The datasets are used to conduct a series of trace-driven evaluations for our proposed learning-based framework, Learning-based Automated Reconfiguration for vRANs (LARV). The detailed experiment setups and how the datasets are collected and utilized can be found in the article:
F. W. Murti, S. Ali, G. Iosifidis, M. Latva-aho, Deep Reinforcement Learning for Orchestrating Cost-aware Reconfigurations of vRANs, IEEE Transactions on Network & Service Management, 2023. (to appear).
Contents:
-
data. Measurement traces that show the relations between the traffic demands and computing resources in platform A and B. Platform A: data_milan. Platform B: data_milan2.
-
topology. We consider a realistic MEC-based Milan topology (N1) and a synthetic topology (N2) generated using the Waxman algorithm. N1: milan21.pickle or milan21.json. N2: waxman21.pickle or waxman21.json