12-2 Voice Ports | Cisco Field Manual[c] Router Configuration

Voice ports on routers provide connectivity between the telephony and data networks.
Telephony signaling is used to pass information about call status, voice port status, telephone numbers , and so forth. Signaling is configurable so that the voice port on a router can match the signaling provided by the telephony device.
Analog voice ports connect to analog two-wire and four-wire telephony circuits:
- Foreign Exchange Office (FXO) ports are used to connect to a PSTN central office (CO) or a PBX. They can function as a trunk or tie line. FXO uses a two-wire circuit.
- Foreign Exchange Station (FXS) ports are used to connect to end- user equipment such as a telephone, fax machine, or modem. FXS uses a two-wire circuit.
- E&M (receive and transmit) ports are used as trunk circuits to connect to a telephone switch or PBX. E&M uses a four-wire circuit, with signaling carried over separate wires from the audio.
Digital voice ports use a single digital interface as a trunk:
- Channelized T1 carries 24 full-duplex voice channels (DS0) or timeslots.
- Channelized E1 carries 30 full-duplex voice channels (DS0) or timeslots.
- ISDN PRI carries 23 B channels plus one D channel (North America and Japan) or 30 B channels plus one D channel (the rest of the world). Each B channel can carry voice or data, and the D channel is used for signaling.
Signaling:
- Loop-start signaling is usually used in residential local loops . It detects a closed circuit for going off-hook.
- Ground-start signaling is usually used for PBXs and trunks. It detects a ground and current flow.
- Wink-start signaling begins with the calling side's seizing the line, followed by a short off-hook "wink" by the called side.
- Immediate-start signaling allows a call to begin immediately after the calling side seizes the line. It is used with E&M trunks.
- Delay-dial signaling begins with the calling side's seizing the line and waiting until the called side is on-hook before sending digits. It is used with E&M trunks.
- Common-channel signaling (CCS) is used with a channelized T1 or E1, sending signaling over a dedicated channel.
- Channel-associated signaling (CAS) is used with a channelized T1 or E1, sending signaling within the voice channel itself. Also known as robbed-bit signaling, CAS uses a bit from every sixth frame of voice data to emulate analog signaling.
- QSIG protocol is an ISDN signaling protocol that takes the place of D-channel signaling in some parts of the world.
Trunk connections can be configured to provide simulated trunks between two PBXs over an IP network between two routers.
PSTN call fallback can be used to force call rerouting over other VoIP or POTS dial peers if the network becomes too congested for good voice quality.
Voice port busyout can be used to force a voice port into an inactive state for a variety of conditions.

Configuration

(Optional) Use a digital voice port.
1. Set the Codec complexity.
  - Select the Codec location:
```
 (global)  voice-card   slot  
```
    -OR-
```
 (global)  dspint dspfarm   slot/port  
```
    Codecs are located on a voice-card in the Cisco 2600 and 3600 routers and on a DSP interface ( dspint dspfarm ) in the 7200 and 7500 series routers.
  - Set the complexity:
```
 (voicecard)  codec complexity  {  high   med  } 
```
    -OR-
```
 (dspfarm)  codec  {  high   med  } 
```
    The Codec capabilities are high (supports G.711, G.726, G.729, G.729 Annex B, G.723.1, G.723.1 Annex A, G.728, and fax relay) or med (the default; supports G.711, G.726, G.729 Annex A, G.729 Annex B with Annex A, and fax relay).
    
    Cisco 2600 and 3600 routers can support up to six voice or fax calls per voice card with high and up to 12 voice or fax calls with med. Cisco 7200 and 7500 routers can support up to two voice calls with high and four calls with med.
  NOTE
  
  Cisco AS5300 access servers have Codec capabilities set within the voice feature cards (VFCs). Cisco AS5800 access servers have Codec configuration performed within dial-peer configuration.
2. (ISDN PRI only) Set the ISDN switch type:
```
 (global)  isdn switch-type   switch-type  
```
  The ISDN switch-type must be set to match the switching equipment being used by the telephony provider. In North America, use basic-5ess (Lucent basic rate switches), basic-dms100 (NT DMS-100 basic rate switches), or basic-ni1 (National ISDN-1). In Australia , use basic-ts013 (TS013). In Europe, use basic-1tr6 (German 1TR6), basic-nwnet3 (Norwegian NET3 phase 1), basic-net3 (NET3), vn2 (French VN2), or vn3 (French VN3). In Japan, use ntt (NTT). In New Zealand, use basic-nznet3 (New Zealand NET3).
  
  NOTE
  
  To use QSIG signaling, use a switch-type of basic-qsig.
3. Configure the T1/E1 controller.
  - Select the controller:
```
 (global)  controller  {  t1   e1  }  slot  /  port  
```
    -OR-
```
 (global)  card type  {  t1   e1  }  slot  
```
    A T1/E1 controller is referenced by controller and slot and port number on 2600 and 3600 routers and by card type and slot number on 7200 and 7500 routers.
  - Set the framing type:
```
 (controller)  framing  {  sf   esf   crc4   no-crc4  } [  australia  ] 
```
    The T1 framing type can be sf (Super Frame, the default) or esf (Extended Super Frame). The E1 framing type can be crc4 (the default), no-crc4, and an optional australia.
  - Set the clock source:
```
 (controller)  clock source  {  line  [  primary   secondary  ]  internal  } 
```
    The controller can derive its clock from line (a CO or an external source) or internal (the controller's internal clock). A line clock can be designated as primary (preferred over other controllers' line clocks) or secondary (used as a backup external clock source).
  - Set the line encoding:
```
 (controller)  linecode  {  ami   b8zs   hdb3  } 
```
    For a T1, the line coding can be set to ami (the default) or b8zs (binary 8 zero substitution). For an E1, it can be set to ami or hdb3 ( high-density bipolar 3, the default).
  - (T1 or E1 only) Define a DS-0 group :
```
 (controller)  ds0-group   ds0-group-no   timeslots   timeslot-list   type   type  
```
    Multiple DS-0 channels can be defined as a single group that can be referenced as a logical voice port. The DS-0 group is given a number, ds0-group-no (T1 is 0 to 23; E1 is 0 to 30). The specific DS-0 timeslots are identified as timeslot-list, a comma-separated list of single DS-0s or one or more ranges of DS-0s.
    
    The type field specifies an emulated signaling type and can be e&m-delay-dial, e&m-fgb (E&M type 2 feature group B), e&m-fgd (E&M type 2 feature group D), e&m-immediate-start, e&m-melcas-delay (E&M Mercury Exchange Limited CAS delay start), e&m-melcas-immed (E&M MELCAS immediate start), e&m-melcas-wink (E&M MELCAS wink start), e&m-wink-start, ext-sig (automatically generate off-hook state), fgd-eana (Feature group D Exchange Access North American), fgd-os (Feature group D Operator Services), fxo-melcas, fxs-melcas, fxs-ground-start, fxs-loop-start, none (null signaling for external call control), p7 (the P7 switch type), r1-itu (R1 ITU), sas-ground-start, or sas-loop-start.
  - (ISDN PRI only) Configure PRI parameters.
    
    First, configure the PRI group:
```
 (controller)  pri-group timeslots   range  
```
    The voice timeslots are identified as a range (numbers 1 to 23 or 1 to 30, separated by a dash or comma).
    
    Next , enable voice calls over the PRI:
```
 (global)  interface   slot  /  number   :  [  23   15  ] (interface)  isdn incoming voice-modem  
```
    The D-channel is selected with the 23 (PRI on T1) or 15 (PRI on E1) keyword. Voice calls will be accepted as if they were modem calls.
Configure a voice port.
1. Select the voice port:
```
 (global)  voice-port   slot/subunit/port  
```
  -OR-
```
 (global)  voice-port   slot/port:ds0-group-no  
```
  A voice port (any type) is selected according to its physical position in the router. Analog ports are referenced by slot/subunit/port, and digital ports are referenced by physical position and the logical DS0 group number, as in slot/port: ds0-group-no.
2. (Optional) Enter a description for the port:
```
 (voiceport)  description   text-string  
```
3. (Analog ports only) Specify the signaling type:
```
 (voiceport)  signal  {  loop-start   ground-start   wink-start   immediate-start   delay-dial  } 
```
  For FXS or FXO, choose loop-start (the default) or ground-start. For E&M, choose wink-start (the default), immediate-start, or delay-dial.
4. (Optional) Specify the call progress tone locale:
```
 (voiceport)  cptone   locale  
```
  The locale is given as a two-letter ISO3166 value ( us is the default).
5. (Optional) Configure the E&M interface.
  - (Analog ports only) Specify the number of wires:
```
 (voiceport)  operation  {  2-wire   4-wire  } 
```
    The 2-wire circuit is the default.
  - Specify the type of circuit:
```
 (voiceport)  type  {  1   2   3   5  } 
```
    The E&M interface can be one of the types shown in Table 12-3.
6. (Optional) Configure ringing operation.
  - (FXS only) Set the ring frequency:
```
 (voiceport)  ring frequency  {  25   50  } 
```
    The ring frequency is set in hertz, to 25 (the default) or 50.
  - (FXS only) Set the ring cadence:
```
 (voiceport)  ring cadence  {[  pattern01   pattern02   ... pattern12  ]    [  define   pulse interval  ]} 
```
    The ring pattern for incoming calls can be set to one of the 12 predefined patterns (the default is selected by the cptone locale) or to a user-defined pattern with the define keyword. The ring cycle is given by pulse ( on-time in hundreds of milliseconds ; 1 to 50) and interval (off-time in hundreds of milliseconds; 1 to 50).
  - (FXO only) Set the number of rings before answering:
```
 (voiceport)  ring number   number  
```
    The router will answer an incoming call after number (1 to 10; the default is 1) rings.
7. (Optional) Use disconnect supervision to detect a disconnected call.
  - Select the disconnect supervision type.
    
    Detect battery reversal (analog ports only):
```
 (voiceport) [  no  ]  battery-reversal  
```
    FXO ports reverse battery upon call connection unless the no keyword is used. FXS ports with loop-start detect a second battery reversal to disconnect a call (the default) unless the no keyword is used.
    
    Detect supervisory disconnects (FXO only):
```
 (voiceport) [  no  ]  supervisory disconnect  
```
    A CO switch normally drops line power for at least 350 milliseconds to signal a call disconnect. The FXO port detects this (the default) unless the no keyword is used.
    
    Use disconnect acknowledgment (FXS only):
```
 (voiceport) [  no  ]  disconnect-ack  
```
    After an FXS port detects a disconnect, it returns an acknowledgment by dropping line power (the default). Use the no keyword to disable the acknowledgment.
  - (Analog FXO only) Configure supervisory disconnect tones.
    
    Create a voice class that contains the tone settings:
```
 (global)  voice class dualtone   tag  
```
    The voice class is labeled as tag (1 to 10000). It contains the parameters for the dual disconnect tones.
    
    Set the disconnect tone frequencies:
```
 (voice-class)  freq-pair   tone-id frequency-1 frequency-2  (voice-class)  freq-max-deviation   frequency  
```
    With freq-pair, the pair of tones is given a unique tone-id (1 to 16) and is set to frequency-1 and frequency-2, in Hz (300 to 3600 Hz). frequency-2 can be set to 0, but random single tones can cause inadvertent disconnects. The freq-max-deviation keyword sets the maximum frequency deviation that will be detected (10 to 125 Hz; the default is 10 Hz).
    
    Set the tone power:
```
 (voice-class)  freq-max-power   dBm0  (voice-class)  freq-min-power   dBm0  (voice-class)  freq-power-twist   dBm0  
```
    The minimum tone power is given by freq-min-power (10 to 35 dBm0; the default is 30). The maximum tone power is freq-max-power (0 to 20 dBm0; the default is 10). The power difference between the two tones is given by freq-power-twist (0 to 15 dBm0; the default is 6).
    
    Set the cadence for a complex tone:
```
 (voice-class)  cadence-min-on-time   time  (voice-class)  cadence-max-off-time   time  (voice-class)  cadence-variation   time  (voice-class)  cadence-list   cadence-id cycle1-ontime cycle1-offtime  cycle2-ontime cycle2-offtime cycle3-ontime cycle3-offtime   cycle4-ontime cycle4-offtime 
```
    The tone is specified as a minimum on time ( cadence-min-on-time; 0 to 100 milliseconds), a maximum off time ( cadence-max-off-time; 0 to 5000 milliseconds), and a maximum variation in detectable on time ( cadence-variation; 0 to 200 milliseconds). The cadence pattern can be specified with a unique cadence-id (1 to 10) and an on-off pattern of four cycle times each (0 to 1000 milliseconds; the default is 0).
    
    Detect supervisory disconnect tones with the voice class:
```
 (voiceport)  supervisory disconnect dualtone  {  mid-call   pre-connect  }  voice-class   tag  
```
    -OR-
```
 (voiceport)  supervisory disconnect anytone  
```
    Tone detection is enabled on the voice port using the voice class tag for tone definitions. Disconnects can be detected during a call ( mid-call ) or only during call setup ( pre-connect ). If the PSTN or PBX cannot provide a disconnect tone, the anytone keyword can be used instead. Any tone used during call setup (busy or dial tone) causes the call to be disconnected.
8. (Optional) Set timeout values:
```
 (voiceport)  timeouts   type value  
```
  The timeout parameters can be set with the following type and value : call-disconnect seconds (0 to 120 seconds; the default is 60), initial seconds (the maximum time between the first and next dialed digit; 0 to 120 seconds; the default is 10), interdigit seconds (the maximum time between dialed digits; 0 to 120 seconds; the default is 10), ringing { seconds infinity } (the time that an outbound call is allowed to ring before disconnecting; 3 to 3600 seconds; the default is 30, or infinite with no disconnect), or wait-release { seconds infinity } (the maximum time that a busy, reorder, or out-of-service tone is sent for a failed call; 3 to 3600 seconds; the default is 30, or infinite).
9. (Optional) Set timing parameters:
```
 (voiceport)  timing   type milliseconds  
```
  An E&M voice port can be fine- tuned with the following timing type, measured in milliseconds: clear-wait (the minimum time between inactive seizure and call clearing; 200 to 2000 ms; the default is 400), delay-duration (the duration of delay-dial signaling; 100 to 5000 ms; the default is 2000), delay-start (the minimum time between outgoing seizure and outdial address; 20 to 2000 ms; the default is 300), pulse (the pulse dialing rate in pulses per seconds; 10 to 20; the default is 20), pulse-interdigit (pulse dialing interdigit timing; 100 to 1000 ms; the default is 500), wink-duration (the maximum wink signal duration; 100 to 400 ms; the default is 200), or wink-wait (the maximum wink wait duration; 100 to 5000 ms; the default is 200).
  
  An FXO voice port can be tuned with the following type, measured in milliseconds: guard-out (how long to wait before seizing a remote FXS port; 300 to 3000 ms; the default is 2000), pulse (the pulse dialing rate in pulses per second; 10 to 20; the default is 20), pulse-digit (the pulse digit signal duration; 10 to 20 ms; the default is 20), or pulse-interdigit (the pulse dialing interdigit timing; 100 to 1000 ms; the default is 500).
  
  Any voice port can be tuned with the following type, measured in milliseconds: dial-pulse min-delay (the time between pulse dialing pulses; 0 to 5000 ms; the default is 300), digit (the DTMF digit duration; 50 to 1000 ms; the default is 100), interdigit (the DTMF interdigit duration; 50 to 500 ms; the default is 100), or hookflash-out (the hookflash duration; 300 to 3000 ms; the default is 300).
10. (Optional) Use Voice Activity Detection (VAD) to reduce bandwidth.
  - Set the music-on-hold threshold:
```
 (voiceport)  music-threshold   dB  
```
    The minimal level of music played on hold can be set to dB (70 to 30 dB; the default is 38) so that VAD is triggered to play the audio.
  - Enable comfort noise generation:
```
 (voiceport)  comfort-noise  
```
    During the silent gaps when VAD doesn't detect a voice, a subtle background noise is played locally on the voice port (this is enabled by default). If this feature is disabled, the silence can make the caller think the call has been disconnected.
11. (Optional) Tune the voice quality.
  - Adjust the jitter buffer.
    
    Set the jitter buffer playout mode:
```
 (voiceport)  playout-delay mode  {  adaptive   fixed  } 
```
    The jitter buffer can operate in two modes: adaptive (the buffer size and delay are adjusted dynamically; this is the default) or fixed (a fixed buffer size; the delay doesn't change). Adaptive mode dynamically adjusts the jitter buffer according to current or changing network conditions.
    
    Rather than configuring the mode on the voice port ( affecting all dial peers using the voice port), you can configure the mode for specific dial peers.
    
    Set the jitter buffer parameters:
```
 (voiceport)  playout-delay  {  maximum   nominal  }  milliseconds  
```
    The jitter buffer can be set for a playout delay of maximum (the upper limit; the default is 160 ms) or nominal (the playout delay used at the beginning of a call; the default is 80 ms) for a delay of milliseconds (40 to 320 milliseconds).
  - Adjust the echo canceler.
    
    Enable the echo canceler:
```
 (voiceport)  echo-cancel enable  
```
    By default, echo cancellation is enabled on all voice interfaces. If it is disabled, the callers might hear an audible echo.
    
    Set the maximum echo cancel duration:
```
 (voiceport)  echo-cancel coverage  {  8   16   24   32  } 
```
    The echo canceler covers a fixed window of the call audio. The window size can be set to 8, 16 (the default), 24, or 32 milliseconds. The coverage window can be made larger if you hear an audible echo.
    
    Use nonlinear echo cancellation:
```
 (voiceport) [  no  ]  non-linear  
```
    By default, the echo canceler uses a nonlinear operation ( residual echo suppression). The nonlinear computation attenuates the signal when a near-end speech (the end of a word) is detected. Use the no keyword to return to linear mode if desired.
  - Adjust the voice level.
    
    Set the input gain:
```
 (voiceport)  input gain   value  
```
    The voice port adjusts the amount of gain at the receiver side to value (6 to 14 decibels; the default is 0). The default value is used to achieve a 6 dB attenuation between phones.
    
    Set the output attenuation:
```
 (voiceport)  output attenuation   value  
```
    The voice port adjusts the amount of attenuation on the transmit side to value (6 to 14 decibels; the default is 0).
    
    Set the voice port impedance (FXO only):
```
 (voiceport)  impedance  {  600c   600r   900c   complex1   complex2  } 
```
    An FXO voice port can be terminated with an impedance of 600c (600 ohms complex), 600r (600 ohms real, the default), 900c (900 ohms complex), complex1, or complex2. Choose a value that matches the specifications of the telephony provider or equipment.

Table 12-3. E&M Interface Types

	E-lead (Output)	M-lead (Input)	Signal Battery	Signal Ground
Type 1 (default)	Relay to ground	Referenced to ground
Type 2	Relay to signal ground	Referenced to ground	Feed for M; connected to 48V	Return for E; isolated from ground
Type 3	Relay to ground	Referenced to ground	Connected to 48V	Connected to ground
Type 5	Relay to ground	Referenced to 48V

(Optional) Use trunk connections with a voice port.
1. Set the trunk-conditioning signaling.
  - Create a voice class as a template:
```
 (global)  voice class permanent   tag  
```
    Identify the voice class with a unique tag (1 to 10000).
  - Set the trunk keepalive interval:
```
 (voice-class)  signal keepalive   seconds  
```
    A keepalive packet is sent at intervals of seconds (1 to 65535; the default is 5 seconds) to the far end of the trunk.
  - Define the signaling sequence that is sent to the PBX:
```
 (voice-class)  signal sequence oos  {  no-action   idle-only   oos-only   both  } 
```
    When a keepalive packet is lost or an AIS message is received from the far end, the router sends a sequence of signaling messages: no-action (no signaling is sent), idle-only (only the idle signal pattern is sent), oos-only (only the out-of-service [OOS] pattern is sent), or both (both idle and OOS patterns are sent; this is the default).
  - Define signaling patterns for idle and OOS states:
```
 (voice-class)  signal pattern  {  idle receive   idle transmit   oos receive   oos transmit  }  bit-pattern  
```
    The signaling pattern for the following conditions is defined as bit-pattern (ABCD, as four 0 or 1 digits): idle receive (an idle message from the network), idle transmit (an idle message from the PBX), oos receive (the network is out of service), or oos transmit (PBX is out of service). The defaults for a near-end voice port are idle receive (E&M: 0000 T1 or 0001 E1; FXO: 0101 loop start or 1111 ground start; FXS: 0101; MELCAS: 1101), idle transmit (E&M: 0000; FXO: 0101; FXS: 0101 loop start or 1111 ground start; MELCAS: 1101), oos receive (E&M: 1111; FXO: 1111 loop start or 0000 ground start; FXS: 1111 loop start or 0101 ground start; MELCAS: 1111), and oos transmit (none).
  - (Optional) Restart a permanent trunk after it has been OOS:
```
 (voice-class)  signal timing oos restart   seconds  
```
    A trunk can be torn down and restarted seconds (0 to 65535) after it has been out of service. By default, trunks are not restarted.
  - (Optional) Return a trunk to standby state after it has been OOS:
```
 (voice-class)  signal timing oos slave-standby   seconds  
```
    A trunk can be returned to its initial standby state seconds (0 to 65535) after it has been out of service. By default, trunks are not returned to the standby state.
  - (Optional) Stop sending packets if the PBX signals an OOS:
```
 (voice-class)  signal timing oos suppress-all   seconds  
```
    If the PBX signals on OOS condition for seconds duration (0 to 65535), the router can stop sending voice and signaling packets. By default, the router does not stop sending.
  - Apply the voice class to a voice port:
```
 (voice-port)  voice-class permanent   tag  
```
    The trunk-conditioning signaling voice class is applied as a template to the voice port.
2. Define the type of trunk connection on a voice port:
```
 (voice-port)  connection  {  plar   tie-line   plar-opx  }  digits  {  trunk   digits  [  answer-mode  ]} 
```
  The trunk connection can be configured as plar (private line automatic ringdown; the caller goes off-hook and digits is automatically dialed), tie-line (a tie-line trunk to a PBX; it is automatically set up and torn down for each call when you dial digits ), or plar-opx (PLAR off-premises extension; an FXO port does not answer until the remote end at digits answers).
  
  The trunk keyword is used to create a permanent trunk between two PBXs connected by two routers. The number digits is dialed to reach the far end of the trunk. If the answer-mode keyword is given, the router waits for an incoming call before initiating the trunk connection. Otherwise, the trunk will be brought up permanently.
(Optional) Use PSTN fallback for call routing during network congestion.
1. Enable call fallback:
```
 (global)  call fallback active  
```
  The router samples the H.323 call requests and attempts to use alternative dial peers if the network congestion is above a threshold.
2. (Optional) Use call fallback for statistics gathering instead of true fallback:
```
 (global)  call fallback monitor  
```
  As soon as the router has gathered fallback statistics, you can display them for planning purposes. Use the show call fallback stats command to see the results.
3. (Optional) Use MD5 encryption keys for fallback probes:
```
 (global)  call fallback key-chain   name-of-chain  
```
  Fallback uses Service Assurance Agent (SAA) probes to determine the state of the network. Use the MD5 key chain named name-of-chain if you are configuring the SAA responder at the far-end router to use MD5. Refer to Section 1-7 for more information about SAA and its configuration.
4. Set the jitter probe parameters.
  - Set the number of jitter packets:
```
 (global)  call fallback jitter-probe num-packets   packets  
```
    The fallback jitter probe uses the specified number of packets (2 to 50; the default is 15 packets). Increase the number of packets to get a better idea of true network conditionsat the expense of additional bandwidth used for the probes.
  - Set the IP Precedence to use on jitter probes:
```
 (global)  call fallback jitter-probe precedence   precedence  
```
    The router can set the IP Precedence value in each jitter probe packet to precedence (0 to 7; the default is 2). The IP Precedence of true VoIP packets is usually set to 5. Setting the probe packets to a more realistic Precedence value helps you measure the conditions that voice packets actually experience.
  - Force the jitter probes to use a strict priority queue:
```
 (global)  call fallback jitter-probe priority-queue  
```
    By default, jitter probes are sent without using a priority queue. If you have a strict priority queue configured using LLQ, the jitter probe packets are sent over the IP RTP priority queue regardless.
  - Adjust the SAA probe timeout:
```
 (global)  call fallback probe-timeout   seconds  
```
    After a timeout period of seconds (1 to 2147483 seconds; the default is 30) elapses without a response to a probe packet, another probe is sent.
  - Use an average of two probes for congestion calculations:
```
 (global)  call fallback instantaneous-value-weight   weight  
```
    Normally, network congestion is measured based on the results of one probe. You can use the weighted average of the current probe with the previous probe to get a more gradual fallback recovery during heavy congestion. The weight (0 to 100; the default is 66 percent) is a percentage used for the current probe so that it can be weighted more than the previous probe statistics.
  - Set the fallback thresholds.
    
    For trigger fallback based on packet delay and loss:
```
 (global)  call fallback threshold delay   delay-value   loss   loss-value  
```
    The fallback threshold is set if the end-to-end delay rises above delay-value (1 to 2,147,483,647 milliseconds; there is no default) and if the percentage of packet loss rises above loss-value (0 to 100 percent; there is no default). The lower you set these values, the higher your expectations for high-quality voice.
    
    For trigger fallback based on the ICPIF threshold:
```
 (global)  call fallback threshold icpif   threshold-value  
```
    The Impairment/Calculated Planning Impairment Factor (ICPIF) threshold is calculated, producing an impairment factor for every network node along a probe's path . With fallback, the ICPIF is calculated using packet loss, delay, and the type of Codecs used. Call fallback is triggered if the ICPIF value rises above threshold-value (0 to 34; the default is 5). The lower the ICPIF value, the better the voice quality. Beware of setting the value to 34, because this indicates 100% packet loss.
5. Enable SAA on the far-end router:
```
 (global)  saa responder  
```
  At a minimum, you must enable the SAA responder on a far-end router so that the local router will receive SAA response packets to the jitter probes.
(Optional) Use local voice busyout.
1. (Optional) Busyout all voice ports on a serial interface:
```
 (interface)  voice-port busyout  
```
  If desired, you can force all voice ports that use the interface into a busyout conditionexcept those configured for specific busyout arrangements.
2. (Optional) Busyout a voice port based on the state of an interface:
```
 (voice-port)  busyout monitor  {  serial   interface-number   ethernet   interface-number  } [  in-service  ] 
```
  A voice port can be configured to monitor the up/down state of a physical interface on the router. When the interface, either serial or ethernet, is up, the voice port is usable. When the interface is down, the voice port moves to busyout so that the calls will be rerouted over another path. Also, you can configure the voice port to busyout when the interface is up by using the in-service keyword.
3. (Optional) Busyout with a seize condition:
```
 (voice-port)  busyout seize  {  ignore   repeat  } 
```
  During busyout, a voice port can be configured to seize the line and either ignore (stay in the busyout state regardless of the reaction of the far end) or repeat (go into busyout; if the far end changes state, cycle back into busyout again).
4. (Optional) Force a voice port busyout:
```
 (voice-port) [  no  ]  busyout forced  
```
  To unconditionally busyout a voice port, use busyout forced. The voice port will stay in the busyout state until the no busyout forced command is used.
5. (Optional) Busyout a voice port according to network conditions:
```
 (voice-port)  busyout monitor probe   ip-address  [  codec   codec-type  ] [  icpif   icpif   loss   percent   delay   milliseconds  ] 
```
  SAA jitter probes are issued toward a target router at ip-address. The probes can mimic certain types of Codecs by using the appropriate packet size. Specify the codec-type as g711a (G.711 A-law), g711u (G.711 U-law, the default), g729 (G.729), or g729a (G.729 Annex A). The ICPIF loss/delay threshold can be specified to trigger a busyout condition based on network congestion. The probes are used to collect active information about end-to-end packet delay and loss. The icpif value (0 to 30) should be chosen to reflect a threshold of poor voice quality. Lower values mean less delay and packet loss.
  
  Otherwise, specific packet loss and delay thresholds can be specified to trigger busyout. If the packet loss measurement rises above percent (1 to 100) or if the end-to-end delay rises above milliseconds (1 to 2147483647), the voice port enters the busyout state.