Mobile networks turn sensors: Ericsson's ISAC drone detection and the private 5G question

The Register reported the announcement on 2 November. It is the most concrete commercial timeline yet for turning cellular infrastructure into a distributed sensor array. Ericsson says the system needs sensing-capable radios, dedicated compute and software. It complements radar rather than replacing it.

How it works is simple in principle. When one cell site transmits, multiple receiving sites pick up the reflection off a drone. The network analyses those reflections to fix a position and follow it.

The Register quotes CCS Insight research director Ian Fogg. He says mobile networks suit this because their antennas already cover low elevations where drones fly. A standards-based 6G method could improve precision by letting different suppliers' base stations interoperate. That interoperability point is the one to watch. A sensing market built on proprietary radios looks very different from one built on a common standard. Fogg's framing puts the precision gains on the standards side, not on any single vendor's hardware.

Two trials, one week

The commercial news landed alongside military testing. Tom's Hardware reported on 2 November that NATO ran its Digital Backbone Experimentation exercise, DiBaX 2026, between 19 and 30 October in Latvia. The exercise investigated whether 5G towers could passively detect and track flying objects. The same piece notes that AT&T and Ericsson demoed drone detection using network sensing over a 5G network near a large sporting venue in the U.S.

That is where the private network industry enters the picture. A detection capability sold to a defence ministry is one product. The same radios inside a factory, port or airport campus are another. Ericsson's own framing in The Register points at the second. Per Fogg, sensing can also track boats in a port, traffic patterns or weather. Tom's Hardware raises the obvious counterargument. It warns that 5G masts could become an even more attractive target if they double as sensors. Neither outlet reports pricing. Ericsson has not published accuracy figures for the small-scale pilots.

The demand side is moving too

The sensing push is not happening in a quiet market. Light Reading reported on 29 October that daily multimodal AI use in India is projected to rise from 22% now to 43% by 2030, against a global forecast of 30%, citing Ericsson ConsumerLab research. The article quotes Nitin Bansal, Ericsson's managing director for India and head of networks for South East Asia, Oceania and India. He says the networks are ready for AI but must be optimized for each application's requirements.

Uplink is weaker indoors and at cell edges because phones transmit at much lower power than base stations.

Arun Menon of MTN Consulting gave that assessment, also in Light Reading. It matters for anyone planning sensing or AI traffic on a private 5G deployment. The weak link is the device side, not the tower.

Optical and quantum work is running in parallel. Light Reading reported on 29 October that Lumen announced three initiatives. One is work with Emerald AI to shift AI workloads between locations based on power availability. Another is research with The Ohio State University on quantum connectivity over fiber. The third is quantum-resistant security through a Fortinet-based SASE product. The same article cites a Gartner prediction. By 2030, data center power limits will force 75% of infrastructure and operations leaders to mandate energy-aware AI workload placement.

What the hardware looks like

On the chip side, Semiconductor Engineering's 3 November roundup lists new papers. One is imec and Vrije Universiteit Brussel work on embodied emissions of 2.5D and 3D scaling in chipletized GPUs. Another is a Materials Center Leoben, PVA TePla and ASE paper on AI-based image enhancement for detecting flip-chip interconnect defects. Its 30 October week-in-review adds that ASE forecasts a doubling of advanced-packaging revenue in 2027. It also says TSMC is reportedly considering a 20% price increase for CoWoS.

None of that is private 5G, but it is the supply chain underneath it. Service providers are already testing the sensing idea on their own turf. Turkcell trialled camera-free 5G sensing on a private network on 22 October, according to a headline in a context list.

What remains unproven is scale. Ericsson's 2027 pilots are small and sensing-only. The company has not said how many customers are signed. NATO's Latvian exercise ran for twelve days. Turning either into coverage across a country, or a factory floor, is the part nobody has demonstrated publicly yet.