VOIP

•                                                                   Voice Over IP

what is voip?

• voice over internet protocol attempts use a LAN and /or WAN (e.g.the internet) carry voice in the same way as the telephone system.

Why?

• save costs
• improve facilities.

Why VOIP?

• companies could save costs by using
• VOIP internally on the LAN (no need to maintain /rent expensive PABX + Infrastructure)
• Voice calls to other HQ/partners over the internet instead of Telecoms.

                                                            

  Why VoIP?

• Companies can device improved methods of working and customer service with integrated  voice/computing.
• Telcoms (eg. BSNL) could save money by having  only one network-the internet (and “pension-    off” the telecoms  network).

THE SOLUTION

Turn your pc into a phone and make calls over internet.

Change in Pattern

Traditional Telecom Model                                                       New Telecom Model

Value added services

Data service

Voice service

Infrastructure

Voice service

Value added services

Data service

Infrastructure

                              

Voice is becoming just another IP application!

PSTN vs INTERNET

We have two large (world wide)digital networks witj the following characteristics:

	Reliability	Quality of service(QOS)	price	size
Telephone network	v. high	v. high	high	World wide static
Data network	variable	variable	low	World wide, increasing.

VOIP problems

·         Reliability of LANS/the internet

·         Reliability-phone system

·         -99.999%reliability(“five 9’s”)

·         i.e 03 seconsds per month.

·         Reliability-internet

        -?99.9%

         Speech quality

·         -delay

·         -packet loss

Quality of service(QOS)

QOS is the level of service expected by the customer.For telephones,the customer expects:

-100% availability

-good speech  quality

-imperceptible delay(e.g.less than 50 ms)

-imperceptible echo

The internet has trouble matching these requirements

Voice Over PSTN

Voice is digitized into 8-bit samples at the (BT)

Exchange every  125us.

Voice over the PSTN

This is known as pulse code modulation(PCM)

11101001

11010100

00110101

00101001

00001011

00111010

Samples every 125us is 8000 times per second.(speech band width of approx. 4khz)

The resulting data stream is 8bitsx8000=64,000bit/sec.(ITU standard G.711)

64KPS speech streams are brought together into larger streams:

-E1:30 speech channels (2Mbps)

-E3:120 speech channels (8Mbps)etc.

These speech channels are switched digitally around the country (or world) with minimal delay at each switching point.

Data over the Internet

The pc attached to router 6 received the packets 1, 4 &3(in that order).

What happens next depends on the transport layer protocol, either:

-TCP (transmission control protocol)

-UDP (User datagram protocol)

TCP will correct the incorrect order and “ask” the missing packet to be re-sent. 

UDP has no packet numbering and hence is unable to ask for missing packets.     

TCP is suitable for:

-web pages

-file transfer etc. (all data to be correctly transferred)

But

-adds considerable extra data in the header (20+octets)

-adds delay: no time to wait for missing packets (it would take too long).

Hence: UDP is used for real time applications

UDP

-Adds only 8 octets to the header

-suitable for real time applications

-requires addition of RTP (real time protocol) to give timing information for play out of packet data.

-occasional RTCP (real time control protocol) packets supervise and report back on quality of the link.    

VOIP uses IP+UDP+RTP as described above

These headers add about 50 octets to the data, so several 8-bit samples are included in each packet.Typically 160(20ms) or 480(60ms).

Hence the header over head is:

-30%(20ms)or 10%(60ms)

Resulting in bit streams of 105 kbps or 88 kbps.

-too high for typical analogue modems!(33-56kbps)

Typical voip packet

13kbps data->36kbps with headers

P P p

I P

U D p

R T p

Voice packet(20ms)

Voip –codecs:

Speech is compressed using a codec(coder/decoder)

-source waveform, eg. ADPCM (adaptive PCM)

a predictive PCM encoder

reduces 64 kbps to 32 or 16 kbps with little loss of quality.

-vocoders

Attempt to model the vocal tract

Very low bit rate

But very unnatural sounding and not generally liked

Not possible to recognize the speaker

Hybrid  eg. CELP (coide excited linear predictive)

Combine the best of the above two methods.

VOIP--Codecs

Typical codecs

	type	Bit rate kbps	Coding delay	Quality(mos)	quality
G.711	PCM	64	<1MS	4.2	GOOD
G.726	ADPCM	32	<1MS	4.0	good
G.728	CELP	16	2ms	4.0	good
GSM	RPE-LTP	13.2	2ms	3.7	Fair good
G.729	CELP	8	5ms	4.0	good
G.723.1	CELP	6.4	7.5	3.8	Fair- good

VOIP - Quality

Speech quality is measured using a method call mean opinion score(MOS).

A subjective measurement based on the results from a number of listening volunteers.

MOS score	Listening quality	Quality of speech
5 4 3 2 1	Excellent Good Fair Poor Bad	Complete relaxation Attention necessary Moderate effort required Considerable effort required No meaning understood

VOIP-quality

Typical MOS:

-4.2(GOOD) "TOLL-QUALITY"SPEECH

-3.7(GOOD) GSM codec

-3-4 hybrids eg.CELP

-1-2 vocoders

Two major problems for VOIP are:
-packet delay

-packet loss

These are inherent in the internet which provides a “best-effort”service only.

VOIP –packet delay

There are many sources of delay to the packets as shown in the typical example below:

                                                                                                           

Delay type	Fixed	Variable
Codec/packetization	30ms
Transmission time	10ms
Network delay	60ms	+1000ms
Jitter buffer(receiver)	1000ms
total	1100ms

Note the addition of ajitter buffer at the receiver to cope with the variation in delay due to:  

-queuing at the LAN router (other packets may be in front)

-variable network delays(busy routers,queued packets,different routes)

This adds to the overall ene-to-end delay.Does this matter.????

VOIP –delay

End-to-end delay results in significant pauses in conversations when one party stops speaking.

This leads to “doyble-talking”where both parties speak together,and ruins a conversation.

The ITU  recommended that one-way speech delay is less than 150ms(with 400ms as an absolute limit).

The target of 150ms for one way to delay is aextremely hard to meet.

The internet was designed as merely a best effort service for file transfer,email,wed browsing etc,where delays up to a few seconds are acceptable(and normal).

VOIP-delay

Methods for reducing delay in IP networks include:

-reserving band width(int-serv)

-prioritizing voice (diff-serv)

Unfortunately there is currently no agreed standard for routers in the u=internet,hence delivery is still "best effort".

The next generation of IP protocol (ipv6) standardizes on  diff-serv,but until that is available end-to-end across a number of ISPs,then VOIP will continue to experience delay and loss problems.

VOIP packet loss

For normal PSTN,call losses would be negligible.

However, for a VOIP call, packet loss must be expected when routers are busy,or jitter –buffer delay times exceeded.

Packet losses of upto 5% may be acceptable,but speech is quite seriously affected beyond that.

Diff serv would alleviate packet loss by giving priority to speech.

Other VOIP problems

Firewalls tend to prevent VOIP(unless specially configured)

Privacy?speech encryption?

Access to emergency services

Criminal activity

“SPIT”:spam over internet Telephony 

call tracing

VOIP protocols and standards

There are two main standards for setting up VOIP calls:

-H.323 from ITU  using ASN.1commands as used by net meeting

-SIP from IETF using text commands as used by  XP messenger

 SIP=session initiation protocol

                                VOIP protocol stack

Within the VOIP suite of protocols,voice packets are commonly referred to as the data plane and signaling packets are commonly referred to as the control plane.

Data plane protocols

Voice sample

CODEC

RTP

UDP

IP

IP/UDP/RTP header (40 bytes)	Voice samples(20-100+bytes)Depends on CODEC
Standard RTP packet
IP/UDP/RTP header (2-4 bytes)	Voice sample(20-100+bytes)depends on CODEC
CRTP packet

RTP:

RTP is the protocol that supports user voice.each RTP packet contains a small sample of the voice conversation.The size of the packet and the size of the voice sample inside the packet will depend on the CODEC used.

CRTP:

Since the IP header is compressed with the UDP  and RTP headers down to a maximum of 4 bytes,there is no room for an IP address.Therefore, the packet cannot be routed. It can only be placed on a point -to -point link that requires no addressing.

 VOIP Interoperability

VOIP is not a protocol.VOIP is acollection of protocols and devices that allow for the encoding,transport and routing of audio calls over IP networks.

PSTN >>VOIP >>PSTN

Native VOIP >>PSTN

Native VOIP >> Native VOIP

CODECS

• Codec-compressed and decompression method to compress voice  while preserving voice quality
• Codecs consists of 4 parameters

-Compressed voice rate -The bit rate for the audio payload

-Complexity-The better the compression and resulting quality,the more CPU processing the 

  codec needs on each end

-Voice Quality-varies from toll to satellite-like

-Delay –More complex codecs often take longer to compress and decompress.

Voice codecs

• G.711

-uncompressed PCM audio stream

-8ks/s of 8 bit values=64kbps

-packet”size”=10,20,30,and 60ms

• G.722-wide band(7khz)
• G.726

-ADPCM-10,20,30.60ms-32kbps

• G.723.1

-MLQ -30ms-5.3 or 6.3 kbps

-Silence suppression G.729

-CS-ACELP-10,20,30ms-8kbps

Packet Encapsulation

H.323

• H.323 is part of a larger body of standards(H.32x)that addresses  multimedia communications over a variety of networks:
• H.310-for multimedia communications over B-ISDN (broadband integrated services digital network).
• H.320-for multimedia communications over narrow band ISDN.
• H.321-for multimedia communications  over ATM
• H-322-for multimedia communications over LANS.
• H.324-for multimedia commincations over PSTN.

H.323

As a minimum.H.323 specifies  protocols for real time point to point audio communicatin betweem two terminals on a packet data based network that do not provide a guaranteed quality of service.

The entire scope of H.323 is much broader and encompasses inter network multi point conferencing among terminals that support not only audio,but also video and data communication.

H.323 Architecture

In a general H.323 implementation ,four logical entities are required

Terminal

Gateways(GW)

Gate keepers(GK)

and multipoint control unit(MCU) 

H.323 Terminal

A terminal or a client is an end point where H.323 data streams and signaling originate and terminate.it may be a multimedia PC with a H.323 complaint stack or a stand alone device such as a USB(universal serial bus)IP telephone ,which provides for real time, two way communications with another H.323 terminal,gateway or MCU.This communication between endpoints consists of speech only,speech and data,speech and video,orspeech,data and video.

Gateway(GW)

Gateway(GW)

A gateway is an optional component in an H.323 enabled network. when communication is required between different networks (e.g., between an IP based network and PSTN),a gate way is needed at the interface.

Gateway(GW)

A H.323 gateway is an H.323 end point that provides for real time,two-way communications between terminals belonging to networks with different protocol stacks.for  example,it is possible for an H.323 terminal to setup conference with terminals based on H.320 or H.324 through an appropriate gateway

A gateway provides data format translation,control signaling translation ,audio and video codec translation,and call set up and termination functionality on both sides of the network.depending on the type of network to which translation is required a gateway may  support H.310,H.320,H.321, OR H.324 endpoints.

Gatekeeper (GK)

Gate keeper(GK)

The gatekeeper provides address translation and control access to the network resources for H.323 terminals,GWs and MCUs.

Gatekeeper(GK)

A gate keeper is  very useful,but optional,component of an H,323 enabled network.

Multipoint control unit - MCU

The MCU ,an optional component of an H.323 enabled network,takes care of establishing multipoint conferences. It consists of:

A mandatory multipoint controller(MC)-used for call signaling and conference control.

An optional multipoint processor(MP)-used for switching /mixing of media streams,and sometimes for real time transcoding of the received audio/video streams.

Although the MCU is a separate logical unit,it may be combined into aterminal,gateway.or gate keeper.

Multipoint control unit(MCU)

The MCU is required in a centralized multipoint conference  where each terminal establishes a point to point  connection with the MCU.The MCU determines the capabilities of each terminal and sends each a mixed media stream.in  the decentralized model of multipoint  conferencing an MC ensures communication compatibility but the media streams are multicast and the mixing is performed at each terminal.

H.323 protocol stack

H.323 is an umbrella  specification for the many different protocols that makeup the overall H.323protocol stack 

                                    

The protocols in the H.323 protocol suite are:

Call controlling and signaling:

-H.225.0:call signaling protocols and media stream packetization(uses a subset of Q.931 signaling protocol).

-H.225.0/RAS:registration,admission and status.

-H.245:control protocol for multimedia communication.

H.323 prtocol stack

T.120 is an ITU-T recommendation that describes a series of communication and application protocols and services that provide support for real time,multipoint data communications.it is used by products such as Microsoft net meeting and lotus sametime to support application sharing,realtime text conferencing and other functions.

H.323 protocol stack

The set of “T.120”recommendations includes:

T.120- data protocols for multiledia conferencing

T.imp- revised implementors guide for the ITU-TT.120 recommendation series

T.121-generic application template

T.122-multipoint communication service-service definition

T.123-network specific data protocols for multimedia conferencing

T.124-generic conference control

T.125-multipoint communication service protocol specification

T.126-  multipoint  still image and annotation protocol

T.127-multipoint application sharing

T.134-text chat application entity

T.135-user –to-reservation transactions with in T..120 conferences

T.136-remote device control application protocol

T.137-virtual meeting room managenet-services and protocol.

PSTN HARMONIZATION

Based on PSTN QSIG lgical layer.ECMA defined formal interworking solutions

H.450 rich suite of services

-call transfer(H 450.2)

-call diversion/forwarding(H.450.3)

-call hold(H.450.4)

-call park and pick up(H.450.5)

-call waiting(H.450.6)

-message waiting (H.450.7)

-name identification(H.450.8)

-call completion on busy(H.450.9)

-call offer(H.450.10  uses H.450.6)

-call intrusion(H.450.11)

-H.450 capabilities and policies (H.450.12)

Enhanced networking solutions

Most efficient call setup-fats start and its enhancements

Reliable signaling-TCPO based since day one

Q.931 multiplexing-performance boost for high capacities

QOS-can request QOS from network(RSVP)

Media streaming-over  UDP/ATM

Service availability-desired protocol requests

Real time fax-on line swithching to fax transmissions

Requirements addressing

Very rich conference services

Advanced PSTN  harmonization

Add on services suite and service creation enabled

Advanced service management

Efficient networking resources usage

Operate today on many environments and chip sets......................

H.323 protocol stack

Audio processing:

-G.711:pulse code modulation of voice frequencies.

-G.722:7khz audio coding within 64kb/s.

-G.723.1:Dual rate speech coders for multimedia communication transmitting at 5.3 and 6.3 kb/s.

-G.728:coding of speech at 16kb/s using low delay code excited linear prediction

-G.729:coding of speech at 8kb/s using conjugate structure algebraic code excite linear prediction.

H.323 protocol stack

Video processing:

-H.261:video codecs for audio visual services at px 64kps.

-H.263:video calling for low bit rate communication security.

-H.235:security and encryption for H. series multimedia terminals.

H.323 PROTOCOL STACK

Data conferencing:

-T.120:this ia a protocol suite(that includes T.123,T.124,T,125)for data transmission  between end  points.it can be used for various applications in the field of collaboration work,such as white boarding,application,application sharing and joint document management.

T.120 utilizes layer architecture similar to OSI model.