Wi-Fi & 5G
Wireless connectivity for distributed, mobile or hard-to-wire vision systems.
Key Advantage
"Enables embedded module deployment where Ethernet cabling is not practical."
Wireless Vision
Our series of embedded processing computers are able to exfiltrate video and processed data out over Wi-Fi / 5G. These devices can connect either to existing infrastructure or to a custom networking bubble (NGCI) produced by our Italian branch.
The Push for Local Networking
NGCIs are self-contained, portable wireless cells that create a private, low-latency communication network independent of any existing telecom infrastructure. This is critical for deployments in remote or contested environments such as construction sites, maritime operations, or temporary security perimeters where public cellular coverage is unreliable or simply absent.
By operating on dedicated spectrum, an NGCI eliminates the unpredictable latency spikes and bandwidth contention inherent to shared public networks. This guarantees a deterministic Quality of Service (QoS) for real-time video telemetry, which is non-negotiable for AI-driven decision loops running on the edge.
NGCI Technology
The NGCI platform is developed and manufactured by SMA-RTY Italy, our sister branch specializing in tactical communications infrastructure.
Architecture & Integration
The wireless radio is integrated directly into the BNX carrier board, providing Wi-Fi 6 or 5G connectivity as a native capability of the processing unit itself. From the application’s perspective, the data pipeline behaves identically to a wired Ethernet deployment, with no additional external hardware or adapters required.
Bandwidth Considerations
Raw video streams demand significant throughput. A single 1080p stream at 30 FPS in H.264 requires roughly 4-8 Mbps, while uncompressed or RAW streams can exceed 1 Gbps. Our systems address this through two strategies:
- On-device compression: The Jetson’s hardware NVENC encoder compresses video in real-time before transmission, reducing bandwidth requirements by orders of magnitude without burdening the CPU.
- Inference-first, stream-second: In many deployments, only the AI inference results (bounding boxes, classifications, alerts) need to leave the device, not the full video stream. This reduces wireless payload to kilobytes per second, making even narrow-band links viable.
Use Cases
- Mobile surveillance: Vehicle-mounted or drone-mounted systems where cabling is physically impossible.
- Temporary deployments: Pop-up security perimeters, event monitoring, or disaster response where infrastructure doesn’t exist yet.
- Industrial retrofits: Factories or warehouses where pulling new Ethernet runs is cost-prohibitive or structurally impractical.
- Maritime & offshore: Ship-to-shore or platform-to-platform vision links over water, where NGCI bubbles provide coverage that public networks cannot.
