Skip to content

AirStack 0.5.0

 

14 September 2021

 

The Deepwave Digital team is happy to announce the release of AirStack 0.5. This is a major release that includes significant performance enhancements, upgrades to critical APIs, and the incorporation of a firmware recovery image to streamline updates.

New Features

  • Continuously Variable (CV) Sample Rates - We have upgraded the AIR-T firmware interpolation and decimation filters to produce a continuously variable sample rate. This feature is available on the AIR-T models with the Xilinx Artix 7 200T model FPGA, i.e., the AIR7201 and the AIR8201. The filters are continuously tunable between 3.9 and 125 MSPS.

  • Recovery Firmware - AirStack now supports multiple firmware images for the onboard FPGA. This feature allows for a recovery image to be installed along side the AirStack firmware as a measure of protection against irrecoverable board states during image loading. Should the firmware installation abruptly exit before finishing, you can now rest assured that the unit can be easily recovered.

  • Improved Conda Support - Full radio functionality, including support for GNU Radio, is now available from within our Conda environments. If using TensorRT or PyTorch you will still be limited to Python 3.6. This is a restriction imposed by JetPack. For all other software, newer versions of Python are now available using Conda (3.6, 3.7, 3.8, and 3.9)

  • Conda Support for GNU Radio - We have enabled the ability to use GNU Radio 3.9.2 in a Conda environment. This provides a significant improvement and allows customers to use newer versions of both Python and deep learning tools such as PyTorch. Conda is now required when using GNU Radio and is also available for all other uses.

  • Soapy SDR Upgrade Soapy SDR has been upgraded to version 0.8.

  • NVIDIA JetPack 4.6 - AirStack 0.5.0 upgrades the base operating system from JetPack 4.4.2 to JetPack 4.6 (L4T 32.6.1). The operating system is Ubuntu 18.04 with Linux kernel 4.9 and also includes significant updates to GPU acceleration libraries such as TensorRT, CUDA, and cuDNN.

  • CUDA Update NVIDIA's CUDA has been upgraded to 10.2. See the release notes for further details.

  • cuDNN Update CUDA Deep Neural Network (cuDNN) is the industry standard software primitives for training an executing deep neural networks on GPUs. In this release, it has been upgraded to cuDNN 8.2.1.

  • TensorRT The AIR-T uses NVIDIA's TensorRT for optimized neural network inference. For this release, TensorRT has been upgraded to 8.0.1, significantly improving the number of neural network layers supported for inference. Additionally, support for workflows based on ONNX models is much improved in this version of TensorRT.

  • AIR-T Edge Series Support - AirStack 0.5 is the first public release that supports the new AIR-T Edge Series.

  • Conda Update - We have updated to Anaconda 4.10.3

  • Improved Transmit Streaming - AIR7201 and AIR8201 have larger transmit buffers in the FPGA firmware to enable continuous streaming at higher sample rates.

New Source Code Examples

  • Creating a GNU Radio Conda Environment - We have created a new tutorial to walk users through the steps of creating a GNU Radio Conda environment on the AIR-T.

  • Deep Learning Inference - An AI Inference on the AIR-T Tutorial has been created with the goal of receiving signals from the AIR-T's front end, preprocess the data, perform neural network inference, and then compare the results against truth data.

Other Notes

  • Python 2 support has been removed

  • GR-WAVELEARNER has been updated and is now supported for GNU Radio 3.9.

  • GNU Radio 3.7 is no longer pre-installed with AirStack and is not supported by conda. We recommend all customers transition to using the newer versions of GNU Radio provided via Conda, as there have been many upstream bugs fixed and performance/usability improvements made.

Bug Fixes

  • Improve clock initialization during startup
  • Restart data streaming on errors for improved GNU Radio support
  • Many Linux upstream security updates

Download

The AIR-T upgraded software and firmware are available for customers with active support plans to download in the Developer Portal.

Please note that upgrading to AirStack 0.5.0 from previous versions of AirStack requires a re-flash of the operating system in addition to the usual firmware update. Please see the installation procedure to apply the software update to your AIR-T, followed by the firmware update procedure.

Last update: August 3, 2023