A linear algorithm for reliable predictive network control

TL;DR

This paper presents a linear, predictive network control algorithm that enhances reliability and performance in network scheduling and routing, demonstrated through simulations comparing favorably to MaxWeight policy.

Contribution

It introduces a novel linear programming-based predictive control approach for reliable network scheduling and routing, expanding traditional models to predict transmission success.

Findings

Improved network performance over MaxWeight policy

Fast execution due to linear programming approach

Enhanced reliability in network control

Abstract

This paper introduces a novel control approach for network scheduling and routing that is predictive and reliable in its nature, yet builds upon a linear program, making it fast in execution. First, we describe the canonical system model and how we expand it to be able to predict the success of transmissions. Furthermore, we define a notion of reliability and then explain the algorithm. With extended simulations, we demonstrate the gains in performance over the well known MaxWeight policy.