<![CDATA[Accurate traffic generation is essential for realistic network simulations in systems such as Content Delivery Networks (CDNs), Information-Centric Networks (ICNs), and the Internet of Things (IoT). These environments handle various types of data traffic—ranging from web pages and videos to sensor data and software updates—making it critical to model traffic patterns effectively. A well-designed traffic generator enables researchers and engineers to simulate real-world workloads, test scalability, and evaluate the performance of caching, routing, and resource allocation strategies under realistic conditions. Each traffic class has unique characteristics, including object size distributions, access patterns, and temporal dynamics. Capturing these differences is key to producing meaningful simulation results. For instance, CDNs require simulation of content popularity and delivery latency, ICNs focus on content retrieval and caching efficiency, while IoT simulations demand modeling of device behavior and intermittent communication. To support such complex scenarios, a traffic generator must not only mimic real user behavior but also allow for flexible scaling, combination, and modification of traffic patterns. This paper presents a method for evaluating synthetic traffic generators by comparing their output to the statistical properties of the Zipf distribution. The focus is on assessing whether synthetic traffic accurately reflects the heavy-tailed nature of real-world traffic as modeled by Zipf’s law. By analyzing the frequency distribution of requests generated by the traffic model and comparing it to theoretical Zipf curves, the study provides insights into the fidelity of the traffic generator. We measure the discrepancy between the simulated network traffic and the theoretical model to evaluate the accuracy and realism of the traffic generation approach.]]>
Copyrights © 2025
