AirStack Core 2.2.0¶

The Deepwave team is pleased to announce the release of AirStack Core 2.2. This is a major release that contains both significant new features and bug fixes. The highlight of this release is an update to all NVIDIA-provided GPU computing libraries.

This release supports the AIR7310 AIR-T hardware. Earlier models should continue to use AirStack 1.0.

New Features¶

• JetPack 6.2 - Jetpack has been upgraded from 6.0 to 6.2. This provides an updated Jetson Linux (Ubuntu 22.04 with kernel 5.15), improved compatibility with Docker 28.x, and refreshed AI libraries including CUDA 12.6, TensorRT 10.3, and cuDNN 9.3 - offering better support for modern AI workloads and development environments. A comprehensive list of these changes can be found here, a few key highlights include:
• CUDA 12.6 - NVIDIA's CUDA has been upgraded to 12.6 (from 12.2).
• TensorRT 10.3 - TensorRT has been upgraded from to 10.3.0 (from 8.6.2). 10.3.0 boosts engine build performance with faster compilation of large matrices, improves quantization fusion, and enhances ONNX model support including better parser compatibility.
• Linux Kernel - Contains all updates up to and including Jetson Linux 36.4.4.

Resolved Issues¶

• USB 3 Support - Resolved issue where USB 3.0+ devices might negotiate at 2.0 speeds.

• Calibration Failure Recovery - In rare cases when synchronizing to an external clock source with a custom master clock frequency, initial radio calibration may not complete successfully. The driver automatically detects and retries calibration as needed, notifying the user with clear informational messages. During this process, this release resolves an issue where certain RF frontend settings—specifically preselector configuration and gain—were not always restored correctly during the retry process. These settings are now reliably preserved, ensuring consistent and correct behavior across all calibration scenarios. We recommend that all users leveraging external clock sources or custom master clock frequencies apply this update to benefit from these reliability enhancements.

• Transmit Underflow Detection - This release improves the accuracy of transmit underflow detection by resolving a firmware issue that could previously lead to undercounting in certain edge cases. The underlying reporting logic has been updated to ensure all relevant underflows are now correctly captured. With this fix, users can reliably detect transmit underflows using any of the following indicators:

  • The transmit channel’s LED will turn off if the radio IC’s sample stream is interrupted.
  • The return code of each writeStream() call in the radio API will accurately reflect underflow conditions.

• System Update - This release enhances system robustness when applying upstream NVIDIA and Ubuntu updates. Previously, minor JetPack patch upgrades installed via apt-get upgrade could, in rare cases, overwrite AirStack Core files. These conflicts have been fully resolved, and users can now safely apply JetPack patch updates and general Ubuntu security upgrades without impacting system stability. Deepwave actively monitors upstream changes to maintain ongoing compatibility with AirStack Core and ensure seamless system maintenance.

• Operating System Status - This release improves system clarity by eliminating misleading warning and error messages that could appear during boot, notifying the user that certain services had failed to start. While these messages had no impact on system functionality, they could cause systemctl status to report a degraded state. The system now boots cleanly, and status reports accurately reflect its fully operational state.

• Previously Fixed Issues - Please reference the release notes from previous AirStack Core releases in this series for additional issues that have been fixed relative to the software preinstalled on your device

Tutorials¶

• Conda and TensorRT tutorials have been updated to reflect the update to JetPack.
• Docker tutorial now shows how to use container base images from third-party distributions/sources, other than NVIDIA.

AirStack Core Compatibility¶

How To Upgrade¶

Customers with an AIR7310 should upgrade their devices by flashing (factory reset) their devices with the new build. Access to the software build is available at the Developer Portal.

After downloading the release, use the included airstack_core_installer GUI to perform the update. Detailed instructions can be found at AirStack Core Installation Guide.