SoftBank demonstrated dynamic nullforming on an aerial base station, letting a non-terrestrial network (NTN) share the same 1.7 GHz spectrum as terrestrial 5G without interference. It addresses real spectrum scarcity and coexistence risk for operators, but nullforming at scale, control-plane overhead, and RAN scheduling logic remain open technical questions.
SoftBank just ran an interesting experiment in Japan: use dynamic nullforming on an aerial base station so a non-terrestrial network (NTN) can reuse the same 1.7 GHz band as a terrestrial 5G site without stepping on its toes. Instead of carving out "special satellite bands", the aerial platform projects a null toward the ground site and keeps retuning that null as it moves, suppressing interference while still serving a wide area below.
If this scales, it changes the usual narrative that NTNs must live in separate spectrum to avoid chaos.
What SoftBank Actually Did
A few details matter here. SoftBank used a light aircraft to emulate a HAPS platform, flying at around 3,000 meters with a maximum ground speed above 200 km/h, and mounted a base station on it that transmitted in the same 1.7 GHz band as a terrestrial site. By dynamically steering a radiation null at the terrestrial base station's location, they could keep interference "stably suppressed" in its vicinity while still providing wide-area aerial coverage.
SoftBank's stated goal is to take this toward HAPS-based mobile services for hard-to-reach areas and emergency communications, with a pre-commercial launch still notionally targeted for 2026 using LTA platforms from Sceye, and HTA platforms later.
Why This Matters For Operators
From an operator's point of view, dynamic nullforming addresses two chronic anxieties around NTNs:
Spectrum scarcity
Reusing existing mid-band (like 1.7 GHz) instead of hunting for "clean" NTN bands could drastically simplify spectrum strategy and regulatory wrangling.
Coexistence risk
Demonstrating that you can fly an aerial base station over an existing macro site, share the band, and still not degrade terrestrial QoE is a strong story for regulators and CFOs.
But this also raises hard questions: how do you industrialize geo-accurate null steering in real-world HAPS or LEO fleets with fast dynamics, multipath, and imperfect location data? What happens when you have dense grids of terrestrial sites plus multiple aerial platforms all trying to carve out dynamic nulls at once?
The Technical Questions We're Not Talking About Enough
If you peel back the PR, several deeper design questions pop up that I'd love to see the community debate:
Nullforming at scale
It's one thing to steer a null at a single known terrestrial BS; it's another to maintain accurate nulls for dozens of sites while your platform is moving at high speed and experiencing drift, attitude changes, and beamforming calibration errors.
CSI and control-plane overhead
How frequently do you need to refresh geometry and channel state to keep the null "tight" enough at 3,000+ meters, and who owns that control loop (NTN controller, terrestrial RAN, joint scheduler)?
RAN scheduling logic
Do we treat NTN cells as "overlay neighbours" with explicit protection rules, or do we push coexistence logic into near-RT RIC apps (e.g., xApps/rApps) that understand both aerial and ground layers?
Business And Policy Implications
SoftBank's trial hints at some non-obvious business and regulatory angles too.
- Spectrum licensing models: if NTNs can reliably share terrestrial bands, do we see new licensing categories where terrestrial MNOs lease "vertical reuse rights" to NTN operators, or do integrated players like SoftBank keep the whole stack?
- Coverage economics: if an aerial layer can reuse the same mid-band spectrum with acceptable interference, does it become a more attractive option for rural coverage and disaster-recovery versus incremental macro build-outs?
- Standardization gap: 3GPP has laid groundwork for NTNs, but do we need more explicit specs for dynamic nullforming-based coexistence, especially around measurement/reporting and cross-layer coordination?
Over To You — Let's Debate
I'm curious how others in the community see this, especially those working on RAN design, NTN, or spectrum policy. A few prompts to kick off the discussion:
- Do you believe dynamic nullforming is resilient enough for large-scale NTN–terrestrial spectrum sharing, or is this only viable in low-density scenarios?
- Where should the intelligence live for this kind of coexistence — in the NTN platform, in the terrestrial RAN (e.g., via RIC/xApps), or in a higher-level spectrum controller?
- If you're an operator, would you rather keep NTNs in separate bands for simplicity, or accept added design complexity in exchange for reusing your existing 4G/5G spectrum?
- How would you model and test worst-case interference events here — e.g., sudden platform misalignment, GNSS failure, or a surge of terrestrial traffic near a protected site?
- What KPIs would convince you that HAPS-based sharing is "production ready" (e.g., % time interference below threshold at macro sites, delta in cell-edge user throughput, outage statistics during aerial maneuvering)?
If you've run similar trials or are exploring NTN integration in your roadmap, I'd love to hear what worked, what broke, and what you'd design differently after seeing SoftBank's results.
