Artificial Intelligence Radio Transceiver (AIR-T)
The first radio frequency system designed for deep learning
An Overview of Deep Learning with the AIR-T

If you are considering the AIR-T for your wireless machine learning, digital signal processing, or high-performance computing application, you will need to know how to program it. The AIR-T is designed to reduce the number of and effort of engineers required to create an intelligent wireless system. Programming the AIR-T is simple and streamlined.

Knowledge Base

Ideally you will have knowledge of how your AI algorithm works and be able to program in a language like Python. You won’t, however, have to worry about buffering, threading, talking to registers, etc. For power users or those wanting to get every last percentage of performance out of the device, various drivers and hardware abstraction APIs are provided so you can write your own custom application.

If you are unfamiliar with deep learning algorithms, the process is typically broken up into two stages: training and inference. Rather than rewriting the great online references covering these processes, here is an excellent blog that discusses training and inference and here is a great introduction to how neural networks operate.

+ Read More

“The AIR-T board will allow researchers and developers to experiment and build artificial intelligence and high performance compute solutions at the RF edge.”
-Arjun Dutt, Director of the Inception program at NVIDIA

A familiar tool

A familiar tool to anyone working in the wireless domain, GNU Radio allows signal processing experts to tie together blocks of functionality using an intuitive GUI. Many of the “in the weeds” details regarding the software implementation are well abstracted so the user can focus on the algorithm instead. More information about GNU Radio may be found here​.

Once an algorithm has been optimized (or a pre-trained algorithm has been downloaded by a 3rd party), the user will reference it in Deepwave’s GR-WAVELEARNER software that provides a TensorRT Inference block for GNU Radio Companion (GRC), as shown to the left.

While all of this may seem new to you, we assure you that the examples provided with the AIR-T will be more than enough to get you up and running quickly.

With the power of deep learning incorporated into wireless technology, the number of engineering hours required to build complex RF system is significantly reduced. For example, the Deep Learning Signal Classifier shown below took approximately 6 hours to train on a single NVIDIA GP100 GPU, resulting in near perfect signal classification. For more information on this classifier, see the Deepwave talk at the GPU Technology Conference.

Schematic & Drawing