Designed for a field where connectivity cannot be assumed.
RiceGuard separates small numeric packets from image traffic and gives each the transport it needs. The station keeps recent information locally, uses LoRa for long-range low-volume communication and sends larger payloads through the cellular gateway.
One station. Two transport paths.
The independent LoRa heartbeat can still signal that the station is present
Small, resilient packets and heavier data do different jobs.
LoRa carries the station heartbeat and small readings. The cellular gateway carries images and larger synchronisation jobs.
Slow and long-range by design.
LoRa operates in Thailand's 920–925 MHz ISM band across eight channels with spreading factors from 7 to 12. Its long-range, low-volume profile suits the station heartbeat and small sensor readings from the field.
Capacity for the larger payload.
An e-SIM cellular gateway carries images and larger data synchronisation over the 4G/5G bands used in Thailand. If cellular service is interrupted, the independent LoRa heartbeat can still indicate that the station is present.
A link interruption delays the data. It does not erase it.
The station is designed to hold seven days of readings locally. If the link drops, information remains at the pole. When connectivity returns, the system can synchronise the delayed history instead of treating the outage as data loss.
Keep working locally, then reconcile stored readings when the network returns.
07 Days retained locally
Local buffer- D1
- D2
- D3
- D4
- D5
- D6
- D7
Route by volume, purpose and continuity need.
RiceGuard routes each type of information according to its size, urgency and role in keeping the station visible.
| Payload | Primary path | Purpose | |
|---|---|---|---|
| 01 | Heartbeat | LoRaWAN | Signals independently that the station is present and reporting |
| 02 | Small numeric readings | LoRa | Moves low-volume field measurements over the long-range link |
| 03 | Images | 4G/5G | Carries camera files and their associated metadata |
| 04 | Bulk synchronisation | 4G/5G | Sends locally retained history after connectivity returns |
Six documented services, each with one clear role.
The architecture keeps incoming device information, sensor history, images, metadata and fusion work in services suited to each job.
- 01 AWS IoT Core
Receives information arriving from field devices.
- 02 AWS Lambda
Routes incoming information to the required processing and storage services.
- 03 Amazon Timestream
Stores timestamped sensor series for analysis through time.
- 04 Amazon S3
Stores camera images and larger visual files.
- 05 Amazon DynamoDB
Keeps metadata, image objects and labels connected.
- 06 Amazon SageMaker
Supports multi-modal fusion and heavier model work.
- Field
- Ingestion
- Storage
- Fusion
A soil reading, camera frame and weather record become more useful together.
The fusion layer combines structured measurements with image features, then connects them with crop and variety knowledge before information reaches the presentation layer.
- 01 Structured measurements
Soil + water + weather
- 02 Image features
Camera frames + labels
- 03 Crop + variety knowledge
Context for interpretation
- 04 Presentation layer
Human-readable result
Cloud services organise the result. The platform makes it usable.
Continue to see how operational, farmer-facing and phenology views present the information to different users.