CSGO: Generalized Optimization for Cold Start in Wireless Collaborative Edge LLM Systems

TL;DR

This paper introduces a latency-aware scheduling framework for edge-based large language models that minimizes cold-start latency by overlapping model loading with inference tasks, optimizing resource utilization.

Contribution

It presents a novel dynamic programming approach to optimize model partitioning and device assignment, effectively reducing cold-start latency in wireless collaborative edge LLM systems.

Findings

Significant reduction in cold-start latency compared to baseline methods

Effective overlap of model loading with computation and communication

Dynamic adjustment of layer partitioning improves resource utilization

Abstract

While deploying large language models on edge devices promises low-latency and privacy-preserving AI services, it is hindered by limited device resources. Although pipeline parallelism facilitates distributed inference, existing approaches often ignore the cold-start latency caused by on-demand model loading. In this paper, we propose a latency-aware scheduling framework that overlaps model loading with computation and communication to minimize total inference latency. Based on device and model parameters, the framework dynamically adjusts layer partitioning and allocation to effectively hide loading time, thereby eliminating as many idle periods as possible. We formulate the problem as a Mixed-Integer Non-Linear Program and design an efficient dynamic programming algorithm to optimize model partitioning and device assignment. Experimental results show that the proposed method significantly reduces cold-start latency compared to baseline strategies.