Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Distributed Offloading in Multi-Access Edge Computing Systems: A Mean-Field Perspective

Shubham Aggarwal, Muhammad Aneeq uz Zaman, Melih Bastopcu, Sennur Ulukus, Tamer Başar

TL;DR

This paper studies optimal task offloading in MEC systems from a mean-field game (MFG) perspective to address timeliness (AoI) and scalability. For equitable access, each of $N$ devices decides local processing versus edge offloading via a scalarized AoI–power objective, with AoI computed through stochastic hybrid system (SHS) modeling and large-$\lambda$ approximations to enable tractable mean-field policies. For priority-based access, a major-minor MFG (MM-MFG) framework is used to model a primary user with ES priority and secondary users sharing the ES, including a pricing mechanism and LCFS-PP discipline; the major player interacts with a continuum of minor players through the mean field. The authors provide distributed algorithms to compute mean-field equilibria (MFE) and major-minor equilibria (MM-MFE), show that MF-based policies closely approximate Nash equilibria from centralized solutions, and demonstrate significant improvements in ES utilization and AoI-power trade-offs across both scenarios. The results offer a clean, scalable design methodology for AoI-aware MEC systems and suggest directions such as graphon-MFGs for networked interactions.

Abstract

Multi-access edge computing (MEC) technology is a promising solution to assist power-constrained IoT devices by providing additional computing resources for time-sensitive tasks. In this paper, we consider the problem of optimal task offloading in MEC systems with due consideration of the timeliness and scalability issues under two scenarios of equitable and priority access to the edge server (ES). In the first scenario, we consider a MEC system consisting of $N$ devices assisted by one ES, where the devices can split task execution between a local processor and the ES, with equitable access to the ES. In the second scenario, we consider a MEC system consisting of one primary user, $N$ secondary users and one ES. The primary user has priority access to the ES while the secondary users have equitable access to the ES amongst themselves. In both scenarios, due to the power consumption associated with utilizing the local resource and task offloading, the devices must optimize their actions. Additionally, since the ES is a shared resource, other users' offloading activity serves to increase latency incurred by each user. We thus model both scenarios using a non-cooperative game framework. However, the presence of a large number of users makes it nearly impossible to compute the equilibrium offloading policies for each user, which would require a significant information exchange overhead between users. Thus, to alleviate such scalability issues, we invoke the paradigm of mean-field games to compute approximate Nash equilibrium policies for each user using their local information, and further study the trade-offs between increasing information freshness and reducing power consumption for each user. Using numerical evaluations, we show that our approach can recover the offloading trends displayed under centralized solutions, and provide additional insights into the results obtained.

Distributed Offloading in Multi-Access Edge Computing Systems: A Mean-Field Perspective

TL;DR

This paper studies optimal task offloading in MEC systems from a mean-field game (MFG) perspective to address timeliness (AoI) and scalability. For equitable access, each of devices decides local processing versus edge offloading via a scalarized AoI–power objective, with AoI computed through stochastic hybrid system (SHS) modeling and large- approximations to enable tractable mean-field policies. For priority-based access, a major-minor MFG (MM-MFG) framework is used to model a primary user with ES priority and secondary users sharing the ES, including a pricing mechanism and LCFS-PP discipline; the major player interacts with a continuum of minor players through the mean field. The authors provide distributed algorithms to compute mean-field equilibria (MFE) and major-minor equilibria (MM-MFE), show that MF-based policies closely approximate Nash equilibria from centralized solutions, and demonstrate significant improvements in ES utilization and AoI-power trade-offs across both scenarios. The results offer a clean, scalable design methodology for AoI-aware MEC systems and suggest directions such as graphon-MFGs for networked interactions.

Abstract

Multi-access edge computing (MEC) technology is a promising solution to assist power-constrained IoT devices by providing additional computing resources for time-sensitive tasks. In this paper, we consider the problem of optimal task offloading in MEC systems with due consideration of the timeliness and scalability issues under two scenarios of equitable and priority access to the edge server (ES). In the first scenario, we consider a MEC system consisting of devices assisted by one ES, where the devices can split task execution between a local processor and the ES, with equitable access to the ES. In the second scenario, we consider a MEC system consisting of one primary user, secondary users and one ES. The primary user has priority access to the ES while the secondary users have equitable access to the ES amongst themselves. In both scenarios, due to the power consumption associated with utilizing the local resource and task offloading, the devices must optimize their actions. Additionally, since the ES is a shared resource, other users' offloading activity serves to increase latency incurred by each user. We thus model both scenarios using a non-cooperative game framework. However, the presence of a large number of users makes it nearly impossible to compute the equilibrium offloading policies for each user, which would require a significant information exchange overhead between users. Thus, to alleviate such scalability issues, we invoke the paradigm of mean-field games to compute approximate Nash equilibrium policies for each user using their local information, and further study the trade-offs between increasing information freshness and reducing power consumption for each user. Using numerical evaluations, we show that our approach can recover the offloading trends displayed under centralized solutions, and provide additional insights into the results obtained.

Paper Structure

This paper contains 17 sections, 4 theorems, 24 equations, 14 figures, 7 tables, 3 algorithms.

Table of Contents

  1. Introduction
  2. Related Works
  3. MEC with Equitable Access--Problem Setup
  4. Age of Information (AoI) Calculations
  5. Stochastic Hybrid Systems (SHS)
  6. Average AoI Calculation
  7. Mean-Field Game (MFG)
  8. Generic Device Optimization Problem
  9. MEC With Priority-Based Access
  10. Average AoI Computation for MEC with Priority-Based Access
  11. Average AoI Analysis for the Primary User
  12. Average AoI Analysis for the Secondary User
  13. Major-Minor Mean-Field Analysis for the MEC System With Priority Access
  14. Performance Evaluation
  15. The MEC System With Equitable Access
  16. ...and 2 more sections

Key Result

Theorem 1

yates2018age Suppose that $\Bar{\pi}$ is the state distribution of the FS-MC and there exists a stationary solution $\Bar{v}\!:=\! [\Bar{v}_1,\! \cdots,\! \Bar{v}_m]$ of the process $v_{\cdot}(t)$ satisfying Then, the average AoI is given by $\Delta:= \sum_{s \in \operatorname{S}}\bar{v}_{s0}$.

Figures (14)

  • Figure 1: The figure shows a prototypical MEC system consisting of an edge server and intelligent applications such as connected autonomy, medical internet-of-things and surveillance that simultaneously utilize edge server for timely computation.
  • Figure 2: The flow of incoming tasks is shown for a system of $N$ devices. In particular, for device $D_i$: $L_i$ and $T_i$ denote the device's local processor and its transmitter, resp.; ES denotes the edge server; $p_i$ and $\bar{p}_i \!=\! 1-p_i$ denote the Bernoulli probability according to which a task is either chosen to be served locally or offloaded to the ES.
  • Figure 3: Evolution of AoI at the receiver
  • Figure 4: Task flow from the perspective of device $D_i$.
  • Figure 5: Task flow for a two-user MEC system with one ES from the perspective of device $D_1$.
  • ...and 9 more figures

Theorems & Definitions (6)

  • Theorem 1
  • Remark 1
  • Theorem 2
  • Remark 2
  • Theorem 3
  • Theorem 4