I am getting voice quality issues off FXS lines on the TA924e. When I do a show voice quality-stats command I am seeing numerous out of order counts on each of the calls. Can you explain what this is detailing? and can this have an affect on voice quality issues that we are experiencing?
Thank you,
Richard D.
Hi Richard:
You're likely to get a much better answer from an ADTRAN engineer soon, but out of order packets in an audio stream will definitely relate to the perceived quality. Specifically, choppiness where parts of a caller's words are missing.
'Out of order' means that some RTP (audio) packets aren't arriving in a timely manner. Every RTP packet has a sequence number so that the TA924e can detect when some are missing or out of order. Generally, packets arriving late--prior to one or more packets arriving with a lower sequence number--are discarded (at least with codecs I'm familiar with).
Do you think there might be a general network issue somewhere between the TA924e and the SIP server/gateway it communicates with? What is the path between them? Is there QoS in place?
Best,
--
CJ
Hi Richard:
You're likely to get a much better answer from an ADTRAN engineer soon, but out of order packets in an audio stream will definitely relate to the perceived quality. Specifically, choppiness where parts of a caller's words are missing.
'Out of order' means that some RTP (audio) packets aren't arriving in a timely manner. Every RTP packet has a sequence number so that the TA924e can detect when some are missing or out of order. Generally, packets arriving late--prior to one or more packets arriving with a lower sequence number--are discarded (at least with codecs I'm familiar with).
Do you think there might be a general network issue somewhere between the TA924e and the SIP server/gateway it communicates with? What is the path between them? Is there QoS in place?
Best,
--
CJ
Richard:
Thank you for your question.
Out of Order Packets can cause voice quality problems in the audio path destined to the ADTRAN unit. The quality problems can be short cut-outs in audio or more severe such as no inbound audio at all.
Below is a sample output from “show voice quality-stats”
Start Out of Discard Delay
ID Time From To Length Codec Order Pkts* Avg Max
--------------------------------------------------------------------------------------------
5989 12:12 PM 5082 3650 4:56 G711u 7 0 50 50
6003 12:14 PM 5080 3123 1:11 G711u 2 0 50 100
6004 12:14 PM 1002 3665 0:38 G711u 0 0 60 90
6005 12:15 PM 3665 8205 0:20 G711u 0 0 56 80
6006 12:15 PM 3665 1002 0:22 G711u 0 0 70 92
6007 12:15 PM 8203 5080 0:59 G711u 0 0 50 50
6008 12:15 PM 3665 5081 1:35 G711u 2 0 51 60
6009 12:15 PM 3669 5562 0:22 G711u 2 0 65 85
6010 12:15 PM 8998 1002 0:25 G711u 0 0 50 50
ID
The local call-id used internally by the switchboard to differentiate call legs.
Start Time
The starting time of the call based on the system clock.
From
The calling party number.
To
The called party number.
Length
The length of time the call is connected.
Codec
The voice codec selected for the call.
Out of Order Packets
An out-of-order packet is counted if the sequence number received is greater than the sequence number expected; such a packet will count as out-of-order AND will also be counted as discarded if it cannot be played out in time. Common causes for these packets are QoS and voice traffic delivery problems to the ADTRAN or duplicate rtp streams destined for the same endpoint in the ADTRAN. In the case of duplicate rtp streams, “ip rtp symmetric filter” is explained at the end of this answer.
Discarded Packets
If a packet arrives before or after the window of the jitter buffer (based on timestamp), it will be considered an early or a late arrival. All early and late arrivals will are logged as discarded packets and will not be played out. If the jitter buffer is set to adaptive and the IAD continues to log packets outside of the jitter window, the DSP will increase the window size in 10ms increments.
Delay (Avg Max)
The Avg and Max delay indicate the size of the jitter buffer. By default, the jitter buffer in the IAD is set to adaptive with a nominal depth of 50ms. If the IAD continues to log packets outside of the 50ms window, the jitter buffer depth will increase up to a max of 100ms by default. The nominal and max jitter buffer size can be configured per FXS/SIP user or E&M/PRI trunk.
TA912(config-9001)#rtp packet-delay nominal 50
TA912(config-9001)#rtp packet-delay maximum 100
The jitter window can also be set to a fixed value by changing the delay mode to “fixed” and configuring the nominal packet delay to the desired window size.
TA912(config-9001)#rtp delay-mode fixed
TA912(config-9001)#rtp packet-delay nominal 150
Note:
ip rtp symmetric-filter
This helps in instances when there are multiple rtp streams with the same ip and port destined for the same endpoint in the ADTRAN. This can happen if a call, that is currently inactive, is still active somewhere in the network. The out-of-order packets will increment in this case. The end result can be 1-way audio.
If you have any additional questions, please let me know.
Geoff
Do out of order packets cause voice quality issues if the jitter buffer plays out in time ?
Does the jitter buffer ever reorder packets before audio play out?
Thanks,
Bill
Hello Bill,
The jitter buffer will try to reorder the out of order packets before they are played out. If the packet sequence numbers are really far off, the jitter buffer may not have time to place the out of order packets in sequence for playback. As a result those packets are dropped.
Generally out of order packets cause audio issues when there is delay involved.
Regards,
Geoff
Thanks for the help, I appreciate it!
Glad to help
Geoff