Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Decentralized AI Service Placement, Selection and Routing in Mobile Networks

Jinkun Zhang, Stefan Vlaski, Kin Leung

TL;DR

In the proposed framework, traffic tunneling is used to support user mobility without costly AI service migrations, and a decentralized Frank--Wolfe algorithm with a novel messaging protocol is developed.

Abstract

The rapid development and usage of large-scale AI models by mobile users will dominate the traffic load in future communication networks. The advent of AI technology also facilitates a decentralized AI ecosystem where small organizations or even individuals can host AI services. In such scenarios, AI service (models) placement, selection, and request routing decisions are tightly coupled, posing a challenging yet fundamental trade-off between service quality and service latency, especially when considering user mobility. Existing solutions for related problems in mobile edge computing (MEC) and data-intensive networks fall short due to restrictive assumptions about network structure or user mobility. To bridge this gap, we propose a decentralized framework that jointly optimizes AI service placement, selection, and request routing. In the proposed framework, we use traffic tunneling to support user mobility without costly AI service migrations. To account for nonlinear queuing delays, we formulate a nonconvex problem to optimize the trade-off between service quality and end-to-end latency. We derive the node-level KKT conditions and develop a decentralized Frank--Wolfe algorithm with a novel messaging protocol. Numerical evaluations validate the proposed approach and show substantial performance improvements over existing methods.

Decentralized AI Service Placement, Selection and Routing in Mobile Networks

TL;DR

In the proposed framework, traffic tunneling is used to support user mobility without costly AI service migrations, and a decentralized Frank--Wolfe algorithm with a novel messaging protocol is developed.

Abstract

The rapid development and usage of large-scale AI models by mobile users will dominate the traffic load in future communication networks. The advent of AI technology also facilitates a decentralized AI ecosystem where small organizations or even individuals can host AI services. In such scenarios, AI service (models) placement, selection, and request routing decisions are tightly coupled, posing a challenging yet fundamental trade-off between service quality and service latency, especially when considering user mobility. Existing solutions for related problems in mobile edge computing (MEC) and data-intensive networks fall short due to restrictive assumptions about network structure or user mobility. To bridge this gap, we propose a decentralized framework that jointly optimizes AI service placement, selection, and request routing. In the proposed framework, we use traffic tunneling to support user mobility without costly AI service migrations. To account for nonlinear queuing delays, we formulate a nonconvex problem to optimize the trade-off between service quality and end-to-end latency. We derive the node-level KKT conditions and develop a decentralized Frank--Wolfe algorithm with a novel messaging protocol. Numerical evaluations validate the proposed approach and show substantial performance improvements over existing methods.

Paper Structure

This paper contains 7 sections, 7 theorems, 41 equations, 9 figures, 1 table, 1 algorithm.

Table of Contents

  1. Introduction
  2. Fixed AI Service Placement
  3. AI-driven network with mobile users
  4. Time-homogeneous formulation
  5. Decentralized Algorithm Design
  6. Optimized AI Service Placement
  7. Numerical Evaluation

Key Result

Proposition 1

With fixed $\{r_i^k\}$, obj_time_invar is equivalent to obj_origin. Specifically, it holds $J = -\left(\sum_{i}\sum_kr_i^k\right)Q$ for any $(\boldsymbol{s},\boldsymbol{\phi})$.

Figures (9)

  • Figure 1: An example edge-cloud vehicular network. Mobile users have multiple pre-train AI model options.
  • Figure 1: Scenarios
  • Figure 2: Traffic tunneling and impact on request latency.
  • Figure 3: Illustration of $\texttt{MSG1}$ and $\texttt{MSG2}$ propagation in DMP.
  • Figure 4: Normalized objective $J$ in all scenarios
  • ...and 4 more figures

Theorems & Definitions (7)

  • Proposition 1
  • Theorem 1
  • Proposition 2
  • Theorem 2
  • Theorem 3
  • Theorem 4
  • Theorem 5: KKT with service placement