Bandwidth Budget on a Shared 100 Mbps Uplink
A logistics depot installs 14 fixed 1080p cameras on a single switch that reaches the control room over a 100 Mbps microwave link shared with office traffic. The integrator initially configures every camera at 25 fps, H.264, 6 Mbps constant bitrate. Office users report the link saturates at midday. The customer wants continuous recording at the NVR and live viewing of 4 cameras at once, and refuses to drop resolution below 1080p. Night scenes are lit by sodium lamps and motion is mostly forklifts moving at walking pace. The NVR has 4 TB usable storage and must retain 30 days.
1. How would you restructure the video streams to fit the link and the storage target, and what do you verify before committing?
Read the worked answer
The arithmetic drives everything. 14 cameras × 6 Mbps = 84 Mbps of camera traffic alone, leaving almost nothing for office data on a 100 Mbps link. Two-lever fix:
- Separate recording from viewing. Enable a second (or third) stream per camera. Record the high-resolution stream at reduced frame rate (12–15 fps is ample for walking-pace forklifts) and use a low-bandwidth stream (e.g. 640×360, 5–8 fps) for the 4-up live view. Live viewing then consumes ~1–2 Mbps total instead of 24 Mbps.
- Switch to H.265 where supported to cut bitrate roughly 30–50% at equal quality, then cap the recording stream with VBR plus a maximum bitrate rather than CBR.
Storage check: 4 TB / 30 days ≈ 133 GB/day ≈ 12.3 Mbps aggregate for all 14 cameras - about 0.9 Mbps per camera, which is tight for 1080p. Either lower the recording frame rate further, accept smart/event-based recording, or add storage. Verify actual bitrate with the camera's live statistics under real scene load (forklift motion, sodium-lamp noise) before signing off; sodium lighting can raise noise and inflate bitrate at night.
2. The customer insists on 25 fps for evidence. What is your response and what do you measure?
Read the worked answer
Explain the trade-off honestly: frame rate aids smoothness of fast motion, but for walking-pace forklifts 12–15 fps preserves identification-relevant detail, and bitrate/storage cost scales nearly linearly with fps. If 25 fps is a hard requirement, the link and storage must grow - quantify it: 14 × 25 fps at a realistic H.265 VBR average (say 3–4 Mbps) is 42–56 Mbps, still leaving headroom only if office traffic is shaped/QoS-prioritized. Measure real average and peak bitrate per camera over 24 h including night noise, confirm switch port utilization and microwave link headroom, and recalculate retention days against 4 TB. Present the customer with a costed choice rather than silently degrading quality.