Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Scalable Cyclic Schedulers for Age of Information Optimization in Large-Scale Status Update Systems

Nail Akar, Sahan Liyanaarachchi, Sennur Ulukus

TL;DR

This work addresses scalable cyclic scheduling for generate-at-will status update systems with heterogeneous service times and packet drops, aiming to minimize weighted AoI and PAoI. It develops an analytical method to compute per-source AoI/PAoI moments under a given pattern and introduces two scalable schedulers: SAMS for AoI and SPMS for PAoI, built on convex optimization and a deficit-round-robin spreading mechanism. A fixed-point iterative approach is proposed for AoI optimization, while PAoI optimization yields closed-form utilization and frequency expressions, enabling pattern construction with low complexity $O(NK)$. Numerical results demonstrate that SAMS closely matches the performance of the IS benchmark in small-scale problems with far lower complexity and that SPMS substantially improves PAoI while surpassing P-GAW$^*$ for AoI, even in large-scale settings with thousands of sources. The framework thus offers practical, scalable scheduling solutions for large-scale status update networks.

Abstract

We study cyclic scheduling for generate-at-will (GAW) multi-source status update systems with heterogeneous service times and packet drop probabilities, with the aim of minimizing the weighted sum age of information (AoI), called system AoI, or the weighted sum peak AoI (PAoI), called system PAoI. In particular, we obtain well-performing cyclic schedulers which can easily scale to thousands of information sources and which also have low online implementation complexity. The proposed schedulers are comparatively studied against existing scheduling algorithms in terms of computational load and system AoI/PAoI performance, to validate their effectiveness.

Scalable Cyclic Schedulers for Age of Information Optimization in Large-Scale Status Update Systems

TL;DR

This work addresses scalable cyclic scheduling for generate-at-will status update systems with heterogeneous service times and packet drops, aiming to minimize weighted AoI and PAoI. It develops an analytical method to compute per-source AoI/PAoI moments under a given pattern and introduces two scalable schedulers: SAMS for AoI and SPMS for PAoI, built on convex optimization and a deficit-round-robin spreading mechanism. A fixed-point iterative approach is proposed for AoI optimization, while PAoI optimization yields closed-form utilization and frequency expressions, enabling pattern construction with low complexity . Numerical results demonstrate that SAMS closely matches the performance of the IS benchmark in small-scale problems with far lower complexity and that SPMS substantially improves PAoI while surpassing P-GAW for AoI, even in large-scale settings with thousands of sources. The framework thus offers practical, scalable scheduling solutions for large-scale status update networks.

Abstract

We study cyclic scheduling for generate-at-will (GAW) multi-source status update systems with heterogeneous service times and packet drop probabilities, with the aim of minimizing the weighted sum age of information (AoI), called system AoI, or the weighted sum peak AoI (PAoI), called system PAoI. In particular, we obtain well-performing cyclic schedulers which can easily scale to thousands of information sources and which also have low online implementation complexity. The proposed schedulers are comparatively studied against existing scheduling algorithms in terms of computational load and system AoI/PAoI performance, to validate their effectiveness.
Paper Structure (8 sections, 23 equations, 5 figures, 2 algorithms)

This paper contains 8 sections, 23 equations, 5 figures, 2 algorithms.

Table of Contents

  1. Introduction
  2. System Model
  3. Analytical Method
  4. System PAoI and AoI Optimization
  5. System PAoI Optimization
  6. System AoI Optimization
  7. Numerical Examples
  8. Summary

Figures (5)

  • Figure 1: Information packets from $N$ information sources are collected by a server, e.g., base station (BS) of a cellular wireless network.
  • Figure 2: System AoI $\mathbb{E} [\Delta]$ as a function of $s_3$ obtained under various policies.
  • Figure 3: System AoI $\mathbb{E} [\Delta]$ as a function of $w_3$ obtained under various policies.
  • Figure 4: System AoI $\mathbb{E} [\Delta]$ for $N=100$ with deterministic and identical service times for 20 different experiments.
  • Figure 5: Variation of system PAoI $\mathbb{E} [\Phi]$ and system AoI $\mathbb{E} [\Delta]$ as a function of $w_3$ with $N=3$ and exponential service times.