跳转至

LEO-Related Papers in SIGCOMM 2025

Tip

updated on 2025-0911

google上都可以直接搜到, 因此不挨个放链接了 :)

SIGCOMM 2025

(1) Enabling Over-the-Air AI for Edge Computing via Metasurface-Driven Physical Neural Networks

有点相关,想看

(2) LeoCC: Making Internet Congestion Control Robust to LEO Satellite Dynamics

必看

(3) Mobile Underwater Backscatter Networking

有点新意,想看

(4) Direct-to-Cell Satellite Network without Satellite Navigation

必看

(5) DeepSpace: Super Resolution Powered Efficient and Reliable Satellite Image Data Acquistion

必看

(6) SaTE: Low-Latency Traffic Engineering for Satellite Networks

必看

(7) Small-scale LEO Satellite Networking for Global-scale Demands

必看

(8) StarCDN: Moving Content Delivery Networks to Space

必看

SIGCOMM 2025 - Workshop: LEONET

Session 1: LEO Internet for All

(1) Optimising LEO Gateway Placement for the People

(2) Simulation and Comparison of Vehicle Satellite Connectivity under a 3D Foliage Environment: Starlink and OneWeb

Session 2: Improving Internet-from-Space

(1) The Internet from Space, Reimagined: Leveraging Altitude for Efficient Global Coverage

(2) A Distributed Data Store in Orbit

Session 3: Performance Characterization & Measurements

(1) FjordLink – Comparison of Starlink and 5G Networks for Tele-operated Vessel Control

(2) Better Fill Up Your Pipe – Revisiting Starlink's Burst Characterization

(3) A Detailed Characterization of Starlink One-way Delay

Session 4: Experimentation & Monitoring Platforms

(1) How LLM Saved Me from Struggling with Experiment Reproduction: LEO Networking as A Case Study

(2) MOSAIC: Piecing Together 5G and LEOs for NTN Integration Experimentation

Session 5: Resilience of LEO Internet

(1) Towards Global Outage Detection for LEO Networks

(2) An investigation of Starlink's performance during the May '24 solar superstorm

(3) A Deep Dive into the Impact of Solar Storms on LEO Satellite Networks

Summary

SIGCOMM25:

论文 核心 推荐系数 亮点与评价
LeoCC 基于 reconfig 信息进行CC 5 RI rather than RTT 的设计太巧妙了!
SN2 手机直连卫星, 不要MEO提供的精确地理信息, 直接LEO进行 导航+通信 5 利用"容许范围"来relax地理/信号的设计很巧妙, 回归本质
DeepSpace 高性能压缩图像与还原, SR + MoE 2 超分辨率SR + MoE分区根据特征还原, 想法很简单粗暴, 实验感觉有点黑魔法
StarCDN 映射关系高速变化下的CDN: 一致性哈希-tag + RelayFetching-逆向流动 3 附赠了一个traffic generator, 积累一下; 想法依旧简单粗暴, 实验也是继续黑魔法路线
TinyLEO 基于SDN的思想, 重构LEO星座 5 非常好的文章, SDN for LEO, 后续可以在dSDN中找找灵感?
SaTE GNN-based TE不再需要DNN就可以正常工作, 同时给出了2个剪枝 4 核心思想 "传统的优化问题 可以转换成 推理过程"; 缺点: 代码怎么还不开源?
MetaAI 数据传输和神经网络处理实现同步 TBD 真物理, 关联不大, 没细看
UWNet 构建水下网络 TBD 真物理, 关联不大, 没细看

SIGCOMM-LEONET25:

论文 核心 推荐系数 亮点与评价
3Denv 考虑植被遮挡的LEO网络仿真器 1 记一下, 积累一些: OMNeT++ + sgp4
EGC MEO优于LEO的可能性分析 2 可能是一个方向? 不太看好, 但不排除可能性
FilledPipe 用UDP测量throughput 2 积累一些: LeoScope + SpeedTest CLI
FjordLink 在潮汐/海浪影响下的StarLink测量 2 积累这个数据集, 未来或许需要
GWPlacement PoP地理分布与地址选择 3 没给出具体做法, 相当于survey. 积累一些: World Population + Py GeoPandas + Starlink Gateway/PoP + PeeringDB - IXPs + StarLink Ground Station Map
InOrbitStore 通过持续流动的方式来“存储”数据 4 "轨道距离的建模"可借鉴, "小量持续流动以实现存储"的idea很有新意
Laser LLM帮助LSN实验复现 4 很有新意, idea来源于当前LLM实验复现的其他应用; 看了一下仓库, 感觉代码质量有待提升...; 提供了不少仿真器仓库, 可积累
May24 太阳风暴对LEO网络性能的影响 1 应该是扩展自 IMC'24 CosmicDance, 不太感兴趣; 积累一下: RIPE Atlas - 全球网络数据实时测量
MOSAIC 5G TN/NTN集成仿真器 5 积累: 汇总好的LEO仿真器(Xeoverse / OpenSN / celestial / ns-3-leo / starlink_unity) + 地面移动网络模拟器(Open5GCore / OpenAirInterface)
One-way Delay 细致测算单向delay 2 积累一些: nanoping + nanoprobe
Roman-HitchHiking 全球LEO网络中断测量系统 3 积累一些: LEO 测试系统(LEOScope Testbed / Google M-Lab) + 工业界网络测量/故障检测(Cisco - ThousandEyes / Cloudflare Radar / DownDetector / IODA)
Solar Storms 太阳风暴对LEO的影响 1 跟上面may24差不多, 不感兴趣...