arXiv subjects

Feiran You

Publications and source records attributed to Feiran You.

6 recordsLinked to original sources

arXiv2026

StarSD: One-for-Many Speculative Decoding

Speculative decoding accelerates autoregressive generation by separating token proposal from verification, but most existing approaches are designed for single-node execution and do not scale well to multi-accelerator clusters used for serving modern Large Language Models (LLMs). We present StarSD, a one-for-many speculative decoding framework that uses a single draft model to serve multiple target models across distributed nodes via a star topology. StarSD decouples drafting and verification, enabling effective sharing of draft computation, and preventing distributed accelerators from remaining idle under bursty workloads. We provide a system-level analysis that characterizes when and why a single draft model can remain fully utilized by multiple verifiers, yielding predictable latency and utilization gains. Extensive experiments in real-world distributed inference settings demonstrate that StarSD simplifies deployment and supports flexible resource allocation across heterogeneous accelerators, while maintaining output quality. These results indicate that StarSD is a practical and scalable framework for bringing speculative decoding to modern cloud and edge inference infrastructures.

eess.SY↗

arXiv2025

Ubiquitous Intelligence Via Wireless Network-Driven LLMs Evolution

We introduce ubiquitous intelligence as a paradigm where Large Language Models (LLMs) evolve within wireless network-driven ecosystems. Unlike static model deployments, this approach enables scalable and continuous intelligence ascension through coordination between networks and LLMs. Wireless networks support system-orchestrated lifelong learning, while LLMs drive the next-generation network development that is more adaptive and responsive. This co-evolution highlights a shift toward self-improving systems, sustaining capability growth across diverse and resource-constrained environments.

cs.NI↗

arXiv2025

ReaCritic: Reasoning Transformer-based DRL Critic-model Scaling For Wireless Networks

Heterogeneous Networks (HetNets) pose critical challenges for intelligent management due to the diverse user requirements and time-varying wireless conditions. These factors introduce significant decision complexity, which limits the adaptability of existing Deep Reinforcement Learning (DRL) methods. In many DRL algorithms, especially those involving value-based or actor-critic structures, the critic component plays a key role in guiding policy learning by estimating value functions. However, conventional critic models often use shallow architectures that map observations directly to scalar estimates, limiting their ability to handle multi-task complexity. In contrast, recent progress in inference-time scaling of Large Language Models (LLMs) has shown that generating intermediate reasoning steps can significantly improve decision quality. Motivated by this, we propose ReaCritic, a reasoning transformer-based critic-model scaling scheme that brings reasoning-like ability into DRL. ReaCritic performs horizontal reasoning over parallel state-action inputs and vertical reasoning through deep transformer stacks. It is compatible with a broad range of value-based and actor-critic DRL algorithms and enhances generalization in dynamic wireless environments. Extensive experiments demonstrate that ReaCritic improves convergence speed and final performance across various HetNet settings and standard OpenAI Gym control tasks. The code of ReaCritic is available at https://github.com/NICE-HKU/ReaCritic.

cs.LG↗

arXiv2025

DRESS: Diffusion Reasoning-based Reward Shaping Scheme For Intelligent Networks

Network optimization remains fundamental in wireless communications, with Artificial Intelligence (AI)-based solutions gaining widespread adoption. As Sixth-Generation (6G) communication networks pursue full-scenario coverage, optimization in complex extreme environments presents unprecedented challenges. The dynamic nature of these environments, combined with physical constraints, makes it difficult for AI solutions such as Deep Reinforcement Learning (DRL) to obtain effective reward feedback for the training process. However, many existing DRL-based network optimization studies overlook this challenge through idealized environment settings. Inspired by the powerful capabilities of Generative AI (GenAI), especially diffusion models, in capturing complex latent distributions, we introduce a novel Diffusion Reasoning-based Reward Shaping Scheme (DRESS) to achieve robust network optimization. By conditioning on observed environmental states and executed actions, DRESS leverages diffusion models' multi-step denoising process as a form of deep reasoning, progressively refining latent representations to generate meaningful auxiliary reward signals that capture patterns of network systems. Moreover, DRESS is designed for seamless integration with any DRL framework, allowing DRESS-aided DRL (DRESSed-DRL) to enable stable and efficient DRL training even under extreme network environments. Experimental results demonstrate that DRESSed-DRL achieves about 1.5x times faster convergence than its original version in sparse-reward wireless environments and significant performance improvements in multiple general DRL benchmark environments compared to baseline methods. The code of DRESS is available at https://github.com/NICE-HKU/DRESS.

cs.NI↗

arXiv2024

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

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%.

cs.NI↗

arXiv2024

JPPO: Joint Power and Prompt Optimization for Accelerated Large Language Model Services

Large Language Models (LLMs) have demonstrated remarkable capabilities in various tasks, leading to their increasing deployment in wireless networks for a wide variety of user services. However, the growing longer prompt setting highlights the crucial issue of computational resource demands and huge communication load. To address this challenge, we propose Joint Power and Prompt Optimization (JPPO), a framework that combines Small Language Model (SLM)-based prompt compression with wireless power allocation optimization. By deploying SLM at user devices for prompt compression and employing Deep Reinforcement Learning for joint optimization of compression ratio and transmission power, JPPO effectively balances service quality with resource efficiency. Experimental results demonstrate that our framework achieves high service fidelity and low bit error rates while optimizing power usage in wireless LLM services. The system reduces response time by about 17%, with the improvement varying based on the length of the original prompt.

eess.AS↗