AI Testing AI: The Future Of 6G Test

Conventional pass-fail limits based on parametric measurements will no longer be sufficient.


The impending arrival of 6G technology promises to revolutionize the way we connect and communicate. With expected data rates of up to 100 times faster than 5G, 6G is poised to enable unprecedented applications, from augmented reality (AR) and virtual reality (VR) to real-time remote surgery and autonomous vehicles with ubiquitous connectivity. A significant facet of 6G’s potential lies in the integration of artificial intelligence (AI) directly into the physical layer of devices. While this innovation holds immense promise, it also brings forth a myriad of test implications that must be carefully addressed. 6G technology aims to deliver ultra-low latency, ultra-high reliability, vastly improved data rates, and massive connectivity. To achieve this, AI is expected to be embedded directly into many parts of the network, including the physical layer of 6G transmitters and receivers. This integration will allow devices to make local decisions, optimize resources, and adapt to changing environments without relying solely on centralized processing.

Challenges in testing AI-integrated 6G devices

Testing AI is different than testing known-function components, as it involves testing how the AI will act when deployed in the infinitely variable conditions of the real world. For instance, AI can correct for degraded performance in components such as power amplifiers, potentially masking failures. AI-enabled devices are designed to adapt to unpredictable real-world scenarios and will operate in dynamic environments with varying signal strengths, interference levels, and user densities. Therefore, AI algorithms must be trained to optimize performance under a sufficiently large set of conditions to ensure reliable performance. Testing will need to include conditions sufficiently different from those in the training set to evaluate performance in changing and unpredictable real-world scenarios.

Ensuring that AI-integrated devices from different manufacturers and devices that have different “learning” can seamlessly communicate and collaborate is crucial. Each device may optimize in a slightly different way, making it difficult to distinguish a device that performs correctly from one that fails to do so.  Because of this, conventional pass-fail limits based on parametric measurements will no longer be sufficient. The hardware and software in the test environment must learn to “recognize” good and bad devices much like a human can recognize the look and behavior of a healthy dog from that of sick one. AI-enabled testers may be needed to test AI-enabled devices.

Addressing the opportunity of testing AI-integrated 6G devices

As mentioned above, testing of 6G will become far more complicated. In fact, there are two keys to validating the trustworthiness of embedded AI:

Data: Having an ample supply of high-quality training and validation data is of utmost importance. In AI-driven systems, data serves as the algorithm itself – it’s the intellectual property and the most valuable asset. There can be no algorithm or design in the absence of a substantial amount of high-quality scenario-based training and validation data. Consequently, data has become significantly more critical and sensitive.  Furthermore, to be able to leverage data effectively, there is a big need for new ways to organize, manage, label, and pre-process that data to get results.

System-level test: Secondly, a test system capable of accurately replicating the variety of conditions that the Device Under Test (DUT) will encounter during its deployment becomes essential. As wireless systems begin to rely more and more on AI/ML algorithms, these systems will need to also perform under the infinite number of scenarios they could experience when deployed. Instead of the conventional stimulus-response testing approach, the question at hand becomes: What are the sets of scenarios we need to incorporate into our test? Developing these intricate systems demands a closer integration between the design and testing phases.

In contrast to the traditional model, which involves building a physical prototype, testing it, and iterating on the design until a satisfactory product is achieved for customer release, organizations are progressively embracing a different approach for these advanced systems.  Organizations are increasingly doing the design and testing of these more advanced systems in a software-based virtual world, where cost and speed are improved. This entails employing software-in-the-loop, model-in-the-loop, and hardware-in-the-loop testing methodologies based on real-world scenarios.

The bottom line: Testing 6G devices with integrated AI will require organizations to re-think their approach to validation

The integration of AI into the physical layer of 6G devices represents a monumental leap in wireless communication technology. It promises to unlock new possibilities and transform industries. However, with great innovation comes significant new challenges, particularly in the realm of testing. The industry must proactively address the test implications of AI integration to ensure that 6G devices deliver on their promises of ultra-low latency, reliability, energy and spectrum efficiency, massive connectivity, and blistering data rates.

As we move closer to the era of 6G, collaboration, standardization, and the development of AI-powered testing solutions will be critical. Establishing industry-wide standards for testing 6G devices with embedded AI is paramount. These standards should encompass AI algorithms, AI training, and AI-based assessment of test results as well as performance metrics and testing methodologies to ensure consistency and reliability. Collaboration among device manufacturers, network operators, and AI experts is essential to address the challenges of AI integration. Sharing best practices and insights can lead to more robust testing methodologies as well as ensuring interoperability with optimum performance.

By addressing these challenges head-on, we can help ensure that AI-integrated 6G devices operate flawlessly, enabling a future where instantaneous communication and transformative applications become a reality. The road ahead is undoubtedly challenging and having a robust and adaptable test strategy is key. To learn more about how to best address the challenges with testing 6G devices, contact NI.

Leave a Reply

(Note: This name will be displayed publicly)