Radiating cable (leaky feeder) for rails, shuttles and cranes

A radiating cable trades the roam for a predictable field that runs along a fixed path. That suits vehicles that never leave their track. It does little for anything that does.

What it is

A radiating cable, or leaky feeder, is a coaxial cable whose outer conductor is intentionally not completely closed, so part of the signal leaks out along its length and a nearby antenna couples into it. The IEC standard for these cables names their home ground: long, narrow, semi-enclosed spaces such as railways, subways and tunnels. The cable acts as one long antenna, fed by an access point at one end.

Designs differ in how they leak. In one maker’s terms, coupled-mode cables fluctuate 20 to 30 dB peak to peak¹ along the run. Radiated-mode cables space their slots so the contributions add in phase, which narrows the gap between median and 95th-percentile coupling loss to about 3 to 8 dB. IEC 61196-4 also defines a uniformly radiating type, with slots that vary along the cable so the energy it radiates is uniform along it within a specific frequency range.

So the field is predictable, not flat. That is still a different problem from a cell edge.

Two losses set the budget

Longitudinal loss is the signal lost inside the cable, in dB per 100 m. It rises with frequency and falls with cable diameter.

Coupling loss is the drop from the power inside the cable to the power at a dipole 2 m away. Datasheets give two values: the median, and the 95th percentile, which is the one to design with.

One maker’s 1/2-inch cable shows how the two add up. At 2,400 MHz it loses 13.4 dB per 100 m, with a 95% coupling loss of 75 dB². At 5,800 MHz it loses 25.1 dB per 100 m, with a 95% coupling loss of 72 dB. So 100 m from the feed, a client 2 m from the cable sits 88.4 dB below the feed power at 2.4 GHz and 97.1 dB below it at 5.8 GHz, at 95% of points. Over 200 m the cable alone takes 26.8 dB at 2.4 GHz and 50.2 dB at 5.8 GHz. That arithmetic is why 5 GHz segments run shorter.

Then the corrections. The same application note adds 20 log(d/2) to the 95% coupling loss for an antenna at distance d, which makes 1 m about 6 dB better than 2 m and 4 m about 6 dB worse. It recommends a 10 dB margin. Compare datasheets with care: the two IEC test methods can differ by more than 10 dB for coupled-mode cables, and a figure for the best antenna orientation runs about 4 dB better than the average of three.

Why it suits vehicles on a track

Coupling depends on the distance and orientation of the mobile antenna from the cable. A vehicle on a rail holds both constant for its whole run. That is the condition the link budget above assumes, and on a track it stays true.

The rest follows. The field goes where the cable goes, including around corners and through narrow passages, which is the geometry of storage aisles, monorail loops and crane runways. With the cable run along the aisle, the path from antenna to cable is short and fixed, so the racking between the vehicle and a distant access point no longer sets the link. Along the cable, the handover between conventional cells can be avoided.

The fits are rail-guided vehicles, storage and retrieval shuttles, transfer cars, overhead cranes, electrified monorails and tunnels. An autonomous mobile robot that picks its own route gains nothing from a cable it may never pass. The automated warehouse page and the cranes and pendants page cover those sites.

Segments, feeds, and where the roam comes back

A cable run is fed by an access point and terminated at the far end. One access point can feed two segments through a power splitter, and crossing between those two is not a roam, because it is the same radio.

Crossing from one access point’s segment to the next is a roam. One industrial Wi-Fi vendor’s monorail example tells you to keep the PROFINET I/O update time no shorter than 16 ms where clients change segments³. A slotted waveguide maker quotes segments of up to 120 m per access point, with longer systems joined by roaming⁴. Put the boundaries where a short gap costs the least, such as a station or a slow section, and measure each one with the roam gap method.

Each segment is also one cell. Every vehicle on it shares its airtime, and the same maker claims up to 20 consumers per segment. The airtime guide applies in full.

MIMO from a cable

Wi-Fi rates depend on spatial streams, and a cable is one antenna. Conventionally, MIMO over leaky cable needs more than one cable, spaced widely enough to keep them uncorrelated.

Research feeds one cable from both ends. A 10 m cable measured that way as 2x2 MIMO in an anechoic chamber, from 5.15 to 5.65 GHz, gave about 82% of channel condition numbers below 10 dB⁵, with degradation near the cable ends in one design. A 2022 tunnel study at 1.8 GHz noted the method had not yet been applied in real deployments. Plan on one stream per cable until a vendor shows you otherwise.

Bands

Read the frequency table, not the word WLAN. A 1-5/8-inch tunnel cable lists WLAN among its applications but is specified only from 30 to 2,800 MHz⁶, which covers 2.4 GHz and nothing above it. One industrial vendor sells separate cables for 2.4 GHz and 5 GHz⁷. The 1/2-inch cable above runs to 7,200 MHz, with a stop band from 1,900 to 2,190 MHz, and its table spans the 5.925 to 7.125 GHz⁸ band that 6 GHz Wi-Fi uses, at 32.1 dB per 100 m by 7,000 MHz.

Slotted waveguide

The related option is a slotted waveguide: a shielded aluminum profile that carries the signal inside it, mounted along the track. One maker builds it into the protective earth conductor rail and claims 802.11n, ac and ax in 5 to 7 GHz. Another lists cranes, monorails and storage systems with travel speeds up to 300 m/min⁹. Both describe low interference with nearby wireless. Treat those as vendor claims, and survey around the profile as you would around any antenna.

Mounting

Distance matters both ways. One maker found that a cable fixed directly against concrete loses 25 to 60% more per 100 m around 2,000 MHz, and humidity and dust add up to 10%. The 1/2-inch datasheet calls for at least 80 mm from the wall, clamps every 0.5 m, and a bend radius no tighter than 200 mm. Turn the slots toward the vehicle, where coupling loss is lowest. The vehicle antenna’s height and offset set the coupling loss, so fix them in the design. The antenna placement page covers the vehicle side.

What it does not do

It does not remove roams. It moves them to segment boundaries, where you can choose and measure them.

It does not help a vehicle that leaves the track. Away from the cable, coupling loss climbs with distance, and the client needs ordinary coverage to stay connected.

It does not make Wi-Fi deterministic, and it is not a safety function. A conductor rail maker listing PROFIsafe compatibility means the protocol runs over it, not that the link stops being a black channel. The safety over wireless guide explains the difference. If you need a second path, a second cable is one way to build it, but feed it on a different channel: one maker warns that two nearby cables carrying the same signal cause large field fluctuations. The redundant radio paths page covers what makes two paths independent.

What to ask a vendor

The 95% coupling loss and longitudinal loss at your channels, and which IEC test method and antenna orientation they assume.
The full frequency range, any stop bands, and whether 5 or 6 GHz is tabulated or only named.
Segment length per access point at your band, and where the roams fall.
How the client roams between segments, and the measured gap.
Mounting distance, clamp spacing, bend radius, and the vehicle antenna position the budget assumes.
How many vehicles share one segment.

A cable gives you a field you can predict along the track. Put the boundaries where you choose, measure each one, and the roam becomes a number you know.