Evolution of Real‑Time Communication (RTC) Technology: From MCU to Global Cloud RTC

Real‑time communication (RTC) has grown from a niche technology in the mid‑1990s to a complex ecosystem spanning network transmission, media processing, signaling, and device integration. The author divides this evolution into five generations:

1. MCU stage (1996‑2004) – Early IP‑multimedia standards H.323/RTP introduced the MCU concept, enabling multi‑party conferencing at the cost of high central processing and limited scalability.

2. P2P stage (2001‑2007) – Broadband (ADSL) and improved PC capabilities fostered peer‑to‑peer voice/video, with SIP/SDP replacing H.323, and new NAT‑traversal techniques (STUN/TURN/ICE) emerging.

3. Single‑SFU stage (2003‑2012) – Centralized SFU services reduced client bandwidth, supporting large‑scale gaming voice channels and early live‑streaming platforms.

4. Cloud‑SFU stage (2010‑2017) – Mobile internet and cloud computing shifted processing to the cloud, introducing BGP‑based edge connectivity, multi‑SFU full‑mesh, and side‑car services such as recording, AI moderation, and CDN integration.

5. Global cloud RTC stage (2016‑present) – Pandemic‑driven demand accelerated cloud‑based PaaS offerings, SD‑RTN architectures, edge‑container deployment, intelligent routing, and multi‑link backup, while AI‑enhanced media processing (beautification, super‑resolution, AI coding) became mainstream.

Throughout these phases, the article discusses protocol evolution (H.323, SIP, RTP, QUIC/RUDP), congestion control (GCC, BBR, PCC), error‑correction (FEC, ARQ/NACK), scalability techniques (Simulcast, SVC), and the emerging need for QoE metrics. It also reflects on why certain technologies (e.g., WebRTC) dominate and why others (e.g., H.264) persist.

The concluding insight is that each major RTC wave is driven not only by advances in terminal compute and bandwidth but also by shifting user‑side scenarios, making the field a blend of deep technical challenges and user‑experience focus.