In today's competitive environment, many organisations are turning to voice over internet protocol (VoIP) as a way to cut costs, boost productivity and increase competitive advantage. More than 70% of organisations use VoIP in some capacity. It is imperative that IT organisations have the proper tools to monitor the quality of service (QoS) of VoIP calls during and after implementation on the network.
VoIP systems present some special IT challenges. First, because end users expect IP phones to work as reliably as land lines, VoIP systems must meet exceptionally high quality and performance standards. Second, even if the system works fine initially, later network changes could affect call quality — or a growing VoIP system could affect other mission-critical applications.
To minimise problems, you need to verify (before deployment) that your infrastructure can support VoIP, thoroughly examine all system elements during deployment and manage your VoIP system proactively after deployment, including ongoing monitoring, troubleshooting and planning for future growth.
Voice quality metrics generally relate to a listening test, which may not take into account factors such as delay. Typical voice quality metrics are mean opinion score (MOS), perpetual speech quality measure (PSQM) or perceptual evaluation of speech quality (PESQ), or R factor.
Factors that affect voice quality can be grouped into three areas:
- network impairments;
- CODEC (a device or program capable of encoding and/or decoding a digital data stream or signal related impairments);
- environmental factors.
Packets containing speech frames can be lost in transit due to buffer overflow, or can be discarded at the receiver if they arrive excessively late. Packet loss rate is affected by the size of the receiver jitter buffer — a larger buffer will result in fewer discarded packets, but may increase overall delay.
The transit time of a packet through the network will vary for a number of reasons. The jitter buffer at the receiver where incoming packets are delayed, re-ordered and forwarded to the CODEC usually removes this effect. This process, while removing the jitter, may increase packet loss and overall delay.
Low levels of delay, under 100 ms, are not noticeable during a voice call. Larger delays cause the conversation to become disjointed. Delay occurs in initial packetisation, voice compression, transmission through the network and the buffering and decoding process.
CODECs are used to convert the analog voice signal to digital and back. The G711 CODEC provides the best voice quality as it does not compress, is not as sensitive to packet loss and has the least delay. Other CODECs, like G729 and G723, consume less bandwidth by performing compression, but reduce clarity, introduce distortion and delay, and make the voice quality extremely sensitive to lost packets.
Voice activity detection, or silence suppression (where packets that contain silence are not transmitted), can sometimes cause clipping at the beginning of talk bursts. While external background noise is outside of the VoIP system, it can affect the user's perception of quality.
Using VoIP analytics to improve call quality
VoIP analytics can be used to provide end-to-end management data, giving IT managers and frontline support personnel complete vision into the network.
Using analytics, calls can be monitored in real time using advanced algorithms to grade the voice quality being delivered. Real-time QoS assessments can then be generated for every call without the need to perform detailed decoding. User configurable alarms can be used to notify engineers when degradation of dropped calls becomes an issue.
Quality grading thresholds can be set for key VoIP QoS parameters, such as jitter, packet drop and call set-up time. The number of calls that fall within each quality grade can be shown for key QoS parameters. Detailed VoIP call information for every call can be clearly shown in a tabular view, to let you quickly identify the route taken and the gateway involved.
Analytics can be used to recognise and decode analysis all major VoIP protocols including:
- the H.323 suite of protocols, specified by the ITU including Q.931, RAS, H.245 and T.120;
- session initiation protocol (SIP), a signalling protocol for internet conferencing, telephony, event notification and instant messaging. The protocol initiates call set-up, routing, authentication and other feature messages to endpoints within an IP domain;
- skinny client control protocol (SCCP), the proprietary signalling and communications protocol in Cisco's architecture for voice, video and integrated data (AVVID);
- media gateway control protocol, used for controlling telephony gateways from core agents;
- all major CODEC protocols, used for VoIP.
* Wayne Allen is program manager Australia and New Zealand, Fluke Networks.