Network infrastructure is evolving with LEO satellite constellations, next-generation cellular networks, and edge computing, yet transport protocols were designed assuming relatively stable network paths, homogeneous link characteristics, and predictable performance. These assumptions break down in modern environments where devices transition between cellular, WiFi, and satellite connectivity; where satellite motion creates dynamic topologies with time-varying latency; and where applications demand both ultra-low latency and high throughput. Traditional single-path protocols leave substantial capacity untapped when multiple network interfaces are available, and protocol parameters tuned for terrestrial networks perform poorly over satellite links with different bandwidth-delay products and loss characteristics. Understanding how protocols actually behave in these emerging environments through measurement, and designing protocol optimizations and extensions that account for their unique characteristics, remains essential research.
Our research develops protocol optimizations for emerging network environments informed by empirical measurements. We study multipath transport protocols that enable simultaneous use of multiple network interfaces—WiFi, cellular, and satellite—for bandwidth aggregation, seamless failover, and mobility support. Work on packet scheduling algorithms, per-path congestion control, and energy-aware multipath operation addresses the challenges of heterogeneous path characteristics. For LEO satellite networks, we explore protocol adaptations including connection establishment optimizations (QUIC's benefits for satellite RTTs), connection migration as satellites move, transport parameter tuning (window scaling, buffer sizing) for satellite link characteristics, and congestion control mechanisms that distinguish satellite-specific losses from network congestion. We investigate satellite-terrestrial hybrid architectures where devices intelligently select paths based on application requirements, and real-time communication protocols optimized for edge computing scenarios. Our protocol research combines measurement-driven understanding of actual network behavior with design of optimizations and extensions suited to emerging infrastructure. We participate in IETF standardization efforts, contributing measurement insights to protocol working groups and engaging with network operators to ensure research addresses real deployment challenges.
