docs(config): add mediasoup SFU env vars and simulator behavior notes

Backend/README.md

@@ -34,9 +34,12 @@ Required env vars:
 | `BETTER_AUTH_BASE_URL` | Public base URL for the backend (e.g., `http://localhost:3000`) |
 | `BETTER_AUTH_TRUSTED_ORIGINS` | Comma-separated list of allowed frontend origins |
 | `PORT` | HTTP port (default `3000`) |
-| `MEDIA_MODE` | Media runtime mode (`legacy` default, `single_server_sfu` scaffold mode) |
+| `MEDIA_MODE` | Media runtime mode (`legacy` default, `single_server_sfu` for in-process SFU path) |
+| `MEDIA_SFU_ENGINE` | SFU engine for `single_server_sfu` mode (`mediasoup` default, `noop` fallback) |
 | `MEDIA_PROVIDER` | Media backend provider (`mock` by default) |
 | `TURN_URLS` / `TURN_USERNAME` / `TURN_CREDENTIAL` | TURN/STUN configuration used by single-server SFU mode |
+| `MEDIA_WEBRTC_LISTEN_IP` / `MEDIA_WEBRTC_ANNOUNCED_IP` | WebRTC transport bind/announce IPs for mediasoup SFU |
+| `MEDIA_RTC_MIN_PORT` / `MEDIA_RTC_MAX_PORT` | UDP/TCP RTP port range for mediasoup worker |
 | `MEDIA_RECORDINGS_DIR` | Local output directory for server-side recording workers (planned in SFU mode) |
 | `MEDIA_MAX_PUBLISHERS` / `MEDIA_MAX_SUBSCRIBERS_PER_ROOM` | Soft concurrency limits for single-server media mode (planned) |
 | `MINIO_*` | Connection settings for the MinIO/S3 endpoint |
@@ -145,15 +148,16 @@ Stream realtime events:
 - Client receives `stream:started` when camera accepts.
 - Both devices receive `stream:ended` when session is closed.
 
-Experimental SFU scaffolding endpoints (`MEDIA_MODE=single_server_sfu`):
+Experimental SFU endpoints (`MEDIA_MODE=single_server_sfu`):
 - `GET /streams/:streamSessionId/sfu/session` – fetch in-memory SFU session state for participant devices
+- `GET /streams/:streamSessionId/sfu/router-rtp-capabilities` – fetch router RTP capabilities used by mediasoup-client `Device.load()`
 - `POST /streams/:streamSessionId/sfu/publish-transport` – camera creates publish transport descriptor
 - `POST /streams/:streamSessionId/sfu/subscribe-transport` – participant creates subscribe transport descriptor
 - `POST /streams/:streamSessionId/sfu/publish-transport/connect` – camera marks publish transport as connected
 - `POST /streams/:streamSessionId/sfu/subscribe-transport/connect` – participant marks subscribe transport as connected
 - `POST /streams/:streamSessionId/sfu/produce` – camera registers media producer on connected publish transport
 - `POST /streams/:streamSessionId/sfu/consume` – participant creates consumer from available producer
-- `public/mobile-sim.js` now prefers this SFU handshake path automatically when `MEDIA_MODE=single_server_sfu`; legacy mode keeps direct WebRTC + frame relay fallback behavior.
+- `public/mobile-sim.js` now uses mediasoup-client handshakes in SFU mode and attaches consumed tracks to `#clientStreamVideo`; legacy mode keeps direct WebRTC + frame relay fallback behavior.
 
 #### Streaming Scale Tradeoffs (Current Prototype)
 - The current implementation is **not production-grade at scale**.
@@ -162,7 +166,7 @@ Experimental SFU scaffolding endpoints (`MEDIA_MODE=single_server_sfu`):
 - Running live transport + fan-out + recording on the same web server is possible for small loads but introduces significant CPU, RAM, and network egress pressure under concurrency.
 - For larger deployments, use a dedicated media plane (managed or self-hosted SFU + recorder) and keep this service focused on auth/session/control APIs.
 - For a pragmatic prototype path that keeps media on the current server, see `docs/streaming-on-web-server-plan.md`.
-- `MEDIA_MODE=single_server_sfu` currently enables scaffolding only (interfaces/config/health visibility), not full SFU media routing yet.
+- `MEDIA_MODE=single_server_sfu` requires a functional SFU engine (`MEDIA_SFU_ENGINE`, default `mediasoup`) and proper network/port exposure for WebRTC.
 
 ### API Docs
 OpenAPI docs are generated from Zod/OpenAPI definitions: