DARTA: Generation of Autocorrelated Random Numbers using Discrete AutoRegression To Anything

Accurate understanding of stochastic processes is crucial for modeling modern communication systems, including machine-to-machine communication, which often exhibit autocorrelation. To effectively model and optimize systems like 5G and future 6G deployments, reliable tools are required to generate autocorrelated processes as inputs for discrete-event simulations or statistical models. The widely used AutoRegression To Anything (ARTA) model generates autocorrelated processes with arbitrary structures. We propose the Discrete AutoRegression To Anything (DARTA) model, an extension of ARTA that enhances performance and numerical computation using discrete random processes with appropriate time discretization. Through a comprehensive parameter study, we evaluate DARTA’s performance, assessing its distribution matching capability, configured autocorrelation, and runtime. Our results demonstrate the effectiveness and practicality of DARTA in efficiently generating discrete autocorrelated stochastic processes. A ready-to-use implementation of our proof-of-concept is provided.