Why BMS/SCADA burn mobile data
Defaults used during commissioning—1–10 s polls, verbose payloads, cloud sync and open remote sessions—often remain in production. On cellular, every byte costs.
Temperature/energy rarely need sub-10 s scans. Aim for 60–300 s, or use Change-of-Value with sensible deadbands (≈0.3 °C, 2% RH).
Uncompressed JSON and frequent TLS handshakes waste bytes. Batch points, compress where supported, and keep TLS sessions alive.
Nightly full uploads dwarf telemetry. Push deltas/aggregates and schedule off-peak or over non-cellular backhaul.
RDP/VNC idles at 50–150 kbps. Use VPN, set timeouts, avoid all-day sessions.
BMS data usage calculator
Estimate daily/weekly/monthly data from telemetry settings. Try 1000 KB payload at 5 s to visualise the 17 GB/day scenario.
Public IP noise: why “blocked” still costs
Routers on fixed public IPs attract constant probes—ICMP pings, TCP SYN scans, HTTP(S) bursts. Packets are counted when they reach the modem, even if the firewall drops them. Whitelisting reduces successful sessions, not the inbound noise itself.
| Source | What it is | Why it adds up | Mitigation |
|---|---|---|---|
| ICMP pings & TCP SYN scans | Automated reachability and port discovery | Dozens to hundreds of sources, all day, every day | Flood/scan protection; hide behind private APN/VPN |
| NTP time sync | Devices polling pool servers | Many small packets at short intervals | Space out NTP; local NTP; reduce retries |
| Keepalives/heartbeats | VPN or platform liveness checks | Short timers (1–10 s) create a steady baseline | Use 60–120 s where possible |
| Reboots/forensic uploads | Logs/state sent after restart | Hundreds of KB per event, multiplied across devices | Reduce crash verbosity; stagger reconnects |
Public IP noise simulator
Simulate unsolicited probes plus routine housekeeping. Compare with the BMS result to understand total exposure.
Busy IP? Try 200–500 sources at 5–10 s intervals. You’ll see how “idle” becomes costly.
What to check in router logs and counters
Track inbound when BMS is quiet. Set caps with 50%/80% alerts.
LAN idle but WAN inbound ticking? That’s likely scans/floods.
Look for “syn flood”, “port scan”, “icmp flood”, “drop”, “invalid”. After enabling protections, drops should stabilise.
Remove unused forwards. Change management ports. Whitelist known IPs only.
Best practice: getting from GB to MB
- BMS tuning: 60–300 s polls; COV with deadbands; batch and compress; avoid all-day RDP/VNC.
- Reduce exposure: Prefer private APN/CGNAT with outbound VPN. If public IP is mandatory, default-deny firewall, minimal forwards, source whitelists, enable flood/scan protections.
- Quiet housekeeping: NTP hourly to 12-hourly; keepalives 60–120 s; reduce management inventory frequency; disable speed tests.
- Alerting & RF: Data caps with alerts; ensure good RSRP/SINR to avoid retransmit overhead.
Frequently asked questions
Why is my BMS using so much mobile data?
Aggressive polling/logging and verbose payloads. A single 1 MB/5 s job is ~17 GB/day. Tune intervals, enable COV, batch/compress.
Does whitelisting stop the background data?
No. It stops unauthorised sessions completing. Probes still arrive and are counted. Flood/scan protections reduce impact; private APN/VPN removes exposure.
What’s a realistic daily target?
Telemetry: ~200–400 MB/day for many sites. Background idle: <10–50 MB/day on private APN/VPN; public IP may be far higher if unfiltered.
Bottom line: measure both sides—BMS telemetry and public exposure. Fix configuration, quiet the housekeeping, and remove exposure where you can. The data bill—and stability—improves fast.