Zero-bit · Zero-watt networks

A new 3D network architecture, driven by tethered drones, redefining zero-energy cellular networks.

ZZNets introduces the Hypercell — an aerial or non-terrestrial coverage layer that takes over blanket coverage at night, letting terrestrial cells sleep without losing service.

Explore the objectives Read the publications

67%Terrestrial energy cut in simulation

12.5%Savings with the shutdown protocol

70×Faster AI channel generation

1stExperimental Hypercell validation

The vision

Cells that sleep, coverage that stays.

Today's energy-saving mechanisms hit a wall: terrestrial coverage cells must stay awake to guarantee service. ZZNets moves that role to the sky, so the ground layer can shut down when nobody needs it.

Zero-bit, zero-energy network vision

Protocols for a Hypercell solution that enables terrestrial cell shutdown and delivers the 6G vision of zero-bit, zero-watt networks.
Tethered drone Hypercells

A proof of concept for a low-energy Hypercell: a 5G base station mounted on a tethered drone providing blanket night-time coverage.
Sky-to-indoor coverage study

Measuring, for the first time, the indoor coverage provided by a drone-assisted base station — critical, since 80% of traffic is consumed indoors.
AI-driven 3D network modelling

Lightweight AI estimators fused with an expert-based network model (ABM theory) to predict energy efficiency for any 3D configuration.
Advanced network optimisation

Tailored AI-based optimisation frameworks for large-scale, stochastic problems, maximising energy efficiency and user experience.

Scientific objectives

Four objectives, one architecture

See them in detail

SO 1

Hypercell energy-saving solution

Protocols and algorithms pairing drone-assisted Hypercells with terrestrial capacity cells.

SO 2

Sky-to-indoor coverage

Quantifying how well a flying base station offloads indoor traffic in dense urban scenarios.

SO 3

AI-based network modelling

Predicting the statistical energy efficiency of any 3D network configuration.

SO 4

AI-based network optimisation

Off-line and on-line frameworks for large-scale, stochastic, multi-objective problems.

Work programme

Activities

All activities

Hypercell architecture and carrier-shutdown protocol

Two candidate architectures, a dynamic cell-pairing algorithm and a distributed shutdown protocol, supported by AI building blocks for traffic-, quality- and energy-aware decisions.

12.5%
Energy saved

Propagation campaigns and the experimental platform

Outdoor-to-indoor measurements across five frequency bands and three floors, plus the first Hypercell platform built on commercial 5G equipment and a tethered drone.

< 4 dB
Model RMSE

AI modelling and the network digital twin

Generative channel modelling, traffic forecasting from NWDAF data and SINR prediction, integrated into the GPU-accelerated Giulia system-level simulator.

70×
Faster on GPU

Optimisation foundations for AI-based Hypercell operation

A theoretical optimisation framework and the BLASTER algorithm, extended with Bayesian optimisation, reinforcement learning and transfer learning alongside UPF and Huawei.

67%
Energy saved

Key results

What the project has proven so far

From a theoretical formulation to a drone flying a commercial 5G cell over the UPV campus.

67% Energy reduction Achieved by the BLASTER algorithm during low-traffic periods, while preserving performance.

< 4 dB Propagation accuracy RMSE of the new multi-floor outdoor-to-indoor model across 2, 3, 7, 10 and 14 GHz.

70× Simulation speed-up Generative AI channel model versus conventional stochastic simulation on GPU.

1–2 dB SINR prediction error From standardised SSB and CSI-RS measurements, enabling proactive shutdown decisions.

Outputs

Dissemination

All outputs

Journal articles

Peer-reviewed papers in IEEE Transactions

Other publications

Conferences, a book chapter and open-access outputs

Conferences and talks

Keynotes, tutorials, invited talks and posters

Doctoral theses

Ongoing at UPV-iTEAM

Internationalisation

Workshop, Huawei project and an MSCA-DN proposal

Knowledge transfer

Impact and transferable results

Partnerships

Built with academia and industry

All collaborations

Academic · 3 research groups

CITIC-UGR, Universitat Pompeu Fabra and ETSIADI-UPV

Professional propagation measurement campaigns, methodological expertise in Bayesian optimisation and reinforcement learning, and the drone operations behind the Hypercell platform.

From research results to industrial validation

Expertise on energy-efficient RAN, carrier shutdown and digital twins — now continuing through a new joint industrial research project on 5.5G/6G simulation.