Two Allen-Bradley controllers separated by about five statute miles of pipeline right-of-way must exchange leak-detection telemetry with timing resolution on the order of tens of milliseconds. Shared VSAT Ethernet that also carries SCADA, video, voice, and crew traffic cannot provide that determinism: GEO latency alone is hundreds of milliseconds, and contention plus rain fade add jitter that Class 1 EtherNet/IP and tightly polled Class 3 messaging will not tolerate. Fiber is often rejected on easement cost. The engineering answer is a dedicated industrial wireless IP hop that the PLC pair can treat like a copper or fiber segment.

Own the spectrum and the RF path. Do not share leak-detection CIP traffic with welfare bandwidth on a satellite hub.

Why shared satellite Ethernet fails

• Bandwidth contention collapses TCP under loss; RPI-driven Class 1 traffic cannot absorb the jitter.
• One-way GEO delay of roughly 250–280 ms plus hub queuing routinely pushes round trips past one second.
• Ku/Ka rain fade along storm cells breaks the path precisely when leak algorithms need continuity.

A point-to-point industrial radio link bounds one-way latency to free-space delay plus radio store-and-forward—typically under five milliseconds at five miles.

Technology families

At five miles with clear LOS, plants usually choose 900 MHz SCADA radios for resilience or 5 GHz PtP bridges when HMI graphics and camera traffic share the hop. 900 MHz yields roughly 100 kbps to a few Mbps usable IP under FCC hopping rules—adequate for leak polling, not for video. 5 GHz bridges deliver tens to hundreds of Mbps but demand cleaner Fresnel clearance.

RF path engineering

Line-of-sight through binoculars is not enough. Keep at least 60% (preferably 80%) of the first Fresnel zone clear. Midpoint Fresnel radius scales with path length and inversely with frequency:

r_mid ≈ 17.32 × sqrt(d_km / (4 × f_GHz))
5 mi (8 km) examples:
  900 MHz → ~26 m radius (~15.5 m at 60% clear)
  5.8 GHz → ~10 m radius (~6 m at 60% clear)

Build the link budget as TX power + antenna gains − FSPL − feedline/connector losses, then require 10–20 dB fade margin for snow and icing. Mount antennas above the obstructing tree line or berm; document tower heights on as-built drawings.

5 GHz operational caveats

Prefer non-DFS channels where regulation allows; DFS vacates can interrupt CIP for minutes. Proprietary airMAX/Mimosa protocols require matched pairs—do not expect a laptop to join the bridge. As Layer-2 devices, transparent bridges pass EtherNet/IP unmodified when no NAT sits in between.

PLC integration steps

1. Place radios on a dedicated VLAN or physically isolated segment for leak-detection traffic.
2. Assign stable IPs to both controllers and both radio management interfaces.
3. Validate ping RTT and sustained CIP MSG or produced/consumed connections for hours across weather changes.
4. Set RPIs and MSG rates within measured throughput, not laboratory Ethernet assumptions.
5. Add supervised heartbeat tags so a radio fade raises a pipeline alarm before algorithms go blind.

Link budget and latency soak test

Commission with a spectrum analyzer or radio RSSI/SNR logs, not only a successful ping. Common failures include Fresnel blockage from “almost LOS” terrain, DFS channel hops, shared switches that reintroduce VSAT-like contention, and undersized 900 MHz throughput for added video. Keep radio spares, connectors, and grounding kits with other remote PLC and PAC assets on the pipeline.