JPPO++: Joint Power and Denoising-inspired Prompt Optimization for Mobile LLM Services

TL;DR

This paper introduces JPPO++, a framework that jointly optimizes prompt compression and wireless transmission power for mobile LLM services, significantly reducing service time and prompt length while maintaining output quality.

Contribution

It proposes a novel denoising-inspired prompt compression scheme combined with DRL-based joint optimization for mobile LLM systems, enhancing efficiency and response speed.

Findings

JPPO++ reduces service time by 46.5% compared to no compression.

Up to 16x prompt length reduction with acceptable accuracy loss.

JPPO with 16x compression cuts total service time by 42.3%.

Abstract

Large Language Models (LLMs) are increasingly integrated into mobile services over wireless networks to support complex user requests. This trend has led to longer prompts, which improve LLMs' performance but increase data transmission costs and require more processing time, thereby reducing overall system efficiency and negatively impacting user experience. To address these challenges, we propose Joint Prompt and Power Optimization (JPPO), a framework that jointly optimizes prompt compression and wireless transmission power for mobile LLM services. JPPO leverages a Small Language Model (SLM) deployed at edge devices to perform lightweight prompt compression, reducing communication load before transmission to the cloud-based LLM. A Deep Reinforcement Learning (DRL) agent dynamically adjusts both the compression ratio and transmission power based on network conditions and service constraints, aiming to minimize service time while preserving response fidelity. We further extend the framework to JPPO++, which introduces a denoising-inspired compression scheme. This design performs iterative prompt refinement by progressively removing less informative tokens, allowing for more aggressive yet controlled compression. Experimental results show that JPPO++ reduces service time by 17% compared to the no-compression baseline while maintaining output quality. Under compression-prioritized settings, a reduction of up to 16x in prompt length can be achieved with an acceptable loss in accuracy. Specifically, JPPO with a 16x ratio reduces total service time by approximately 42.3%, and JPPO++ further improves this reduction to 46.5%.