Quantifying Skype User Satisfaction

1
Quantifying Skype User Satisfaction

• CHEN, Hanliu
• Sept 2, 2006

2
Key Issue

• Skype a successful peer-to-peer application,
  which is in many ways bringing a revolution in
  voice communications, and driving Service
  Providers crazy in their effort to identify Skype
  calls using their facilities for free.
• Key issue of this paper the quantification of
  Skype user satisfaction
• Major challenge the lack of an easily accessible
  and objective index to quantify the degree of
  user satisfaction

3
3 contributions of this paper

• An objective and perceptual user satisfaction
  index is devised, in which the parameters are all
  easily measurable and computable online
• the index is validated with an independent set of
  metrics for voice interactivity derived from user
  conversation patterns
• quantifying the influence of the bit rate, jitter
  and loss, RTT on call duration, which provides
  hints about the priority of the metrics to tune
  for optimal user satisfaction with Skype

4
QoS factors

• Based on a solid statistical study (including
  survival analysis, correlations, regression
  analysis and hypothesis testing), the paper
  conclude that
• Call duration is a key measure of user
  satisfaction (short calls are usually bad, since
  the cost is not an issue here!)
• The most important QoS measures impacting the
  quality of a Skype call (its duration)
• (1) source bit rate
• (2) jitter (measured by standard deviation
  of the bit rate sampled every second at the
  receiver)
• (3) RTT (measured by sending RTT probe
  packets)

5
Survival analysis result
6
User satisfaction index
Based on the Cox model, USI is proposed to
evaluate Skype users satisfaction levels
USI2.15log(bit rate)-1.55log(jitter)-0.36RTT
On average, the degrees of user dissatisfaction
caused by the bit rate, jitter, and roundtrip
time are in the proportion of 46531.
7
The validation of the quantitative USI

• Basic idea validate USI by an independent set of
  metrics that quantify the interactivity and
  smoothness of a conversation
• Basic method check packet size to identify if
  the conversation is ON
• Responsiveness the proportion of OFF periods
  that coincide with ON periods of the other side,
  greater responsiveness indicates better
  satisfaction
• Response Delay longer response delay indicates
  poorer conversation quality
• Talk Burst Length longer bursts indicate poorer
  quality

8
The validation of the quantitative USI
9
My comment on this paper

• The USI model can also be used to quantify VoIP
  applications other than Skype
• In addition to evaluating the performance of
  QoS-sensitive applications, the USI can be
  implemented as part of applications to allow
  adaptation for optimal user satisfaction in real
  time
• The USI can be used give an optimal choice of the
  tradeoff between bit rate, jitter and RTT in
  congestion control

10

• Thank you!