Modeling Network Traffic and Exploring Distribution Fitting: A Case Study on Spotify

As Internet traffic grows exponentially due to user demands and economic competition, the characterization of network workload and its resource utilization becomes more complex but also more important. This study aims to gain insights into the network's dynamics and explore various aspects of its behavior and performance. It presents an empirical analysis of network traffic based on a 24-hour dataset obtained at a university campus from internet users on the ‘Spotify’ platform. The analysis is built on data features, and it reveals several traffic models. More specifically, using the Cullen and Frey graphs, it investigates which are the optimal distributions fitting to data traffic. It focuses on patterns observed in the number of connections per minute and brings out an initial burst of activity followed by a stable level of connectivity. The study also examines the duration of TCP (Transmission Control Protocol) connections and the size of transfer data, indicating that most connections are short-lived and data transfers predominantly in-volve smaller sizes. These findings contribute to a comprehensive understanding of the network's behavior and can shape network management strategies for optimizing performance and resource allocation.