A Comparative Study of Call-By-Call and Two-Call Traffic Distribution Models in Computer Networks

Abstract

Naldi's Markov chain model offers an analytical framework to understand user behaviors in competitive environments involving two operators, such as Internet Service Providers (ISPs). The model investigates the influence of blocking probability and initial user preferences on traffic distribution between operators. In its original form, the model assumes a "call-by-call" basis, where users switch operators after a failed call attempt, favoring operators with lower blocking probabilities.

This paper extends the model by introducing a "two-call" attempt framework, allowing users to make two consecutive attempts with the same operator before switching. Results show that this modification significantly enhances traffic share for operators, particularly those with higher blocking probabilities, as it reduces immediate traffic loss following a single failure.

A comparative analysis between the call-by-call and two-call models reveals that the latter improves traffic retention, increases connection success rates, and decreases churn. This study underscores the critical impact of blocking probabilities on user behaviors and demonstrates how the two-call model mitigates these challenges to optimize traffic distribution.

The findings highlight the value of user behaviors modelling for network optimization, providing actionable insights for ISPs and operators. By adopting the two-call approach, operators can improve quality of service (QoS), manage congestion effectively, and retain users more successfully in competitive markets.