Understanding call quality for Internet voice applications
It’s been a long time since Voice over Internet protocol (VoIP) became an alternative to traditional telephony. Initially used by enterprises to by-pass expensive national and international call routes VoIP became the mainstream when consumers started to use it, and subsequently small to mid-sized business in the form of telephony solutions such as Cloud PBX. For many people, quality is still a common question when considering voice over IP services. The technology itself was designed with quality of service in mind however, a lot early adopters of VoIP technology failed to configure their services well enough.
This poor experience is well described by Gartner’s Hype Cycle in the trough of disillusionment where early adopters view the service as sub-standard.
Nowadays IP telephony and Cloud PBX can deliver much better quality than traditional telephony. Faster and larger links, technology improvements, new high definition wideband codecs [such as G.722] and better network designs almost eliminate quality problems in VoIP solutions. This allows telephony to move into the cloud along with the other services making it an attractive and scalable business phone system option for businesses of all sizes.
Delivering voice over the Internet in some situations may still cause voice quality issues especially if transported over slow and congested ADSL links with oversubscribed DSLAMs. Being able to monitor the quality of every call to ensure call quality provides transparency and peace of mind for customers, it also allows quick identification of potential issues in a network that can affect call quality.
Access4’s approach is to monitor voice quality across its Unified Communications platform for Cloud PBX and Hosted Contact Centre users. For Access4 partners access to the same quality data is available for their customers. This real-time information helps in making sure any potential problems are proactively identified.
The voice call quality dashboard in SASBOSS™ simplifies monitoring and also facilitates a faster troubleshooting process where partners need to escalate to Access4 technical support. The tool provides access to specific call data and unique call identifiers to make the triage process considerably easier for the partner and improve the customer’s experience. Access4’s Unified Communications solutions now interrogate phones and softphones connected to the BroadSoft® platform to provide feedback on call quality at the end of each call.
As with many things, call quality can be subjective and has historically been measured by people giving their opinion of call quality on a scale from 1 to 5. This became known as the Mean Opinion Score (MOS).
MOS became an industry standard for measuring phone call voice quality. In VoIP networks, MOS is calculated based on different parameters such as packet loss, round trip delay, jitter, the codec used, signal and noise levels, etc. In reality, MOS’ never reach scores of five (5) however values between four (4) and five (5) indicate very good quality. Despite MOS being a calculated value, it is still an opinion score which means its absolute value cannot be used alone without a trending view to understanding changes over time and identifying minimum and maximum values. For example, three phones from three different vendors, connected to the same network next to each other could report slightly different quality values. If all the phones consistently report scores of 4.1 but today one phone reports a score of 3, the customer will be experiencing the poorer quality and perhaps something is wrong with their network or connection.
Calculated MOS is reported by the phone via RTCP-XR SIP Publish messages back to Access4’s centralised collector that consolidates all the reports into a database for further processing. Phones report two (2) MOS’ – one for listening quality (MOS-LQ) and one for conversational quality (MOS-CQ). Listening quality reflects the quality for a person who listens on the phone and conversational quality reflects an overall quality. Splitting LQ and SQ helps to determine the reason for the bad quality. If LQ quality is low while CQ is higher that means the network is not handling voice towards the phone while upstream is fine. If LQ is fine but CQ is affected you may question network performance in the other direction, from the phone to the cloud platform. In an ideal network environment, CQ and LQ should be identical.
SASBOSS™ call quality feature aggregates all the reports for individual calls and visually graphs the average call quality for a requested period for all your customers, individual enterprises or even at a single service level. Extended call quality Call Detail Records (CDRs) can be pulled for individual services for detailed analysis. MOS is only reported for the calls longer than 10 seconds; however detailed CDRs are available even for the calls that did not get an RTCP-XR report back from the phone.
How to use the SASBOSS™ quality monitoring feature:
There are a number of different ways to access voice quality graphs in SASBOSS™; through the call quality dashboard available under the ‘Dashboard’ main menu, through enterprise “actions” menu by selecting “view call quality” to view quality metrics for that enterprise; or through service “actions” menu to view voice quality of a particular service.
Call quality graphs will aggregate RTCP-XR reports for a specified timeframe and show the number of calls used to render the graph.
Each point can be hovered over to obtain additional information such as exact values calculated for the specific point or zoomed in by clicking on it. Once clicked, the graph will automatically select a narrower timeframe and graph individual calls.
More detailed information is available by clicking on the data points for individual calls.
This data record allows quick identification of the customer, group and service the call belongs to as well as many other factors that can affect the quality of experience:
- Phone vendor and model – some devices provide better quality
- Firmware version – we recommend to keep the phones on the latest Access4 supported version;
- Device IP address (and NAT IP if the device is behind NAT) – it is important to understand if there is a firewall between the phone and SBC as firewalls need to be configured to support VoIP properly;
- SBC IP address – the actual IP address where their RTP stream carrying the voice was sent to or received from, this information may help network engineers to investigate network issues related to voice quality;
- MOS LQ and CQ.
If a partner has a call issue that requires escalation to Access4, they can easily do so by clicking email support in the ‘Call Grid Data Record’. SASBOSS™ will populate all the important data removing the data entry error including the unique reference to the call. This allows Access4 engineers to go directly to the call speeding up the resolution process for customers.