AIR-T: When Artificial intelligence meets with Radio – Transceiver 

AIR-T: When Artificial intelligence meets with Radio – Transceiver 


The concept of Software-defined radio is not new, it has been around for years. It is basically about moving complex signal handling over to a digital software platform via PCs and smart systems.

While we have seen Artificial Intelligence already been implemented in a host of applications like speech recognition, gaming and autonomous vehicles, we are yet to see it being adapted for software-defined radio or incorporated into an RF hardware solution. This is why Deepwave Digital, a hardware and software solutions provider, has taken it upon themselves to come up with a Software-defined Radio with deep-learning muscle, called AIR-T.

AIR-T is a board that combines both high-end hardware and RF deep learning for SDR. AIR-T is equipped with three signal processors: a NVIDIA Jetson TX2 for signal processing and deep learning applications, a Xilinx Artix 7 FPGA for real-time operations and integrated CPUs for I/O, hardware, and software applications. The system also has a dual-channel MIMO transceiver with two 100MHz RX channels and two 250MHz TX channels, along with a rich set of connecting options including GPS Sync via 1 PPS and 10MHz, Ethernet, Wi-Fi, Bluetooth, HDMI, USB 2.0/3.0, SATA, and high-speed digital I/O.

“This versatile system can function as a highly parallel SDR, data recorder, or inference engine for deep learning algorithms. The embedded GPU allows for SDR applications to process bandwidths greater than 200 MHz in real-time.” says Deepwave. “The AIR-T also uses zero copy memory access to overcome the data transfer overhead typically associated with GPU processing.”

Features and Specifications include:

  • Dual Channel MIMO Transceiver with:
    • 300 MHz to 6 GHz
    • A pair of 100 MHz bandwidth Rx channels, and,
    • Two 250 MHz bandwidth Tx channels
  • Xilinx Artix 7 FPGA
  • NVIDIA Jetson TX2
  • Quad-core ARM Cortex-A57 CPU
  • Dual-core NVIDIA Denver2 CPU
  • 256-core NVIDIA Pascal GPU
  • 8 GB of memory
  • USB 3.0, USB 2.0/3.0,
  • 1x SATA port
  • High-speed digital GPIO/UART
  • 1 Gbps Ethernet
  • WiFi
  • Bluetooth
  • GPS Sync via 1 PPS and 10 MHz
  • Dual Power Mode: 22/14 W
  • Dimensions: 17.0 cm x 17.0 cm x 3.5 cm
  • Weight: 0.35 kg
  • Operating System: Ubuntu 16.04

AIR-T is mentioned to be Ubuntu 16.04 along with a custom GNURadio block, with an option of programming the GPU using Python or C++.

The AI radio transceiver is currently available through the crowdfunding campaign that was launched on crowd supply with a USD 20,000 funding goal. The campaign was funded about three years ago and has so far raised over 900% of its goal which is more than USD 193,000.

More information on the AIR-T can be found on the project’s page on crowd supply.

About Emmanuel Odunlade

Hardware Design Engineer | #IoT Consultant |All things #ML | Entrepreneur | Serial Writer | Passionate about Innovation and technology as tools for solving problems in developing countries. Spare time is spent around writing and advocacy for the growth of the Maker/DIY Culture in Africa.

view all posts by emmanuel
Notify of

1 Comment
Inline Feedbacks
View all comments

The content of your blog is exactly what I needed, I like your blog, I sincerely hope that your blog a rapid increase in traffic density, which help promote your blog and we hope that your blog is being updated.

Subscribe to Blog via Email

After subscribing you can choose how often you will receive our updates:

Join 97,601 other subscribers