Connection Reliability
William C. Hardy · 2001
Describes and analyzes users' concerns with their ability to effect connections through a telecommunications service. Defines the generic measure for perceived QoS to be the probability of experiencing and expected outcome, regardless of its cause. Contrasts the resultant measure, Normal Completion Rate (NCR) with more commonly used measures of connection success probability, such as the call completion rate and answer/seizure ratio. Recommends the probability distribution of runs of connection failures as the appropriate quantifier for purposes of evaluation, and describes evaluation criteria based on that quantifier. Supporting Appendix C analyzes the practical problems involved in trying to use answer/seizure ratio to compare connection reliabilities.