SPAIK: мікросервісна платформа стримінгу реального часу з власними медіасервісами

Abstract

This paper presents the concept of SPAIK (Streaming Platform for Adaptive Interactive Communications), a low-latency streaming platform evolved from a course project that originally relied on third-party media components. The diploma-stage goal is to replace external streaming infrastructure with in-house microservices responsible for signaling, real-time media routing, recording, and adaptive delivery. We propose a hybrid media pipeline: WebRTC for interactive «glass-to-glass» communication and lowlatency HTTP-based streaming (LL-HLS / MPEG-DASH) for scalable audience distribution. A key research focus is minimizing end-to-end latency under volatile networks while preserving perceptual quality. We analyze ABR approaches (buffer-based BOLA vs. learning-based policies) and outline an MLassisted bandwidth prediction module to reduce rebuffering and quality oscillations. The expected outcome is an experimentally validated architecture and a reproducible evaluation framework for latency and QoE trade-offs.