<!doctype html><html lang="en" class="no-js"><head><meta charset="utf-8"> <!-- begin SEO --><title>FlowContrast: Few-Shot Multimedia Traffic Representation Learning in LEO Satellite Networks - Mahdi Rahimi</title><meta property="og:locale" content="en-US"><meta property="og:site_name" content="Mahdi Rahimi"><meta property="og:title" content="FlowContrast: Few-Shot Multimedia Traffic Representation Learning in LEO Satellite Networks"><link rel="canonical" href="https://mmaahhddii1995.github.io/publication/ICME.md"><meta property="og:url" content="https://mmaahhddii1995.github.io/publication/ICME.md"><meta property="og:description" name="description" content="Although the adoption of encryption protocols is essential for ensuring data privacy, it introduces significant challenges for traffic classification on the Internet, particularly in low Earth orbit (LEO) satellite networks such as Starlink, where fluctuating RTT, burst jitter, and reordering frequently distort flow patterns. Meanwhile, modern Internet traffic is dominated by multimedia applications (video streaming, social media, audio services), whose burst structures and temporal dynamics differ significantly from conventional terrestrial flows. To address encrypted traffic classification, prior deep learning approaches rely on large labeled datasets and are primarily tailored for terrestrial networks, limiting their generalizability under satellite-specific dynamics. To close this gap, we propose FlowContrast, a framework that combines self-supervised contrastive pre-training with few-shot learning, tailored for satellite multimedia traffic. FlowContrast applies domain-aware augmentations directly on raw flows, simulating satellite-induced perturbations before flow-to-image conversion, preserving temporal dependencies and packet-level semantics. Using massive unlabeled traffic, FlowContrast learns transferable representations adapted with minimal labeled samples. We also release SMA40, the first publicly available dataset of Starlink multimedia applications traffic. Experiments on multiple benchmarks show FlowContrast consistently outperforms state-of-the-art methods, achieving superior cross-domain generalization with reduced label dependency. The SMA40 dataset will be released to facilitate future research"><meta property="og:type" content="article"><meta property="article:published_time" content="2026-06-26T00:00:00+00:00"> <script type="application/ld+json"> { "@context" : "http://schema.org", "@type" : "Person", "name" : "Mahdi Rahimi", "url" : "https://mmaahhddii1995.github.io", "sameAs" : null } </script> <!-- end SEO --><link href="https://mmaahhddii1995.github.io/feed.xml" type="application/atom+xml" rel="alternate" title="Mahdi Rahimi Feed"> <!-- http://t.co/dKP3o1e --><meta name="HandheldFriendly" content="True"><meta name="MobileOptimized" content="320"><meta name="viewport" content="width=device-width, initial-scale=1.0"> <script> document.documentElement.className = document.documentElement.className.replace(/\bno-js\b/g, '') + ' js '; </script> <!-- For all browsers --><link rel="stylesheet" href="https://mmaahhddii1995.github.io/assets/css/main.css"> <!-- Custom CSS to force white background and black text --><link rel="stylesheet" href="https://mmaahhddii1995.github.io/assets/css/custom.css"> <!-- Force moon (dark mode) as default on page load --> <script> if (!localStorage.getItem('theme')) { document.documentElement.setAttribute('data-theme', 'light'); localStorage.setItem('theme', 'light'); } </script><meta http-equiv="cleartype" content="on"> <!-- start custom head snippets --> <!-- favicon from https://commons.wikimedia.org/wiki/File:OOjs_UI_icon_academic-progressive.svg --><link rel="apple-touch-icon" sizes="180x180" href="https://mmaahhddii1995.github.io/images/apple-touch-icon-180x180.png"/><link rel="icon" type="image/svg+xml" href="https://mmaahhddii1995.github.io/images/favicon.svg"/><link rel="icon" type="image/png" href="https://mmaahhddii1995.github.io/images/favicon-32x32.png" sizes="32x32"/><link rel="icon" type="image/png" href="https://mmaahhddii1995.github.io/images/favicon-192x192.png" sizes="192x192"/><link rel="manifest" href="https://mmaahhddii1995.github.io/images/manifest.json"/><link rel="icon" href="/images/favicon.ico"/><meta name="theme-color" content="#ffffff"/><link rel="stylesheet" href="https://mmaahhddii1995.github.io/assets/css/academicons.css"/> <!-- Support for MatJax --> <script src="https://cdnjs.cloudflare.com/polyfill/v3/polyfill.min.js?features=es6"></script> <script id="MathJax-script" async src="https://cdn.jsdelivr.net/npm/mathjax@3/es5/tex-mml-chtml.js"></script> <!-- end custom head snippets --></head><body> <!--[if lt IE 9]><div class="notice--danger align-center" style="margin: 0;">You are using an <strong>outdated</strong> browser. Please <a href="http://browsehappy.com/">upgrade your browser</a> to improve your experience.</div><![endif]--><div class="masthead"><div class="masthead__inner-wrap"><div class="masthead__menu"><nav id="site-nav" class="greedy-nav"> <button><div class="navicon"></div></button><ul class="visible-links"><li class="masthead__menu-item masthead__menu-item--lg persist"> <a href="https://mmaahhddii1995.github.io/">Mahdi Rahimi</a></li><li class="masthead__menu-item"> <a href="https://mmaahhddii1995.github.io/">Home</a></li><li class="masthead__menu-item"> <a href="https://mmaahhddii1995.github.io/about/">About</a></li><li class="masthead__menu-item"> <a href="https://mmaahhddii1995.github.io/cv/">CV</a></li><li class="masthead__menu-item"> <a href="https://mmaahhddii1995.github.io/publications/">Publications</a></li><li class="masthead__menu-item"> <a href="https://mmaahhddii1995.github.io/students/">Teaching & Supervision</a></li><li class="masthead__menu-item"> <a href="https://mmaahhddii1995.github.io/NEWS/">Recent News</a></li><!-- Theme toggle removed to force permanent white/black mode --></ul><ul class="hidden-links hidden"></ul></nav></div></div></div><div id="main" role="main"><div class="sidebar sticky"><div itemscope itemtype="http://schema.org/Person"><div class="author__avatar"> <img src="https://mmaahhddii1995.github.io/images/profile.png" class="author__avatar" alt=""></div><div class="author__content"><h3 class="author__name"></h3><p class="author__bio">Academic researcher in anonymous communications, privacy-preserving systems, and applied cryptography.</p></div><div class="author__urls-wrapper"> <button class="btn btn--inverse">Follow</button><ul class="author__urls social-icons"> <!-- Font Awesome icons / Biographic information --><li class="author__desktop"><i class="fa-solid fa-location-dot icon-pad-right" aria-hidden="true"></i>Leuven, Belgium</li><li class="author__desktop"><i class="fas fa-fw fa-building-columns icon-pad-right" aria-hidden="true"></i>COSIC, KU Leuven</li><li><a href="mailto:mahdi.rahimi@esat.kuleuven.be"><i class="fas fa-fw fa-envelope icon-pad-right" aria-hidden="true"></i>Email</a></li><!-- Font Awesome and Academicons icons / Academic websites --><li><a href="https://scholar.google.com/citations?user=cwhrpkQAAAAJ&hl=en"><i class="ai ai-google-scholar icon-pad-right"></i>Google Scholar</a></li><li><a href="https://orcid.org/0009-0003-0223-9082"><i class="ai ai-orcid ai-fw icon-pad-right"></i>ORCID</a></li><!-- Font Awesome icons / Repositories and software development --><li><a href="https://github.com/Mmaahhddii1995"><i class="fab fa-fw fa-github icon-pad-right" aria-hidden="true"></i>GitHub</a></li><!-- Font Awesome icons / Social media --><li><a href="https://www.linkedin.com/in/mahdirahimi95"><i class="fab fa-fw fa-linkedin icon-pad-right" aria-hidden="true"></i>LinkedIn</a></li></ul></div></div></div><article class="page" itemscope itemtype="http://schema.org/CreativeWork"><meta itemprop="headline" content="FlowContrast: Few-Shot Multimedia Traffic Representation Learning in LEO Satellite Networks"><meta itemprop="description" content="Although the adoption of encryption protocols is essential for ensuring data privacy, it introduces significant challenges for traffic classification on the Internet, particularly in low Earth orbit (LEO) satellite networks such as Starlink, where fluctuating RTT, burst jitter, and reordering frequently distort flow patterns. Meanwhile, modern Internet traffic is dominated by multimedia applications (video streaming, social media, audio services), whose burst structures and temporal dynamics differ significantly from conventional terrestrial flows. To address encrypted traffic classification, prior deep learning approaches rely on large labeled datasets and are primarily tailored for terrestrial networks, limiting their generalizability under satellite-specific dynamics. To close this gap, we propose FlowContrast, a framework that combines self-supervised contrastive pre-training with few-shot learning, tailored for satellite multimedia traffic. FlowContrast applies domain-aware augmentations directly on raw flows, simulating satellite-induced perturbations before flow-to-image conversion, preserving temporal dependencies and packet-level semantics. Using massive unlabeled traffic, FlowContrast learns transferable representations adapted with minimal labeled samples. We also release SMA40, the first publicly available dataset of Starlink multimedia applications traffic. Experiments on multiple benchmarks show FlowContrast consistently outperforms state-of-the-art methods, achieving superior cross-domain generalization with reduced label dependency. The SMA40 dataset will be released to facilitate future research"><meta itemprop="datePublished" content="June 26, 2026"><div class="page__inner-wrap"><header><h1 class="page__title" itemprop="headline">FlowContrast: Few-Shot Multimedia Traffic Representation Learning in LEO Satellite Networks</h1><p>Published in <i>IEEE International Conference on Multimedia and Expo (ICME)</i>, 2026</p></header><section class="page__content" itemprop="text"><p>The contents above will be part of a list of publications, if the user clicks the link for the publication than the contents of section will be rendered as a full page, allowing you to provide more information about the paper for the reader. When publications are displayed as a single page, the contents of the above “citation” field will automatically be included below this section in a smaller font.</p><p style="font-size: smaller">Recommended citation: H. Zhang, G. Xiong, G. Gou, Z. Li, W. Xia, Z. Chen, H. Li, and M.Rahimi, "FlowContrast: Few-Shot Multimedia Traffic Representation Learning in LEO Satellite Networks," in ICME 2026: IEEE International Conference on Multimedia and Expo.</p></section><footer class="page__meta"></footer><section class="page__share"><h4 class="page__share-title">Share on</h4><a href="https://bsky.app/intent/compose?text=https://mmaahhddii1995.github.io/publication/ICME.md" class="btn btn--bluesky" title="Share on Bluesky"><i class="fab fa-bluesky" aria-hidden="true"></i><span> Bluesky</span></a> <a href="https://www.facebook.com/sharer/sharer.php?u=https://mmaahhddii1995.github.io/publication/ICME.md" class="btn btn--facebook" title="Share on Facebook"><i class="fab fa-facebook" aria-hidden="true"></i><span> Facebook</span></a> <a href="https://www.linkedin.com/shareArticle?mini=true&url=https://mmaahhddii1995.github.io/publication/ICME.md" class="btn btn--linkedin" title="Share on LinkedIn"><i class="fab fa-linkedin" aria-hidden="true"></i><span> LinkedIn</span></a> <a href="https://x.com/intent/post?text=https://mmaahhddii1995.github.io/publication/ICME.md" class="btn btn--x" title="Share on X"><i class="fab fa-x-twitter" aria-hidden="true"></i><span> X (formerly Twitter)</span></a></section><nav class="pagination"> <a href="https://mmaahhddii1995.github.io/publication/WhenMixnetsFail.md" class="pagination--pager" title="When Mixnets Fail: Evaluating, Quantifying, and Mitigating the Impact of Adversarial Nodes in Mix Networks ">Previous</a> <a href="#" class="pagination--pager disabled">Next</a></nav></div></article></div><div class="page__footer"><footer> <!-- Sitemap & follow removed --> <!-- Footer removed --></footer></div><script src="https://mmaahhddii1995.github.io/assets/js/main.min.js"></script></body></html>
