Background
compliance-automation toolsge runs a national 5G network across a multi-vendor RAN. Their optimization loop leaned on drive-test campaigns and per-vendor tools that each modeled a slice of the network, so a change in one place had unpredictable effects elsewhere — and validating coverage meant putting trucks on the road.
The challenge
Drive-testing was slow, expensive and stale by the time it was analyzed. KPI-by-KPI manual RAN tuning improved one metric and quietly hurt another. And an energy-saving push kept tripping over the risk of dropping coverage or breaching an SLA, with no safe way to test the trade-off first.
What we did
- Built a network twin from the topology, terrain, multi-vendor config and live counters, with ray-traced RF propagation over the real geography.
- Validated the twin against existing drive-test and counter data until its coverage and KPI predictions tracked the real network.
- Ran closed-loop RAN optimization — scheduler, power, tilt and handover — against the twin for throughput, latency and energy together, not one KPI at a time.
- Tested every energy-saving and slicing change on the twin before it touched a live cell.
“We used to find coverage problems with trucks. Now the twin finds them before we roll one — and it proved the energy savings wouldn't cost us an SLA.”
The outcome
- Most routine drive-testing was retired in favour of ray-traced predictions validated against periodic ground truth.
- RAN changes were tuned for throughput, latency and energy jointly, with the twin catching the regressions manual tuning used to ship.
- An energy-saving programme went live after being proven on the twin to hold coverage and SLA, rather than being shelved as too risky.
- The twin now runs continuously, with live counters returning into it so its predictions sharpen over time.
One detail worth calling out
The believable moment was a back-test: fed only last quarter's config and counters, the twin reproduced a coverage hole the network had actually developed after a change. Once the RF team saw it predict a problem they'd lived through, they let it vet changes before the field did. The energy programme was the payoff — it shipped because the twin could prove the SLA held.