VoIP monitoring

[Monitoring principles]

General:
VoIP networks are grown rapidly in past decades because of opportunity they are giving. You can easily deploy, manage and extend your VoIP network over an existing IP networks and also it gives you bigger variety of features than regular telephony for the lower cost of services and cheaper price for international destination. Thus all of this benefits, VoIP networks are extremely rigid in front of network quality. In most case if you aim to have a decent voice quality it's strict necessary your network parameters to be on high level network availability in best case scenario up to 99.999% and strict latency. But currently IP networks can't meet this requirements 100%. According to this situation we must be able to notice quality degradation and to take measures to avoid or sort out this issue. That is why we need to monitor strictly quality of our network especially in case when we are using the same network for VoIP communication and data transfer. Therefor exists many methods and tools to detect and estimate quality of VoIP traffic .

Type of VoIP quality estimation:
PESQ: was particularly developed to model subjective tests commonly used in telecommunications (e.g. ITU-T P.800) to assess the voice quality by human beings. Consequently, PESQ employs true voice samples as test signals. In general it’s an complex method to estimate voice quality it use an sample recorded on talker side and compared to sample recorded on listener side this approach gives quite better results in voice quality monitoring but it’s much more expensive and complex to realize than others.
MOS: provides a numerical indication of the perceived quality of received media after compression and/or transmission. The MOS is expressed as a single number in the range 1 to 5, where 1 is lowest perceived audio quality, and 5 is the highest perceived. A drawback of obtaining MOS estimations is that it may be more time-consuming and expensive as it requires hiring experts to make estimations.
R-Factor: A value derived from metrics such as latency, jitter, and packet loss per ITU‑T Recommendation G.107E- model, the R-Factor value helps you quickly assess the quality-of-experience for VoIP calls on your network. Typical scores range from 50 (bad) to 90 (excellent). This values can be easily converted to an MOS equivalent.

[Network parameters which are influenced on voice quality]

In this article we will refer to parameters we can estimate within our network, other parameters related with analog setup at end-user device or hybrid transit are not considering, also we didn't consider analog voice deviation like Voice Activity Detectors , Noise , Echo, External environmental factors etc. The following parameters we can estimate and based on them to calculate VoIP calls qulity.
Latency calculations are supported by synthetic RTP packets (probe requests) which are issued by testing environment at a configurable interval when an RTP flow is active. The header of the synthetic RTP packet contains a timestamps, sequence number which are used to measure the round-trip time between the two packets. For instance, if the sum of measured latency is 500 ms and the number of latency samples is 25, then the average latency is 20 ms.
Packet Loss (%) Calculation of packet loss or packets that are not received usually is performed per RFC 3550 using RTP header sequence numbers. Formula is 100 x (1- packets seen / packets expected), where packets expected = highest RTP sequence number - lowest RTP sequence number +1.
Average Jitter (milliseconds) Jitter, that is, the variation in the delay of received packets in a flow, is measured by comparing the interval when RTP packets were sent to the interval at which they were received. For instance, if packet nr. one and packet nr. two leave 40 milliseconds apart and arrive 65 milliseconds apart, then the jitter is 25 milliseconds.

[Active network monitoring]

Active probes send emulated VoIP traffic through the network and measure the service quality of the network. Currently in our system we are not using an active monitoring system based on probes we are using an IP network monitoring tool called SmokePing (see fig. 1) which is measuring Network quality , its the same way like with RTP probes but in our case ICMP packets are sent. Using this tool we can get information about latency , jittering and packet loss based on which we can estimate network quality, for an good voice quality network must be - measurements for jitter ( < 20 - 50 ms) and latency (< 100 ms) and "zero" measurements for packet loss, duplicated and out-of-order packets. Check network qulity based on ICMP packets it's not the best way to estimate voice quality but it's decent way to estimate network quality by which we can rough determine voice quality.
Fig1 example of graph from Smoke Ping
For big VoIP networks we are recommended to use tool from www.sevana.fi called Asterisk VQM which will provide an great web interface and ability to use VOIP RTP probes for your network active monitoring.

[Passive network monitoring]

Passive probes provide a different perspective of the network as we have with active probes. Passive monitoring is working based on following principles - all VoIP traffic is captured and analyzed. This kind of monitoring tool is searching through the packets on the controlled server or router or any link on which is installed to identify individual VoIP calls and compute the service quality received by each one of them. Passive probes can be used to continuously monitor the performance of the network for the actual VoIP calls traffic, as opposed to the measurements for synthetic traffic using active probes.
Voipmonitor. Currently we are using an tool based on voipmonitor 3.0.1 (consult voipmonitor.org ) traffic sniffer and an web tool developed by us, which can give you an MOS value for each call in particularity calculated based on ITU-T G.107 E-model. The E-model is based on a mathematical algorithm, which get individual transmission parameters that are transformed into different individual "impairment factors" that are assumed to be additive on a psychological scale. The algorithm of the E-model also takes into account the combination effects for those impairments in the connection which occur simultaneously, as well as some masking effects. To the extent that impairments are present for which psychological additivity is not maintained, E-model predictions may be inaccurate.

Tshark. Also you can use tshark which will give you some details regarding network status based on which you can determine quality of Voice service on which you are running this tool. In our day by day work we are using mostly tshark to determine current calls status,you can use an tcpdump filters to get only necessary information.

tshark -n -q -z rtp,streams -f "host xx.xx.xx.xx and host yy.yy.yy.yy" 

Finally you will got following statical information :
============================ RTP Streams ==================================================

Src IP addr Port  Dest IP addr Port  Payload    Pkts Lost Delta(ms) Jitter(ms) MJitter(ms) Prob?

yy.yy.yy.yy 16390 xx.xx.xx.xx  13386 G.711 1038 0 (0.0%)  21.12     0.41         0.27

xx.xx.xx.xx 13386 yy.yy.yy.yy  16390 G.711 1036 0 (0.0%)  31.67     1.25         0.41 

==========================================================================================

Asterisk variables. You can determine call qulity by using Asterisk variables ${RTPAUDIOQOS} and ${RTPAUDIOQOSJITTERBRIDGED} or ${CHANNEL(rtpqos,audio,all)}(in new version) which will give you an information related RTP traffic statistic like loss, jittering and delays – you can easily store these results in CDR database for future analyze. You will receive the following information:

ssrc=254186206;themssrc=2024901615;lp=0;rxjitter=0.020917;rxcount=150;txjitter=0.000000;txcount=83;rlp=0;rtt=14818.715000 

each row explication is below listed , where *src is the source of a stream of RTP packets, identified by a 32-bit numeric SSRC identifier carried in the RTP header so as not to be dependent upon the network address.

• Our Receiver
  - ssrc - our ssrc
  - rxcount - no. received packets/Received packets
  - lp - lost packets/Lost packets rxjitter - our calculated jitter(rx)/Jitter
• Our Sender
  - themssrc - their ssrc
  - txcount - transmitted packets/Sent packet
  - rlp - remote lost packets/Lost packets
  - txjitter - reported jitter of the other end/Jitter
  - rtt - round trip time/RTT

[Advices to improve your VoIP network]

• Upgrade Your Internet Connection and Improve your contention ratio – for this you will need to calculate your band width depend on codec you are using , search for voip bandwidth calculation in google, and consul you INTERNET provider, make needed measurements and improve you link to VoIP service provider.
• Train all of your staff – it's seems obvious, but we should to spend a lot of time on it. Because this is where many VoIP migrations run into a lot of trouble. For example - some thinks that some user of VoIP network may know, but usually not - Pause Any Downloads While On Call.
• If you are using WiFi find useful articles related with Improve Your WiFi Signal or Use a DECT VoIP Phone or if it's possible drop WiFi, Use Ethernet (wire).
• Give voice traffic priority by configuring QoS. VOIP QoS bottleneck and the place that causes most lack of the quality is first outbound link. This is usually the slowest link, being a *DSL or cable modem link of a little as 128k upstream. If you can get out that first link with delay and jitter under control, you can get quality VOIP performance without problem.
• Submit support tickets to Your VoIP Provider and discuss with them possible improvements.

[Conclusion]

Active and passive probes, when used in combination to monitor network performance, give us the capability of effectively monitoring our network for VoIP performance. Estimation of VoIP network has an major benefits regarding thus who are using VoIP,and it gives to you as an system administrator possibility to predict some problem and eliminate them before you clients will notify it and complain about this. It's depends on you which method of VoIP monitoring you will chose but it's quiet important to use it.