TRADITIONAL TELEPHONE

Well setup
Proprietary PBX
Uses TDM/FDM

VOICE OVER IP
Open ish- can use different vendors for different network components
Unified network and staff- same staff for all network
Internet telephone service provider- Both data and communication
Toll bypass- using internet for both data and communication
Phased migration supported-using tradition PBX systems along with voice gateways to get all the
benefits of VoIP.

VOICE GATEWAYS
Entry and exit point for systems
Channel between internal network nodes to maybe TDM, T1, ITSP etc
Provide a line of security
Provide VoIP services
They act as a bridge between traditional connections (like fxo, fxs, t1 etc) and VoIP

VOICE PROCESSING
Brain of the network
Device control
Dial plans and permissions
Phone features

SINGLE SITE
CUCM/call manager publisher- R/W – copies of the database that are used for IP telephony
CUCM subscriber- R/O
Unity- voicemail

CENTRALISED MULTISITE
In addition to single site features it has SRST(survivable remote site telephony-support features which
are provided when connection to CUCM goes down) and even the router is able to connect phone calls
in case the connection with the CUCM is broken

FXO- analog port to the PSTN

FXS(foreign exchange station)- connecting analog devices to the voice gateway so as to use VoIP services

MULTISITE DISTRIBUTED
Has independent Single sites each having their own subscribers and publishers
GK(gate keeper)- provides control functions in the multisite distributed network that is common for
everyone. Provides bandwidth management, acts like a phonebook
The GK can divide the network into different zones. It is a CENTRALISED MANAGEMENT. All the
configuration and setup goes into the GATE KEEPER.
SPLIT CLUSTER
The CUCM publisher is kept at one site and the subscriber is kept at another. So the copy of the
database is made at the subscriber at the other site. But this works only if the networks are very close to
each other and have high speed connectivity because the replication process has a certain time limit.

Voice switches – provide features like power over the Ethernet so that IP phones can be connected to
them. They also provide Quality of service(QoS) so that phone calls take priority over data

ATA(Analog telephone adapters)- for connecting the analog phones to the Ethernet.

***********cisco voice design models

***********Cisco voice gateways

2----------------------------------------------------------------------
FXS (foreign exchange station)- a port which allows analog devices to be connected to the router and on
the other port we have Ethernet port so that that analog devices can be used as a VoIP devices.
Connects analog phones(generally)

FXO (foreign exchange office)- used to connect the PSTN network to the VoIP network. Its an analog
connection. Connects to central office(generally).

Analog = 1 call per port.

A router becomes a voice gateway because of modules like FXS, FXO and also the software feature
inside the ios.

A router is installed with modules like the DSP(digital signal processor) which convert the incoming voice
into tiny audio packets

The DSP takes the analog voice , digitizes it, then converts them into packets and add IP headers which
are then sent over the internet.

E&M(Ear and Mouth) – it is used to bridge PBX. But the PBX system must have E&M cards installed in
them. The voice gateways(voice routers) also have E&M cards. Then again the voice gateways are
connected to the IP phones. The PBX systems also have FXS and FXO ports which can be connected to
FXO and FXS ports on the voice gateway respectively.

CAMA(centralized automated message accounting) – used solely for emergency calling. Connected to
the PSTN and delivers ANI/Caller-ID to the right authorities.

SIGNALING
Informational – busy, ring etc
Supervision – on/off hook
Addressing – dialing

In case of FXO and FXS signaling is of two types namely loop start and ground start
In loop start when we lift the receiver the circuit loop is connected and current begins to flow through
the device. In ground start we have to ground the connection so as to get the dial tone.

Whatever signaling is used by PBX or PSTN or any other traditional network, in order to connect to
them we have to match that same signaling technique at out voice gateway. We also have to match
the type and the operation(number of wires etc)

PSTN uses loop start PBX uses ground start (most of the time)

COMMANDS
Router#sho ver (showing the version)
Router#show voice port summary (summary of the ports)
Router#conf t (global configuration command)
Router<config>#voice-port 1/0/0
Router<config-voiceport>#?

Router<config-voiceport>#no comfort-noise
Router<config-voiceport>#signal ?
Router<config-voiceport>#cptone ?
Router<config-voiceport>#cptone us (call progress tone united states)
Router<config-voiceport>#ring cadence ?
Router<config-voiceport>#ring cadence define 10 ?
Router<config-voiceport>#shutdown
Router<config-voiceport>#no shutdown

We can’t assign dial plan/phone number

Assigning a phone number

Router<config-voiceport>#exit
Router<config>#dial-peer voice ?
Router<config>#dial-peer voice 1000 ?
Router<config>#dial-peer voice 1000 pots
Router<config-dial-peer>#destination-pattern 1000
Router<config-dial-peer>#port 1/0/0
Router<config-dial-peer>#^Z
Router#csim ?
Router#csim start ?
Router#csim start 1000
For setting a dial number as soon as we lift the phone
Router#conf t
Router<config>#voice-port 1/0/0
Router<config-voice-port>#connection plar ?
Router<config-voice-port>#connection plar 1000

Gns3.net (emulated routers)

3------------------------------------------------------------------------------------------------
Voice gateways basically translates between VoIP and analog signals (one of the function)

The IP phones already contain DSP which converts the voice into digitized packets along with the
addition of IP headers which are then sent to the voice gateways. The VG strips the IP header and
delivers the packet to the concerned system using a suitable PSTN signaling.

VWIC – voice and WAN interface cards- can understand both voice as well as data

Digital ports allow multiple calls as opposed to analog which allows only one call per line.

Digitals ports 3 types- T1 (used in US japan etc), E1 (used everywhere), ISDN (for connecting to PSTN PBX
etc).

CAS (channel associated signaling)- associated with T1 and E1. Each channel is 64Kbps – every 6th bit is
used for signaling. It is just a framing standard not the actual signaling.
Types – SF (super frame)- signaling is used to control 12 frames. ESF (extended super frame)- 24 frames
are controlled by signaling.
Advantages-all the channels are available for data. Simultaneous 24 calls can be accommodated because
we don’t dedicate a single channel for signaling but used a single bit from every channel for that
purpose.
Disadvantage-reduce user bandwidth.

CCS (common channel signaling)- dedicates an entire channel to signaling.

Disadvantage- you loose one entire channel
Advantages-more bandwidth. And hence supports more features. Used in PBX systems so as to get more
features and services with the help of advanced signaling

ISDN protocols.
Q.921 (layer 2) – error detection/correction
Q.931 (layer 3) – call setup, teardown, information such as caller ID etc.

NFAS (non facility associated signaling) – uses a single ‘D’ channel, for signaling, for multiple PRI
interfaces.

QSIG – used to connects PBXs from different vendors.

CONFIGURATION
Router>en
Router#

LEARN ROUTER CONFIGURATION SETUP

T1-port 23(actually it is the 24th port but the counting starts from 0) is the signaling port
E1- port 15(16th port) is the signaling port

5-----------------------------------------------------------------------

ECHO
It happens everywhere every time. Its just that the delay is so small that we are not able to notice it. It
usually happens at the 2 wire-4 wire interface. The 2 wire coming out of our telephones when come in
contact with the 4 wires of the telephone company, there is a reflection at the interface. Usually the
delay is small but as the communication distance increases the delay becomes significant. The IP
network itself can add its own delay, generally during TOLL BYPASS, before it reaches the PSTN network.

In order to cancel the echo, the voice gateway/voice routers make a copy of the outgoing signal and if it
receives the exact same signal in a given amount of time say 24ms then it completely blocks that signal
it is an ECHO signal. This feature is enabled by default in the routers. The coverage can be increased.
This feature consumes resource on your router and by reducing the coverage we can save ourselves
some resources if we want.

CROSS CONNECTING VOICE PORTS

we ca take our digital voice port such as T1 and break it into individual DSO channels and then bridge
them to analog ports such as FXS or FXO. As soon as we lift the phone on an FXS line it automatically
grabs a channel on the T1 line.

Rules for such a connection

The analog and the digital ports must be on the same module.
Up to maximum of 4 analog ports per digital port
DS0 group has to be one time slot. Signaling must match. In case of FXS and FXO, an FXS port must be
paired up with FXO and vice versa.

6--------------------------------------------------------------------------------------

CODEC- stands for coder decoder. It is a method used to turn our audio or voice into packets

Codec BW(kbps) MOS char. PPS bytes sample size

G.711 64 4.3 uncompressed 33/50 240/160 30/20
G.729 8 3.92 compressed 33/50 30/20 30/20
iLBC 15.2 4.14 open source 33/50 50/38 30/20
G.711- standard codec for voice communication. Completely compatible with old style of
communication. Uncompressed audio of 64Kbps is compatible with channels of a T1 port. So the voice
gateways do not have to perform major translations between info coming from a G.711 phone and a T1
line. G.711 can be used to easily connect VoIP and old school analog phones. It is a high quality CODEC.
It also hides the packets lost, the DSPs fill those missing packets by extending the adjacent sample.

G.729- allows us the compress our audio to 8Kbps. especially geared for Latin based languages. It builds
a codebook and then sends it. It divides an audio into packets of 20ms. Now the adjacent packets have
more or less the same sound content. So this codec keeps a single sample and discards the rest and then
plays that sample for say 100ms, thereby saving a lot of bandwidth. Not good with music on hold. Not
good with non latin based languages. If there is a packet loss then sound starts to break very easily and
the quality drops drastically.

iLBC- it tolerates packet loss a lot better than G.729.

using 30ms sample size increases the efficiency as the less no of packets which less header info to
process. Less BW being used. Drawback is when we loose a packet it is more impactful. It also adds more
delay.

QUALITY EVALUATION METHODS

MOS (mean opinion score) – done manually by trained ears.

PESQ – speech quality. Done automatically by software by taking into account the delay, packet loss etc.
PEAQ – audio quality. Includes everything

Transcoding – changing from one codec to another eg. Changing from G.711 to G.729 or the other way
around which is most commonly used when attempting toll bypass.

CALCULATING PER CALL BANDWIDTH

1. Bytes per packet you need for your CODEC
2. Add 40 bytes per packet for layer 3 headers
3. Add in layer 2 header overhead (see the table in the video)
4. Multiply by PPS
5. Multiply by 8 to convert bytes to bits

G.711 is not as efficient as traditional methods bandwidth wise but its still okay because we have a lot of
bandwidth at our disposal.

RTP is the protocol of sound.

DSPs offload media processing from voice gateways.

Some functions of DSPs are as follows:-
1. Coding – take audio voice and use G.711 or G.729 ect to convert to packets. Also provides
functions like echo cancelation, VAD (voice activity detection).
 2. Transcoding
3. Media termination points – so that VGs can hold the call during a call transfer.
4. Conferencing- phones have enough resources to support 3 person conference call. for more
number of people the DSP take audio from all the users, mixes them up and then sends a single
audio to everyone.

DSPs are limited in the number of calls they can handle. It depends upon the codec complexity. In order
to tackle this limitation we use DSPfarming. In DSP farming we allocate portions of DSP resources from
various devices for a particular task such as conferencing. So when a conferencing request comes the
voice gateway access those assigned resources to complete the task.

7-------------------------------------------------------------------------------

Nyquist theorem – the theorem says that if we sample and analog signal twice as much as the highest
frequency then we can regenerate the signal fairly from their digital representation.

Human voice frequency range- 20-9000 hz

Core of human voice which contains the most info- 20-4000hz

So sampling the human voice 4000*2=8000 times gives a digital signal from which the original can be
recovered fairly. Now 8000*8 gives 64000 bits=64Kbps which is the bandwidth required to transmit
human voice.

PCM (pulse code modulation) – converts the amplitude scale into binary digits. It samples the audio non
linearly because more info is stored in the lower frequency spectrum than the in the higher freq.
spectrum. PCM scale lies in -127 to +127.

The samples are 8bit binary value with the leftmost bit representing whether it is positive or negative
and the rest of the bits represents the amplitude.

Assigning a number to each sample is known as quantization. Converting these numbers to binary is
done by PCM.

Compression is optional.

8--------------------------------------------------------------------------------

Call legs – they are the hops or links that are encountered while the signal travels to its destination IP.
They are of two types namely 1. POTS call leg 2. VoIP call leg

POTS call legs are used for traditional devices and systems/basically devices that do not have IP address.
VoIP call legs are associated with devices that have an IP address.
CME- call manager express. Installed on the router which starts behaving like a CUCM and perform call
processing and support other features.

The dial peers not just defines the voice routing or phone numbers but also the characteristics of a call
such as the CODEC to be used, features like VAD (voice activity detection). In simple terms they specify
which port to ring when a certain phone number arrives at the router.

_____________________related to configuraton______________
The tag name given during dial-peer command doesn’t mean anything as such. They just must be unique
on each router. It’s a good practice to match the destination pattern with the tag.

Everything in the pots dial-peer is associated with a port

Everything in the VoIP dial peer is associated with an IP address

COMMANDS
Router#sh start
Router#conf t
Router<config>#hostname R1
R1<config>#line con 0
R1<config-line>#logg s
R1<config-line>#no exec-t
R1<config-line>#exit
R1<config>#dial-peer ?
R1<config>#dial-peer voice ?
R1<config>#dial-peer voice 101(tag) ?
R1<config>#dial-peer voice 101 pots
R1<config-dial-peer>#?
R1<config-dial-peer>#destination-pattern ?
R1<config-dial-peer>#destination-pattern 102
R1<config-dial-peer>#port 1/0/1
R1<config-dial-peer>#exit
R1<config>#do show dial-peer voice sum
R1<config>#dial-peer voice 103 pots 2 dial peers on the same phone
R1<config-dial-peer>#destination pattern 103
R1<config-dial-peer>#port 1/0/1
R1<config-dial-peer>#exit
R1<config>#exit
R1#debug voip dial-peer
R1#u all turn off debugging mode
R1#conf t
R1<config>#no dial-peer vo 103 removing the dial peer
R1<config>#dial-peer voice 101 no pots as 101 has been defined before
R1<config-dial-peer>#destination-pattern 101
R1<config-dial-peer>#port 1/0/0
R1<config-dial-peer>#exit
R1<config>#dial-peer voice 102 pots
R1<config-dial-peer>#destination-pattern 102
R1<config-dial-peer>#port 1/0/1
R1<config-dial-peer>#do debug voip dialpeer

R1<config-dial-peer>#exit
R1<config>#dial-peer voice 1 pots
R1<config-dial-peer>#destination pattern 1..
R1<config-dial-peer>#port 1/0/1

***********ASK DENNIS ABOUT THE CONFIGUATION OF THE 2ND STRUCTURE.

9--------------------------------------------------------------------------------------------------

WILDCARDS

. (dot) matches a single digit.eg 2.. , 5…3, 2….1..1 etc

[] (square brackets) indicates a range of numbers, single digit [2-9].. , [13].. {1,3}, [2-49] {2-4,9}
^ negates a range. Eg 9[^1]…… {any number except 1 including *}
T indicates any number of variable length. Destination-pattern T. wait for inter digit timeout(IT)
# (pound) indicates end of dialing. Eg 973179823#
+ (plus) indicates that the number adheres to E.164 standards.

POTS dial-peers automatically strips any number defined explicitly

Outbound dial peer-ambiguous situation

1. Closest match
2. Explicit preference
3. Random selection

Inbound dial peer-ambiguous situation

1. Incoming called number(DNIS-dialed number identification service)- look for a dial peer that
matches the incoming called number.
2. Answer address(ANI-automatic number indentification)-route call coming from a specific
number to a specific port or other number.
3. Destination pattern(ANI)- compare caller ID to the destination pattern to get a match.
4. Port assignment (NOT CLEAR ASK DENNIS)- route the incoming calls to a port defined in the dial
peer
 5. If all the above methods fail use dial peer 0 (this is the deafult). It can’t be configured.

 Works for any CODEC

 Has VAD (voice activity detection) enabled
 Marks traffic as IP preference 0 ie. No QoS (quality of service)
 Has no RSVP (reservation protocol-reserve a certain amount of bandwidth to get a high
quality of service) support
 Support Fax-rate services.

COMMANDS

R1<config-dial-peer>#preference 1 (set preferences under ambiguous situations)

R1<config>#dial-peer hunt ?
R1<config>#dial-peer hunt 1 (0 is by default)

10--------------------------------------------------------------------------------

ALL COMMANDS

11--------------------------------------------------------------------------------

Hunt group commands

1. Preference
2. Huntstop
3. Dial-peer hunt

The PSTN(call company in this case) gives our router a range of DID numbers. So if anyone call one of
those number through the PSTN(a cell phone maybe) it directly comes to our router which in return
rings out phones according to the dial peers.

BASIC DIGIT MANIPULATION COMMANDS

1. Forward-digits – the forward all the typed digits. Used in case of POTS dial-peer because POTS
strips off exclusively typed numbers

2. Prefix- for adding numbers to the dialed numbers. Eg prefix 986626

3. Num-exp – for translating one number to another.

4. Translation rule.
12--------------------------------------------------------------------------

Some commands to change caller ID

1. Station-ID(name/number) - associated with voice port

2. CLID network-number(number)- associated with dial peer
3. CLID(caller ID) strip name - associated with dial peer
4. CLID restrict - associated with dial peer

Translation rules:-
Allows us to change calling(ANI) or called(DNIS) information. Uses regular expressions to transform.
They are created, put into a profile and then assigned to anything.

Syntax for writing translation rules.

Rule 1 /4000/ /6500/

Wildcards
^ matches beginning of the string
$ matches end of the string eg /^4000$/
. matches single character just like dial-peer
\ removes special meaning eg /\*1/
[] just like dial peer eg 43[2-6]..
? matches character to the left 0 or 1 times
*(asterik) matches character to the left 0 or more times. Eg/.*/ matches blank dialing
+ matches character to the left 1 or more times
|(or) logical or statement eg /4000 | 5000/

& is used to carry the dialed number to the change set

Rule 1 /4…/ /5000/

Rule 2 /^[4-5]…$/ /480847493/
Rule 4 /^7…/ /1808986&/ matches 7123418 and changes to 18089867123. Notice 418 is dropped
Rule 5 /^7…/ /1808986\0/ same as previous. Set 0 matching
Rule 6 /\(9\) \(555…\)/ /\1480\2/ number 9 becomes set 1 and 555… is set 2.
Rule 9 /\(123\)456\(7\)\(8\)/ /\3\3645\1/
Rule 7 /.*/ /1000/ matches blank dialing and changes to 1000
Rule 8 /.+/ /1000/ we have to dial at least one number.

13------------------------------------------------------------------------------
20------------------------------------------------------------------------------

RTP - (real-time transport protocol) it is a protocol responsible for delivering our voice. Our voice is
first converted into digital packets to which the RTP adds its own header which contains information
such as time-stamp (T.S.) and the sequence number of the packet. After this the UDP header is added.
Delivers audio and video streams over data networks.
cRTP – this flavor of RTP strips of redundant information on the packet headers to reduce its size. It
usually strips down a 40 byte packet to merely 2-4 byte packet. However it is recommended only for low
speed connections as it puts a processing toll on the router which may overwhelm it.
sRTP – provides encryption, authentication and integrity services to make our voice packets secure. It is
more effective and flexible than VPN.
VAD – provides bandwidth saving by eliminating voice of silence. It will not send the silent packets
thereby saving bandwidth.

RTCP - (real-time transport control protocol) it is a statistical protocol. It provides statistical information
about an ongoing call such as # of packets dropped, codec, receiver size etc. it goes hand in hand with
the RTP. Has lesser priority than RTP.

H.323, SIP and MGCP are signaling protocols.

H.323 is a heavy protocol. H.323 is more established and is more widely used than SIP. It has gone
through much review and testing phases.

SIP is lightweight as compared to H.323. it is easy to understand as it is very simple. But it also is
evolving, new and not supported by all devices.

H.323 and SIP have peer-to-peer architecture. Meaning that the participating devices viz. gateways and
routers are intelligent. They have all the configurations at their local site. All the dial peer are stored at
their concerned routers/gateways. In other words all the gateways have full blown configuration on
them so that they are intelligent. It does not have a centralized device which acts as a brain for all the
gateways in a network which is the case with MGCP.

MGCP has a client server model in which the server acts as the brain providing all the required
functionality to the gateways or the routers. All the configuration and dial peers are stored at the server
which is the CUCM. Drawback is that if the link between the gateway and the CUCM is down for some
reason then the gateway looses its brain. In order to cope with this the gateways have SRST (survivable
remote site telephony) in which simple configurations are put so as to have a basic call functionality.

VOICE CALL PROCESSING STAGES

1. Sampling
2. Quantization
3. Encoding
4. CODEC compression
5. VoIP encapsulation
 6. Network transport
7. VoIP decapsulation
8. Decoding
9. Modulation

21------------------------------------------------------------------------------------------------------

H.323 has 3 main sub protocols namely

1. H.225 – responsible for call setup/teardown.

2. H.245 – responsible for feature exchange. Feature exchange includes exchange of CODEC info
between gateways so that they agree on a common CODEC etc.

3. H.225 RAS (registration admission and status) – used by gate keeper (GK). The gate keeper holds the
dial plans for the end points (gateways) so as to increase scalability in a peer-to-peer structure because
putting all the files on each end point manually is a tedious job. The gate keeper also provides
bandwidth control. It can be achieved by dividing the network into zones and then assigning each zone a
bandwidth limit. If a zone has exhausted its bandwidth then no more calls can be admitted to it (this is
admission part of the RAS).

H.323 components :-
1. H.323 terminals- these are end points which speak H.323 directly. They don’t need a gate keeper
because H.323 devices are self intelligent.

2. gateways that speak H.323

3. gate keeper (GK)

4. MCU (multi-point control unit) – used for conference calling. It takes in the audio or video streams,
combines them and then sends them out as a single stream to each of the end points

Dial-peers by default use H.323

H.323 call flow

Direct-
H.225 setup
H.225 proceeding
 H.225 alerting (giving the ringtone tone to the calling party)
connect
H.245 (feature exchange. CODEC to be used is decided however if a common ground is not achieved
then the call disconnects after lifting the hand set)
RTP and RTCP
end (245)
release 225

With gate keeper

(GK)ARQ (admission request)
ACF (admission confirm) or ARJ (admission reject)
(GW) H.225 setup
(GK from 2 GW)ARQ
ACF
proceeding
…
DRQ (delete request)
DCF (delete confirm)

In case of multiple GKs we use LRQ (location request) and LCF (location confirm) to find the correct end
point location.

22---------------------------------------------------------------------------------------

Features of SIP (session initiation protocol)

1. Very simple to read and understand.
2. Uses text based ASCII communication. Easy to debug
3. Easy loop detection in SIP in which one GK is connected to another which in turn is connected to
another and at the end a loop is created. Loop detection capabilities are not present in H.323.
4. SIP is more efficient than H.323
5. H.323 maintains call state info for every call making H.323 GK a potential bottleneck.