Resource: Paper Carnival 2025

Carnival 2025

Time: 2025-08-27 ~ 2025-08-28

Location: A518

Day 1

Session 1: Networked System

1. LLM Cache Optimization - Qinyong Li "9:00-9:40"

• [Eurosys'25] CacheBlend: Fast Large Language Model Serving for RAG with Cached Knowledge Fusion
• [SigComm'24] CacheGen: KV Cache Compression and Streaming for Fast Large Language Model Serving

2. Disaggregated OS - Haifeng "9:40-10:00"

• [NSDI'25] Beehive: A Scalable Disaggregated Memory Runtime Exploiting Asynchrony of Multithreaded Programs

---

Session 2: Video Analytics for AIoT

1. ML for VA - Xinyan "10:05-10:45"

• [ISCA'24] DACAPO: Accelerating Continuous Learning in Autonomous Systems for Video Analytics
• [NSDI'23] RECL: Responsive Resource-Efficient Continuous Learning for Video Analytics
• [MM'23] Edge-Assisted On-Device Model Update for Video Analytics in Adverse Environments

2. Image Offloading Revisit - Yi Zhou "10:45-11:05"

• [MobiSys 2024] LimitNet: Progressive, Content-Aware Image Offloading for Extremely Weak Devices

3. Vision Drones - Jiahao "11:05-11:25"

• [T-RO'25] FAPP: Fast and Adaptive Perception and Planning for UAVs in Dynamic Cluttered Environments

---

Session 3: Networking

1. Quantum Networks - Yaliang "14:00-14:40"

• [INFOCOM'22] E2E Fidelity Aware Routing and Purification for Throughput Maximization in Quantum Networks
• [JSAC'24] On Optimum Entanglement Purification Scheduling in Quantum Networks
• [INFOCOM'25] Link Configuration for Fidelity-Constrained Entanglement Routing in Quantum Networks

2. V2V Networks - Zhenguo Bi "14:40-15:00"

• [INFOCOM'25] RoCooper: Robust Cooperative Perception under Vehicle-to-Vehicle Communication Impairments

---

Session 4: LoRa

1. ML in LoRa Reception - Kai Chen"15:05-15:45"

• [ICNP'23] Hi²LoRa: Exploring Highly Dimensional and Highly Accurate Features to Push LoRaWAN Concurrency Limits with Low Implementation Cost
• [SenSys'24] Enhancing LoRa Reception with Generative Models: Channel-Aware Denoising of LoRaPHY Signals
• [ICNP'24] Deepdetangle: Deep Learning-Based Fusion of Chirp-Level and Packet-Level Features for Lora Parallel Decoding

2. Performance - Mengyu "15:45-16:15"

• [ToN'24] RALoRa: Rateless-Enabled Link Adaptation for LoRa Networking
• [TMC'25] Enhancing Link Performance for Mobile LoRa Networks

---

Session 5: LLM Code Generation

1. Use of CodeGen - Youwei "16:20-16:40"

• [Mobicom'25] AutoDroid-V2: Boosting SLM-based GUI Agents via Code Generation

2. Code Translation - Bairong "16:40-17:00"

• [ICSE'25] InterTrans: Leveraging Transitive Intermediate Translations to Enhance LLM-based Code Translation
• [Arxiv] AutoIOT: LLM-Driven Automated Natural Language Programming for AIoT Applications

---

Day 2

Session 6: Volumetric video

1.Volume Video- Mengfan "9:00-9:20"

• [Mobicom'24] An End-to-End, Low-Cost, and High-Fidelity 3D Video Pipeline for Mobile Devices

2.Volume Video- Jiyi "9:20-10:00"

• [TOG'23] 3D Gaussian Splatting for Real-Time Radiance Field Rendering
• [TOG'24] V³: Viewing Volumetric Videos on Mobiles via Streamable 2D Dynamic Gaussians

---

Session 7: Edge Deployment and Inference

1. Edge LLM - Junzhe "10:05-10:25"

• [INFOCOM'25] TensAllo: Adaptive Deployment of LLMs on Resource-Constrained Heterogeneous Edge Devices

2. Inference Optimization - Ruizheng "10:25-10:45"

• [INFOCOM'25] DUNE: Distributed Inference in the User Plane

---