VOIP between (2) 1552 Wireless Bridge. Should I use G.729?

[Matthew Michaluk]

Hello all,

I have this scenario where we have about 15 cisco phones at a remote site which is pretty much across the street and they are connected via 2 1552 wireless bridges outdoor.  Currently we are using CUCM 8.5 with that region is using G.711.  I started to notice some quality issues between this site.  I was thinking of dropping it down to G.729 as the 1552 Wireless Bridge do not support Qos.  It is best effort only.  Any input is appreciated.

[paolo bevilacqua]

Contrary to common belief, G.729 behaves worst than G.711 in presence of packet loss. A better codec, perhaps, would be iLBC. However I have many customer using wirelss and everything works perfectly without any QoS configures, so I recommend you check the antenna for perfect alignement, etc.

[Matthew Michaluk]

The symptom is the called party will experience dead air in the middle of caller speaking from this remote site.  The user at this site does not noticed though.  The link between the two sites does not have high utilization.  This issue has been a bit tricky.

[paolo bevilacqua]

Update AP software and give network a throught check.

[Matthew Michaluk]

What is your shoot from the hip reaction?  AP and controller software?  Or something in the path?

[paolo bevilacqua]

It can be anything. Sudden one-way voice is not packet loss, but some device that stops forwarding media flow. You have to find which one, that may be not easy, so before taking the forensic route, better put things shipshape anyway.

[Matthew Michaluk]

When doing a packet capture from the affect phones to a phone that is working fine, I noticed that I only see UDP traffic in the capture and not RTP.  Is this to be expected coming from the point to point 1552 wireless bridges which do not have Qos?

[paolo bevilacqua]

RTP transport layer is UDP.

[Matthew Michaluk]

Right.  But normally I see RTP instead of just UDP in the packet capture.  This is kind of telling me that it is losing it markings along the path.  Right?

[paolo bevilacqua]

Not necessarily. Wireshark use euristics to determined what is RTP. DSCP can be set to EF, or not, and the packet still would be a valid RTP.

[Matthew Michaluk]

So if the DSCP is not set to EF it would only should up as UDP in WireShark?  And would the DSCP EF carry over the wireless bridges to the remote end point?

[paolo bevilacqua]

A correctly working euristic would still show it as RTP. This being said I don't know how it would actually show in Wireshark.

Normally network devices do not change DSCP value unless configured to do so.