Portable Lightweight 4K Live Streaming Camera Headset for Surgeons

Abstract

Surgical education depends on direct observation of live procedures, but the operating roomcan only hold a handful of students at a time, leaving the rest to wait their turn. This thesispresents the design, implementation, and validation of a secure, low-latency system thatstreams a surgeon’s first-person view to remote browsers, built as a software demonstratoron a Rockchip RK3588 single-board computer with a 4K camera. The video path runs overWebRTC with hardware-accelerated H.264 encoding. To keep the security-critical codesurface small within a fixed timeline, the design favours configuration-driven, minimal-code components. A YAML-configured media server (MediaMTX) and stateless WHIP/WHEPsignalling replace a custom signalling server. Because WHEP is receive-only, two-way audiofor surgeon–student interaction is obtained through a dual-path design that opens aseparate WHIP session for the return channel. Transport security is provided through TLSand DTLS-SRTP, while access control is handled using credential-based authentication andviewer logging. The device operates autonomously through a Mealy finite state machinethat manages start-up, starting and closing the stream, error recovery, and shutdown. A 5GHz LAN the prototype delivered 4K resolution at 30 fps with a mean end-to-end latencyof 277 ms, confirmed two-way audio, and held the audio/video offset to 70.5 ms, which iswithin the ITU-R BT.1359 acceptability bound. Sixteen of nineteen verifiable requirementswere fully met; the principal shortfall is the absence of an external signaling server, whichconfines operation to a local network. The results establish the system architecture as aviable foundation for remote surgical observation, with external signaling, production-grade security, and multi-user scaling identified as the main remaining steps before clinicaldeployment.