Ensuring Quality in
the Contact CenterJoseph DumontProduct ManagerContact Center ServicesEmpirix, Inc.
Abstract: This presentation will focus on assuring IP telephony voice quality in the contact center and the enterprise. The presenters will address concepts such as active call monitoring and multilayer monitoring, and discuss how these strategies can be used to determine when a failure has occurred or is about to occur in an IP telephony network. The conversation will include a discussion of how performance management reporting can be used to assure that service level objectives are met and how recurring failures can be identified and prevented in real-time.
Proactive multi-channel testing and monitoring of contact center and communication solutions and their role in improving performance and customer experience will also be addressed. This is a shared session with IQ Services. You will each have time for a 20 minute presentation, with a 5 minute Q&A at the end.
• How do I know I have a problem?– Is the voice quality on my
network acceptable?– Is everything working
properly?
• What’s wrong &how do I fix it?– Is increased traffic
effecting voice quality?
– Is something broken? What is it?
– Did something change that effects voice quality?
– How do I get to the source of the problem?
• How do I detect trends and know when my next problem will occur?
– Are things trending toward degrading quality?
– Do I have recurring problems? Every Monday morning? Some other event?
Assuring IPT Voice Quality
LifecycleAssurance
Initial Baseline New
BusinessInitiatives
(ie: CTI,
Speech Rec)
Identifying a Problem
What Could Possibly Go Wrong?User• Acoustic echo; handset speaker to microphone• Hybrid echo; impedance mismatching (handset or local loop)• Distortion; hybrid analog-to-digital conversion and compression
Network• Network faults• Over capacity• Circuit noise; switching• External noise; crosstalk • Signaling protocol mis-matches• Transcoding; compression• Echo suppression• Delay (Latency); packet, speech• Jitter; packet, speech• Out of order packets• Loss of packets• Duplicate packets
IPT Infrastructure Management
ManagementSystem
UC App Servers
http access
SNMP Monitoring
CDRs/CMRs
Backend Systems
Gateway Stats
VVVV
PSTN
Troubleshooting
Troubleshooting
Passive MonitoringWhat is it?
• Insert Network Probe– Locate probes around the network– Probes “listen” to all calls and report call quality
• Solution types– Protocol Analyzers– VoIP specific monitoring probes
Pros• See “real” calls, trace
problems to specific endpoints
• Possible to monitor every call
• Reactive - must wait for a “bad” call to detect a problem
• Difficult troubleshooting - hard to determine what caused the problem
• Expensive – need lots of probes to cover large networks
Cons
Active MonitoringWhat is it?
• Place monitoring calls and measure call statistics– On and off the network
Pros• Simulates a real call,
follows actual call path• Accurate measurements• Proactive – detect
problems early• Extensive testing, IVRs,
CTI, etc.
• Cannot see every call• Takes resources
Bandwidth Directory Numbers
Cons
Infrastructure MonitoringWhat is it?
• Collect call statistics from switches and gateways– CDRs, CMRs– Phone Stats– RTCP collectors
Pros• Inexpensive• Collect information as seen
by the infrastructure – consistent reporting
• Reactive - must wait for calls to all locations
• Difficult to determine what caused the problem
• Limited diagnostic capabilities
Cons
TroubleshootingWhat do I need to test and
monitor?• Infrastructure– Switch / Media Gateway handoffs– Media Gateway echo insertion / loss– Agent “Presence” or registrations– Voice quality over WAN / Toll-bypass links– Conference Bridge Link Voice Quality
• Application– IP IVR embedded CTI– Speech Voice activity detection (VAD) impact on
Speech enabled applications– DTMF Acceptance parameters
(In-band vs. Out-of-band signaling)
Performance Management
Performance ReportingWhat Are the Trends?
• Voice quality getting worse?• Recurring problems?• Poor VQ correlated with other factors?
Summary• Identify a problem– Passive Monitoring– Active Monitoring
• Troubleshoot to find the cause of the problem
• Performance management– Find trends– Identify trouble periods
Supplementary Slides
Ways to Detect Problems
• Users complain• Customers
complain• CEO complains
Much Better
Not Good
• Test before deployment• Passive monitoring• Gather information
from the infrastructure• Active monitoring
Measuring Voice Quality
• Specified by ITU-T P.830 in 1996
• Range of results: 1 to 5• Absolute Category
Rating (ACR) is: 5 is excellent
4 is good
3 is fair
2 is poor
1 is bad
Mean Opinion Score (MOS)
• Individuals provide an ACR score
• Resultant MOS is an average of the scores
Panel of Human Listeners
• Expensive to set up• Cumbersome maintenance• Inconsistent results• Not practical or scalable
for live networks or frequent regression testing
Problems
MOS = (3+4+3+5)/4 = 3.75
IP Voice QualityFactors Impacting Quality
Gateway Gateway/Phone
Lost Packets
CODEC-introduces some distortion
Jitter
Late packets are discardedby jitter buffer
Round trip delay
Passive, Non-reference-Based E-model
Gateway Gateway/Phone
Lost PacketsCODEC-introduces some distortion
Jitter
Late packets are discardedby jitter buffer
Round trip delay
E-model
R Factor
Active, Reference-basedPESQ
FFT FFT
Perceptually weightedObjective score
Test (speech) fileSent thru’ network
500 MIPS processor
PESQ algorithm (ITU P.862) based on and replacesprevious (ITU P.861 PSQM, BT PAMS) algorithms
VoIPService
VoiceQualityTester