SustainabilityPower Optimization: Waking Up Nodes on Demand
How we configured custom environmental sensors to sleep deep and wake up only on drone passes, achieving over 12 months of battery life.
Loading...
We specialize in developing cutting-edge AI analytics, autonomous systems, IoT networks, and satellite communication solutions that transform industries.
Zorzos Projects (Zorzos Konstantinos Single Member Private Company) is a multidisciplinary technology company based in Athens, Greece. We collaborate with high-caliber scientists and systems engineers to develop robust, customized hardware and software platforms that solve real-world challenges.
From custom embedded sensor systems to high-speed communication routing and real-time edge intelligence, we help clients research, prototype, and scale technology solutions for complex industrial and agricultural applications.
Custom low-power sensing nodes with RF wake-up, PCB design, firmware development, and wireless sensor network deployment for industrial and agricultural applications.
UAV mission planning, flight controller configuration (ArduPilot/SITL), path optimization using TSP heuristics, and autonomous robotic fleet coordination.
Non-terrestrial network integration via GEO satellites, high-altitude relay systems, encrypted OpenVPN tunnels, and payload compression for bandwidth efficiency.
Machine learning inference on edge devices, computer vision for crop disease detection, predictive analytics, and real-time anomaly detection systems.
Long-range, low-power wireless networks using LoRa/LoRaWAN protocols. Custom protocol stacks, gateways, and mesh networking for remote deployments.
End-to-end data pipelines, real-time dashboards, custom API development, and visualization platforms that transform raw telemetry into actionable intelligence.
An innovative remote monitoring system using a UAV to fly over fields, activate dormant ground nodes via LoRa, retrieve environmental telemetry, and synchronize data through a GEO satellite link to an edge processing server.
Deploying high-altitude network relays over software-defined radio arrays to provide direct-to-cell coverage and emergency text alerts in remote regions without cellular service.
Deploying neural networks on low-power edge gateways to perform real-time image recognition on plant leaves, automatically classifying plant stressors and scheduling irrigation.
SustainabilityHow we configured custom environmental sensors to sleep deep and wake up only on drone passes, achieving over 12 months of battery life.
NavigationTransitioning from complex MAVLink broadcast parsing to direct REST HTTP triggers during autonomous TSP flight paths.
Smart FarmingAn analysis of our decision to pivot away from ground-probing NPK sensors in favor of hyper-accurate ambient BME280 metrics.
6G RelaysExploring the VPN tunneling requirements for transmitting compressed sensor telemetry over high-latency Agrarian Testbed links.
SecurityA breakdown of our end-to-end encryption pipeline between the ground nodes, the Pi 5 proxy, and the cloud database.
AI & VisionHow we deployed YOLOv8 on low-power Edge TPUs to automatically identify rust and blight on agricultural plant life.
MilestonesReviewing our recent field trial results, verifying our KPIs, and establishing the roadmap for the final 5 months.
Have a project in mind? Reach out and our engineering team will get back to you.
Alexandroupoleos 16, Metamorfosi
Athens, Greece, PC 14452
info@zorzosprojects.com
Zorzos Konstantinos Single Member P.C.
VAT: EL96752247
We are always looking for new challenges. Send us an email with your project details and we will get back to you.
Email UsData-driven precision farming for remote settings lacking terrestrial network infrastructure.
Low-power nodes scattered in the field measure environmental telemetry. Nodes stay in deep sleep to maximize battery longevity.
As the UAV flies its route, it emits a LoRa RF wake command, triggering the dormant nodes to wake up on-demand.
Woken nodes log telemetry locally to MicroSD for redundancy, then transmit the data package back to the UAV via LoRa.
The companion computer (Raspberry Pi 5) on the drone encrypts and compresses the data, routing it via a secure GEO satellite OpenVPN link.
The edge processing server receives the satellite stream, parses the measurements, and updates the farmer's dashboard in real-time.
Remote farmlands lack internet and cellular coverage. Conventional IoT designs need per-sensor cellular connections, making wide deployment uneconomical.
An optimized autonomous UAV flight cycle coordinates with ultra-low-power ground sensors. Sensors stay in deep sleep until the drone emits a LoRa RF wake command, collects telemetry, and forwards it via GEO satellite.
868 MHz LoRa broadcast activates dormant sensors
MicroSD logging ensures data redundancy
HTTP-to-serial middleware on RPi 5
Encrypted OpenVPN tunnels via GEO
Detailed architecture, hardware wiring, and protocol diagrams for the AGRIDRONE platform.
Communication pipeline routing telemetry from deep-sleep sensor nodes through LoRa, RPi 5 companion computer, and GEO satellite relay to the cloud database.
Sequence diagram showcasing the RF LoRa wake-up command trigger, telemetry packet assembly, local logging, and satellite proxy forwarding stages.
Pico 2 W target board mapping to the SX1262 LoRa module transceiver and BME280 sensor breakout pinouts.
Physical trace routing and double-SPI SPI0/SPI1 bus isolation preventing clock collisions between the LoRa transceiver and MicroSD logger.
M5 Milestone — all operational KPIs met or exceeded
Sensor activation success during UAV passes
Missions completed without intervention
Sensor node operational life
Collection to dashboard visualization
Successful satellite data delivery
Field coverage per flight mission
| Category | Metric | Target | Achieved | Status |
|---|---|---|---|---|
| System Perf. | Wake Success | >95% | 96.5% | Achieved |
| Flight Completion | >90% | 95.2% | Achieved | |
| Sat Transmission | >80% | 99.5% | Achieved | |
| System Uptime | >99.5% | 99.8% | Achieved | |
| Technical | Battery Life | >10 mo | 12+ mo | Achieved |
| Coverage Area | >5 ha | 5.2 ha | Achieved | |
| Data Latency | <30 min | 11 min | Achieved | |
| Dissemination | Website Visitors | 100+ | 85 | In Progress |
| Social Posts | 5+ | 5 | Achieved |
10-month plan from February to November 2026.
Requirements, architecture, protocol definition.
UAV, sensors, satellite integration.
ML models, APIs, application.
Field test, final validation.
Deploying high-altitude network relays over software-defined radio arrays to provide direct-to-cell coverage and emergency text alerts in remote regions without cellular service.
Deploy lightweight solar-powered relays at 20km altitude to establish wide area coverage cells.
Configure software-defined radios dynamically to bind direct-to-cell frequencies without hardware swaps.
Synchronize remote field sensor transmitters to the balloon mesh using low-latency handshakes.
Tunnel encrypted data streams over secure OpenVPN channels to the edge cloud gateways.
Deliver emergency alert push SMS to standard handheld mobile devices located in regional blackout zones.
Theoretical designs and link budget diagrams for high-altitude non-terrestrial cells.
Design overview of the HAPS transceiver payloads routing high-frequency channels to satellites.
Software-defined frequency boundaries allocating bandwidth dynamically based on active local devices.
Target Key Performance Metrics
High-throughput data link speed
Operating spectrum frequency band
Network availability rate
Lab-testing frequency bounds and link budget simulators.
Fitting solar transceiver array structures onto high-altitude platforms.
Deploying test balloon relays to confirm localized emergency cell alerts.
Deploying neural networks on low-power edge gateways to perform real-time image recognition on plant leaves, automatically classifying plant stressors and scheduling irrigation.
Unmanned drones sweep crop fields to capture high-definition leaf image frames.
Run bounding-box filtering locally on the camera streams to crop and isolate leaf targets.
Evaluate Convolutional Neural Network weights locally on an Edge TPU (YOLO v8 format).
Determine plant health metrics (dehydration, rust, blight) to generate soil directive flags.
Broadcast instructions to localized drip irrigation systems to adjust water release dynamically.
CNN tensor layouts and vision pipeline diagrams optimized for low-power edge accelerators.
Visualization of layer reductions (convolution, maxpooling, and dense classifications) configured for TPU compiling.
Data routing schematic showcasing local image pre-processing, quantization, TPU execution, and solenoid feedback loop.
Target Edge Performance KPIs
In-field classification success rate
Leaf classification inference time
Hardware accelerator power draw
Compiling leaf dataset libraries and optimizing weights in YOLO.
Quantizing and compiling CNN models to low-power Edge TPU binary profiles.
Deploying onto custom gateways to automate irrigation schedules based on leaf telemetry.
