AI device for data transport in Harsh environment

Problem. Sensors in fields, greenhouses and remote buildings often work where Wi-Fi and mobile coverage is weak or crowded. When a wireless link degrades, readings silently go missing, and nobody knows whether the cause is distance, interference or broken hardware. Solution. Aglink is an wireless data transport system with the model that analyzes the packet reception history of a single node on a single channel over the last 30 minutes. Based on this, it estimates the probability that the channel will lose 3 or more packets out of 30 in the next 5 minutes. Audience. Farmers, greenhouse operators, researchers and students who need reliable low-cost sensing and want to understand their wireless link. Status. Working prototype. The node and gateway firmware are written. The packet format, CRC check and data output were tested on a computer with simulated radios, and the circuit is built and simulated. Field tests on real hardware are the next step. Why it deserves the nomination. It makes a cheap radio measurable and creates the data foundation for AI-based link prediction.

Номинация

AI-стартап года

Компания

Ruta

Проект

Aglink

Сфера

AgriTech / сельское хозяйство

Стадия

MVP

AI-компонент

The ML model that analyzes the packet reception history of a single node on a single channel over the last 30 minutes. Based on this, it estimates the probability that the channel will lose 3 or more packets out of 30 in the next 5 minutes.

Влияние / результаты

Predicts packet loss 5 minutes before it happens, for each node and radio channel. - Accuracy ≥90%, precision ≥75%, recall ≥70% (7 of 10 loss events flagged in advance). - Average precision ≥0.80 (prototype today: 0.71 on simulated data), ≥20% better than a rule-based baseline.

Инновация

The innovation of the project lies in integrating AI with diversity reception,which provides redundant security for data transmission and significatly minimizes data loss rates

Связь с Азербайджаном

Built in Baku by RUTA -the model is designed and trained in-house by a local team. Serves agriculture, one of the four priority sectors of Azerbaijan's AI Strategy for 2025-2028. Covers a known gap: Azerbaijani agriculture has no field data for AI. AgLink keeps sensor data coming from fields with poor mobile coverage, providing local farm AI (irrigation, disease detection) with data to learn from. Edge AI: predictions are made on the modem itself, so field data does not have to leave the country for AI processing.

Команда

1–5 человек

Бизнес-модель

B2B2C модель

Дата запуска

2026-09

← Все номинанты