used by a single

user on the access network.

Implements control over the

resources used by a single

core network user.

Implements the control

over the resources used by

the entire core network.

2.2.3 Committed Access Rate

The committed access rate (CAR) function is an important and efficient bandwidth

management function. It is usually configured on a device at the network border and enables

the device to restrict the incoming packet rate, which helps carriers provide guaranteed QoS.

The SE2900 detects the bandwidth resources used by media streams. If the media streams

from a user consume excessive bandwidth resources, the SE2900 discards subsequent media

packets, preventing bandwidth theft.

Figure 2-6 shows the implementation of CAR using token bucket model:

Figure 2-6 CAR mechanism

Tokens

p B - Burst Size

p - Token Arrival Rate

Overflow

Tokens

Packets

Arriving Conform

Exceed




2.2.4 Voice Activity Detection/Comfort Noise Generation

Voice activity detection (VAD) is a technique used in speech processing in which the presence

or absence of human speech is detected. It makes an important component of IP telephony by

helping retain service quality, reduce the use of voice circuits, and reduce the E2E latency.

Users do not occupy the voice channels all the time when placing calls. According to the

statistics on the traditional telephone services, a party actually occupies not more than 40% of

the conversation period during a call. This is caused by the following reasons: One party is

listening to the other party. Pauses in the talk are inevitable due to thinking and breaks. Pauses

may be caused by hesitation, breathing, and stammer. The first instance has a long pause and

is rare. The third instance has a short pause and is common. The pause and frequency of the

second instance is between the first and the third instances. The basic principle of VAD is to

judge the voice signal energy. When the energy is below a certain threshold, the UMG8900

considers that the party is in the silence state.

There is one issue in VAD applications. In the mute periods, the comfort noise generation

(CNG) technology is used to generate comfortable noise so that the users will not feel

unnatural in conversation or regard call interruption by mistake. Therefore, a comfort noise

generator must be configured at the receiving end. The receiver adopts a mechanism to

regenerate the main features of the background noise. The background noise parameters are

specified by the transmitting end. The noise generation methods must guarantee the

synchronization of the decoder and the encoder. Even if the encoder does not send any bit in a

certain period, the decoder can identify the absence of bit, which ensures the smooth

assembling of voice segments and silence segments when the voice is restored.

2.3 Forwarding Priority Control

2.3.1 Differentiated Services Code Point

The IP network uses the best-effort mechanism to transmit IP packets. Consequently, when

real-time audio data, video data, or control data are transmitted over the IP network, it cannot

ensure the reliable data transmission. When the network is congested, the quality of services

with a high demand for the QoS cannot be guaranteed.

The differentiated services code point (DSCP) enables UEs to apply for services at various

levels by setting the DiffServ field. The first six bits in the DiffServ field define a DSCP value.

Packets with the same DSCP values are in a set called a behavior aggregate (BA). For each

DSCP value, a PHB ID is configured on the router to identify the behavior, such as traffic

monitoring, traffic throttling, or queuing, that meets certain forwarding requirements. When a

packet arrives, the router distributes the packet into a BA and selects a PHB based on the

DSCP value carried in the packet and forwards the packet accordingly.

The SE2900 supports the DiffServ technique and sets different DSCP values for signaling and

media packets so that packets with high DSCP values are preferentially forwarded, ensuring

service quality.

Figure 2-7 shows the DSCP mechanism.




Figure 2-7 DSCP mechanism

QoS high

SE2900

UE

QoS processing

SE2900IP network

QoS low

…

…

QoS medium

DSCP

XXX

DSCP

YYY

DSCP

ZZZ

DSCP

XXX

DSCP

YYY

DSCP

ZZZ

1

2

3

1

2

3

The packets marked with higher priorities are processed and forwarded first.

2.4 QoS Monitoring

2.4.1 Voice Quality Reporting

Voice quality reporting enables the SE2900 to measure voice quality of calls in real time,

including the packet loss rate, jitter, round-trip delay, number of received/sent RTP packets,

number of bytes of received/sent RTP packets, and mean opinion score (MOS).Voice quality

reporting enables the SE2900 to implement the following functions:

1. Reports QoS statistics to the EMS so that carriers can monitor the network performance

based on the performance measurement, adjust and optimize network, and improve

service quality.

2. Reports QoS statistics to the EMS using user tracing messages for real-time visualization

of the service quality.

3. Reports QoS statistics to the LMT.

4. The SE2900 includes QoS information in INFO requests.

5. The SE2900 includes QoS information in BYE requests.

Voice quality statistics are calculated based on the fields related to RTP and RTCP.

Figure 2-8 shows an example RTP packet:




Figure 2-8 An example RTP packet

Figure 2-9 shows an example RTCP-SR packet:




Figure 2-9 An example RTCP packet

The calculation involves the following fields:

Sequence number:

The sequence number starts from a random number and increments by one when an RTP

packet is sent. The receiving end detects packet loss and assembles packets based on the

sequence number.

Time stamp




The time stamp indicates the sampling time of the first bit in an RTP packet. The sampling

time precision must meet the synchronization requirement to facilitate the synchronization

and jitter calculation. The start value of the time stamp is random. The time stamp value

ascends with the data size.

Last sender report (LSR):

The LSR is a field in RTCP packets and contains the 32 bits in the middle of the 64-bit

NTP timestamp reported by the latest sender of the RTCP packet from SSRC_n. If no SR

is received, this field is set to 0.

Delay since last sender report (DLSR):

The DLSR refers to the duration between the receiving and sending of the SR from

SSRC_n. If no SR is received from SSRC_n, this field is set to 0.

Packet loss calculation method

The packet loss rate is the total number of packets received in a call within a specific

duration divided by the difference between the sequence number of the last received

packet and that of the first received packet within the duration.

Jitter calculation method

Assume that Si is the RTP time stamp of packet i and Ri is the arrival time of packet i in

the RTP time stamp, for packet i and packet j, D(i,j) = (Rj - Sj) - (Ri - Si). Use the

difference (D) between packet i and its preceding packet i-1 (calculated in arrival order,

not in sequence number order) to consecutively calculate the arrival time jitter after each

packet i is received from the source SSRC_n. The calculation formula is J = J +

(|D(i-1,i)| - J)/16.

Round-trip delay calculation method

Assume that SSRC_r is the receiver of an SR message. Time point A when the SSRC_n

receives the SR message from SSRC_r can be used to calculate the round-trip delay.

Specifically, use the LSR to calculate the duration of the round trip (A-LSR), and then

subtract DLSR from A-LSR to obtain the round-trip delay.

1.1 Multimedia Priority Service (MPS) • During the registration of a user, the P-CSCF obtains priority information of the user

from the S-CSCF.

• After receiving the INVITE request from the user, the P-CSCF delivers the obtained

priority information of the user to the PCRF so that the PCRF ensures the quality of

services available for the user based on the priority information.

• The P-CSCF includes the Resource-Priority header in the INVITE request before

forwarding the request to the core network so that the core server can identify the

priority of the user.

• The P-CSCF preferentially ensures the quality of services available for high-priority

users.




I/S-CSCF HSS

SE2900

P-CSCF/ATCF/ATGW

2

MME

PCRF

S/P-GW

eNodeB

1

1) Store the MPS of

subscribers

3

4

4) Map MPS to ARP

3

1) Implement ARP policy

2) Get MPS from HSS and

synchronize to other NEs

3) Indicate MPS to PCRF

and IMS-Core during call

Resource-Priority

Technical White Paper for QoS on SE2900 V300R002

Documents

Transcript of Technical White Paper for QoS on SE2900 V300R002