Transcript Slide 1

CCNA VOICE OFFICIAL EXAM CERTIFICATION
•
•
CHAPTER 7
Gateway and Trunk Concepts
Converting Analog Voice to Digital:
•
The average human can hear frequencies of 20-20,000 Hz
• Human speech uses frequencies from 200-9000 Hz
• Telephone channels typically transmit frequencies of 300-3400 Hz
• The Nyquist theorem is able to reproduce frequencies of 300-4000 Hz
Converting Analog Voice to Digital continued:
• Sample at twice the highest frequency to reproduce accurately
(Nyquist)
• Quantization is the term used to describe the process of
converting an analog signal into a numeric quantity
• Since an eight (8) bit binary number can represent a value from
zero (0) through two-hundred fifty-five (255) we use the Most
Significant Digit (MSD) to represent positive/negative value
• A zero (0) in the MSD represents a positive (+) value
• A one (1) in the MSD represents a negative (-) value
• The result is a range of zero through positive one-hundred
twenty-seven (0 through +127) and negative one through negative
one-hundred twenty-seven (-1 through -127)
1
• Answer: -76
0
1
1
0
1
0
0
Converting Analog Voice to Digital continued:
• Codec’s convert Analog voice into Digital
transmissions.
• Different Codec’s convert in different methods with
more or less complexity
• Available Codec’s:
 G.711
 Internet low Bitrate Codec (iLBC)
 G.729
 G.726
 G.729a
 G.728
• Is the Codec supported in the system
• How many Digital Signal Processors (DSP’s) are used
Converting Analog Voice to Digital continued:
•
•
•
•
Does the Codec meet satisfactory quality levels
How much bandwidth does the Codec consume
How does the Codec handle packet loss
Does the Codec support multiple sample size
Codec’s:






Codec
Bandwidth
Consumed
MOS
G.711
Internet Low
Bitrate Codec (ilBC)
G.729
G.726
G.729a
G.728
64 Kbps
15.2 Kbps
4.1
4.1
8 Kbps
32 Kbps
8 Kbps
16 Kbps
3.92
3.85
3.7
3.61
• MOS (Mean Opinion Score) is determined by listeners listening to the
phrase “Nowadays, a chicken leg is a rare dish.” and scoring the quality
of the connection on a one to five scale.
Calculating Total Bandwidth Needed per Call:
• Determine sample size: A larger sample is more
efficient (Example: 30 bytes of voice to 50 bytes of
overhead 30/80x100%=37.5% is Voice)(Example: 20
bytes of voice to 50 bytes of overhead
20/70x100%=28.5% is voice)
• A larger sample takes longer to prepare, so in
circuits with delay the voice call will not be as good.
• Bandwidth can be saved using Voice Activity
Detection (VAD) where no packets are sent during a
time when there is no voice
• VAD can account for 35-40% of total call time
• RTP header compression does not repeat the header
after the first packet since the information will stay
the same for the length of the call saving 40%
Calculating Total Bandwidth Needed per Call
continued:
• Determine CODEC used
• Determine sample size
• Determine layer overhead
 Layer 2 datalink
 Ethernet:
 Frame-Relay:
 Point-to-point Protocol (PPP):
20 bytes
4-6 bytes
6 bytes
 Layer 3 and 4, network and transport
 IP:
 UDP:
 Real-time Transport Protocol (RTP):
Typically layers 3 and 4 are always 40 bytes
20 bytes
8 bytes
12 bytes
Calculating Total Bandwidth Needed per Call
continued:
•
Bytes-per-packet = (Sample_size * Codec_bandwidth) / 8
• Total_bandwidth = Packet_size * Packets_per_second
• Add



any additional overhead:
GRE/L2TP:
MPLS:
Ipsec:
• Call A:
30 mSec Sample size
G.711 Codec
Ethernet network
24 bytes
4 bytes
50-57 bytes
Call B:
20 mSec Sample size
G.729 Codec
Frame-relay network (4 byte)
Calculating Total Bandwidth Needed per Call
continued:
• Call A:
(.03 * 64Kbps) = 1.92Kbps / 8 = 240 bytes
240 + 20 (ethernet) + 40 (layer 3 and 4) = 300 bytes
300 * (1 / .03) = 10K bytes per second
10K * 8 = 80Kbps
•
Call B:
(.02 * 8Kbps) = 160bps / 8 = 20 bytes
20 + 4 (frame-relay) + 40 (layer 3 and 4) = 64 bytes
64 * (1 / .02) = 3.2K bytes per second
3.2K * 8 = 25.6Kbps
Calculating Total Bandwidth Needed per Call
Compared continued:
• Call B: G.729
(.02 * 8Kbps) = 160bps / 8 = 20 bytes
20 + 4 (frame-relay) + 40 (layer 3 and 4) = 64 bytes
64 * (1 / .02) = 3.2K bytes per second
3.2K * 8 = 25.6Kbps
•
Call B: G.711
(.02 * 64Kbps) = 128Kbps / 8 = 160 bytes
160 + 4 (frame-relay) + 40 (layer 3 and 4) = 204 bytes
204 * (1 / .02) = 10.2K bytes per second
10.2K * 8 = 81.6Kbps
• Savings of 68.6% using the G.729 Codec!
Digital Signal processors:
• DSP’s perform the function of sampling, encoding, and
compression of all audio signals coming into the router.
• DSP’s might be located on the routers motherboard
• DSP’s might also be add on modules similar to SIMM memory
modules on the motherboard called Packet Voice DSP Modules
(PVDM)
• DSP modules can contain multiple DSP circuits
 PVDM2-8: Provides .5 DSP chip
 PVDM2-16: Provides 1 DSP chip
 PVDM2-32: Provides 2 DSP chips
 PVDM2-48: Provides 3 DSP chips
 PVDM2-64: Provides 4 DSP chips
• Codec’s G.711 (a-law and u-law) (u-law is United States,
Japan) (a-law All others), G.726, G.729a, and G.729ab are all
of medium complexity
• Codec’s G.728, G.723, G.729, G.729b and iLBC are all high
complexity
Digital Signal processors:
• To calculate the number of DSP’s needed use the Cisco DSP
calculator http://www.cisco.com/cgi-bin/Support/DSP/dspcalc.pl (Must have Cisco CCO account)
RTP and RTCP:
• Real-time Transport Protocol (RTP) operates at the transport
layer (layer 4) of the OSI model
• Real-time Transport Control Protocol (RTCP) also operates at
the transport layer (layer 4) of the OSI model
• They both work on top of User datagram Protocol (UDP)
• Two transport layer protocols simultaneously working is highly
unusual but is what happens with voice and video!
• UDP works as normal to provide port numbers and header
checksums
• RTP adds time stamps, sequence numbers, and header
information
Data Link
IP
RTP
UDP
Payload
Type
Sequence
Number
Time
Stamp
Audio
Payload
RTP and RTCP continued:
• The payload will specify if the packet is handling voice or
video
• Once established RTP will use even numbered port from
between 16,384 and 32,767
• RTP streams are one-way! If a two-way communication takes
place then a second session is established
• RTCP also engages at the same time and establishes a session
using an odd numbered port from the same range that follows
the RTC even numbered port chosen
• RTCP will account for:
 Packet Count
 Packet Delay
 Packet Loss
 Jitter (delay variations)
• RTP carries the voice while RTCP does the accounting
• RTCP is used to evaluate if there is enough bandwidth or
services to complete a call of good quality
Internet Low Bitrate Codec (iLBC):
• Industry nonproprietary compression codec that is universally
supported
• Developed in 2000 to provide high-quality, bandwidth-savvy,
available to all industry vendors
• Provides a bit rate of 15.2 Kbps when coded using a 20 mSec
sample size, and 13.3 Kbps when using a 30 mSec sample size
• Is a high complexity codec (more DSP required)
• High quality approaching G.711 (64 Kbps). The best of any
compression codec
• Limited support at this time. Cisco phone models that support
iLBC: 7906G, 7911G, 7921G, 7942G, 7945G, 7962G, 7965G,
and 7975G
Trunking the PSTN to CME:
• Foreign Exchange Station (FXS) ports typically connect analog
phones, fax machines, and modems to the CME router
• Foreign Exchange Office (FXO) ports normally connect the
PSTN to the CME router, or PBX system
• Earth and Magneto (E&M) or Ear and Mouth connects from the
PSTN directly to a PBX system
Digital Connections:
• Channel Associated Signaling (CAS) uses robbed bits from the
voice data flow for signaling and control functions. Does affect
the voice quality slightly (in-band-signaling)
• Common Channel Signaling (CCS) uses a separate channel for all
signaling and control functions (out-of-band signaling)
Trunking Connections Between Systems:
• Common language must be used or conversion between
languages
• Available languages are H.323, Session Initiation protocol
(SIP), Media Gateway Control protocol (MGCP), and Skinny Client
Control Protocol (SCCP)
• SCCP is Cisco proprietary
H.323:
• International Telecommunications Union (ITU) accepted in
1996.
• Designed to carry multimedia over Integrated Services Digital
Network (ISDN)
• Based or modeled on the Q.931 protocol
• Cryptic messages based in binary
• Designed as a peer-to-peer protocol so each station functions
independently
• More configuration tasks
• Each gateway needs a full knowledge of the system
• Can configure a single H.323 Gatekeeper that has all system
information
• Each end system can contact the gatekeeper before making a
connection
• Gatekeeper can perform Call Admission Control (CAC) to
determine if resources are available before a call is accepted
• Gatekeeper and Gateway can be the same device
SIP:
• SIP was designed by the IETF as an alternative to H.323
• SIP is a single protocol whereas H.323 is a suite of protocols
as FTP is a single protocol within the TCP/IP protocol suite
• SIP is designed to set up connections between multimedia
endpoints
• Uses other protocols (UDP, RTP, TCP….) to transfer voice or
video data
• Messaging is in clear ASCII text
• Vendors can create their own “add-ons” to the SIP protocol
• SIP is still evolving
• SIP is destined to become the only voice and video protocol
MGCP:
• IETF standard with developmental aid from Cisco
•
•
•
•
•
All devices under a central control
Voice gateway becomes a dumb terminal
Allows minimal local configuration
Single point of failure
Uses UDP port 2427
SCCP:
• Often called “skinny” protocol
• Cisco proprietary
• Similar to MGCP in that it is a stimulus/response protocol
• Allows Cisco to deploy new features in their phones
• Cisco phones must work with Cisco systems (CME,
CUCM,CUCME…)
• Cisco phones can also use other protocols such as SIP or MGCP
with downloaded firmware
Internet Telephone Service Providers:
• New service providers that provide phone services over the
internet (Vonage)
• They interface with the traditional phone service providers
(PSTN)
• Bundle voice and data together
End of Chapter 7