How do we keep radio coverage working in tunnels and deep excavation works on an MRT site?

A fixed rooftop antenna or a normal above-ground repeater setup stops being useful the moment a heading, shaft or bored drive goes more than a couple of hundred metres past line of sight — the tunnel curves, the face advances, and steel lining segments start acting like a Faraday cage around anyone working near the cutterhead. The fix used on every Singapore tunnelling contract I've worked is radiating cable (leaky feeder) run behind the face and fed from a repeater anchored at the shaft head-end, not chased down the tunnel on the machine itself. The cable is extended in fixed lengths as the drive advances, in-line amplifiers go in roughly every 350–500 m, and the uplink from a 4 W handheld back to that repeater — not the downlink — is almost always the path that runs out of margin first.

Why an open-air antenna design doesn't survive past the shaft

Above ground, a two-way radio system fails gracefully: signal fades with distance and you get static before you get silence. Underground it fails abruptly. A fixed antenna radiates in every direction except the one that matters — down a curving bore — so once the face turns a few degrees off the antenna's line of sight, coverage doesn't degrade, it ends. Add the precast lining segments a tunnel boring machine (TBM) leaves behind it, each ring tied with close-centred rebar, and you get the same shielding effect we've measured in reinforced-concrete basements, except now it wraps the crew on every side rather than just overhead and underfoot. Ventilation fans, dewatering pumps and the cutterhead drive also push the noise floor up in the same band a UHF portable uses, so a system that measures fine on signal level alone can still fail on audio quality.

Radiating cable is the only medium that keeps pace with an advancing face

Leaky feeder is a coaxial cable with slots cut into its outer conductor so RF leaks in and out continuously along the run, rather than from one point like an antenna. That property is what makes it the standard, not a compromise, for a tunnelling contract: it doesn't need line of sight, it follows every curve in the bore, and it can be physically extended metre by metre as the drive advances instead of requiring a fresh antenna survey every time the face moves. In practice the cable is run along the crown or sidewall behind the trailing gear, terminated in a matched load at the open end so the unterminated tail doesn't set up standing waves, and spliced onward as each new ring of segments goes up.

The numbers that make or break the design:

• UHF two-way radio and DMR sit at 400–470 MHz; half-inch plenum coax used for the trunk feed loses roughly 4.5 dB per 100 m at 450 MHz.

• Radiating cable needs an in-line amplifier roughly every 350–500 m to overcome its own attenuation and keep the coupled signal above threshold along the whole run.

• A 40 W (46 dBm) repeater at the shaft head-end is the standard on-site source; the design threshold for reliable DMR voice is −95 dBm at the portable, with a minimum 10 dB fade margin (15 dB where the system also has to support rescue communications).

• The uplink — a 4 W portable talking back through hundreds of metres of cable and amplifier chain to the repeater — is the path to budget first; it is very rarely the downlink that fails.

Keep the repeater at the shaft, not on the machine

The mistake I see most often on a first-time tunnelling mobilisation is trying to put active equipment — a repeater, a booster, sometimes the head-end itself — on the TBM to "keep it close to the face." Don't. The machine vibrates constantly, runs hot, and every metre it advances is a metre further from wherever a technician needs to reach it for maintenance. The working pattern is a fixed head-end at the shaft or portal, feeding a radiating cable extended and spliced as the drive proceeds; only passive cable and amplifiers travel with the face, mounted so they can be serviced from the invert without stopping the drive. This also keeps the system's IMDA station licence tied to one fixed location rather than a moving one, which is the simpler answer at application time.

Cross-passages and the confined-space communication case

Cross-passages between twin bores, adits and any enclosed space a worker enters and can't simply walk out of in an emergency sit inside Singapore's confined-space safety framework, which requires a working means of communication between the person inside and a standby person before entry is permitted — check the current Workplace Safety and Health (Confined Spaces) Regulations text for the exact wording that applies to your contract. In practice that means the radiating cable or repeater coverage has to already reach the cross-passage before anyone works in it, not be discovered afterwards to have a gap. I treat every cross-passage, adit and sump as its own coverage point, walk-tested before the space is signed off for entry, rather than assuming "the tunnel is covered" automatically covers every branch off it.

What changes once the tunnel is lined and handed over

The construction-phase radiating cable system is temporary infrastructure and is normally decommissioned once the running tunnel is lined, fitted out and handed to the rail operator's own systems — it doesn't need to be integrated into the permanent railway communications network, and trying to make it do double duty usually costs more than running two separate systems. The one planning point worth flagging early is the overlap period near TOP, when construction crews, testing teams and the operator's own staff can all be working in the same tunnel section on different systems at once; that's worth a short-lived, deliberately separate channel plan rather than trying to merge everyone onto one before commissioning is complete. For context on how large these drives now get in Singapore: LTA's Cross Island Line Phase 2 contract deploys a 12.8 m diameter TBM reaching depths of up to 90 m over a roughly 5 km drive, underlining why a system that has to be physically extended alongside the machine, rather than surveyed once and left alone, is the only workable approach.

Related questions

Does the radiating cable have to be re-terminated every time the tunnel is extended? Yes — the far end carries a matched load, and that load moves to the new end of the run each time the cable is spliced forward, otherwise the open end reflects signal back down the line and creates standing waves that show up as patchy coverage near the splice.

Can a Wi-Fi or LTE push-to-talk app replace radiating cable underground? No, for the same reason cellular signal doesn't reach a basement carpark: there's no backhaul down a bore that has none of its own. A PoC or Wi-Fi PTT app still needs its own leaky feeder or repeated network run into the tunnel to have anything to connect to, so it doesn't remove the RF design problem, it just adds a second network on top of it.

How is coverage in an active drive actually verified before crews rely on it? With the same grid walk-test method used above ground — measured signal level at portable height against the −95 dBm threshold and required fade margin — but run incrementally as the face advances rather than once at the end, since the last few hundred metres behind the cutterhead are the section most likely to be short of amplifier reach. See our guide on how in-building radio coverage is measured and passed for the survey method itself.

How Suneast approaches this

Suneast designs and installs radiating-cable and repeater systems for tunnelling, deep excavation and shaft works across Singapore, sized and extended in step with the drive rather than surveyed once and forgotten. We anchor the head-end at the shaft, budget uplink and downlink separately against the −95 dBm threshold and the required fade margin, and walk-test every cross-passage and adit before it's signed off for entry — the same discipline we apply to our permanent car-park and tunnel DAS work, covered in our piece on leaky feeder versus discrete antennas for car parks and tunnels.