Scalable Performance Evaluation of Fully Connected Mesh Networks Using Mininet: A Deterministic Benchmark for TCP, UDP, and ICMP Traffic

Mesh networks have become essential in contemporary communication systems due to their decentralized architecture, resilience, and inherent redundancy. Despite significant research progress, the impact of topology scaling on transport-layer performance remains insufficiently isolated in controlled environments. This study introduces a reproducible emulation framework using Mininet to evaluate fully connected mesh networks of three sizes-۵, ۱۰, and ۱۵ nodes-under TCP, UDP, and ICMP traffic conditions. By applying identical bandwidth, delay, and queueing parameters across all topologies, the experiments isolate node-count effects from external influences such as wireless interference, routing variability, or hardware constraints. The results indicate that Round-Trip Time (RTT) remains stable at approximately ۱۲.۲ ms across all scales, while TCP consistently reaches near-maximum link utilization (۹.۵۵ Mbps). UDP traffic demonstrates predictable linear throughput scaling with minimal jitter and negligible packet loss. These findings confirm that, within deterministic fully connected topologies, performance is governed primarily by link-layer properties rather than network size. The framework provides a clean, upper-bound performance baseline that complements prior simulation-based, analytical, and hardware studies, and serves as a foundation for future research in adaptive routing, topology-aware optimization, and reinforcement-learning-based congestion control in large-scale mesh networks.

کلیدواژه ها:

Mesh networks ، Mininet ، network emulation ، TCP performance ، UDP jitter