AI is popping up in conversations everywhere, sometimes centering on musings of how the technology might develop and what it means for the average person – but getting down into the nitty gritty of how AI will affect wireless communications more specifically is another matter. To delve deeper, IoT Insider spoke to Houman Zarrinkoub, Principal Product Manager at MathWorks.
Zarrinkoub demonstrated his particular belief in the adoption of AI-native wireless systems as he said, “Engineers tasked with designing modern wireless systems have realised that integrating AI is no longer optional; it is essential.”
Why AI? And what does an AI-native wireless system look like? Luckily, Zarrinkoub has the answers: “An AI-native wireless system inherently incorporates AI algorithms into its operational framework,” he explained. The reasons for doing this include, “better coverage, higher capacity and reliable robustness.”
Because of AI’s capacity to learn from its environment through the data it is trained on, AI-native wireless systems are designed to learn and adapt accordingly. “This approach differs significantly from traditional designs based on more rigid, predefined models that have scalability limitations and often require costly, time-consuming signal processing resources,” said Zarrinkoub.
Echoing a point made by Rachel Johnson, Principal Product Manager at MathWorks in an IoT Unplugged podcast episode on integrating AI where she stressed the importance of collecting data properly, Zarrinkoub said: “Engineers designing AI-native systems need large, real-world measured data sets. Most of this data is sourced from physical prototypes or by measuring real-world signals. However, most engineers use digital twins … to augment data to train AI-native systems.”
Designing and developing this system
“This is a complex process that involves creating a design workflow that includes gathering data, training and testing the model, and implementing and integrating the model into the wireless system,” he summarised.
Data collection is “the first step”, by using over-the-air (OTA) signals or synthesising data drawn from a digital twin. This data is used to train the AI model, to make sure it’s capable of real-world performance.
“An AI model is only useful when it is implemented as part of a real-world system,” Zarrinkoub explained, in reference to the implementation and integration part of the process. “The first step involves scaling and resource assessment, this involves evaluating the processing power, memory requirements and data throughput needed for the AI models to operate efficiently.”
Integrating AI is not necessarily an easy matter, and in engineering design, “optimising one metric often compromises another … For instance, increasing network throughput may raise power consumption and latency, necessitating trade-offs to maintain energy efficiency.
“Engineers can employ modelling and simulation to explore various scenarios … One solution is to simulate the integrated system before full-scale deployment to ensure AI components will interoperate properly with legacy systems.”
For integrating and implementing such a system, this involves scaling and resource assessment; using automatic code generation and integrating AI models into the wireless systems, Zarrinkoub explained. “Engineers must verify interoperability by analysing the end-to-end performance before full-scale integration.”
Wireless communications of the future
As already detailed by Zarrinkoub, the process involved in designing, developing, integrating and, finally, implementing a system is a complex process deserving of lots of time and attention.
“The wireless industry is at a critical juncture,” emphasised Zarrinkoub. “With the upcoming rollout of 5G Advanced and 6G standards, the next generation of wireless systems will deploy more AI-native technologies.”
Mobile communications standards organisation 3GPP has been “vocal” about the role AI is anticipated to play in upcoming 5G and 6G standards, said Zarrinkoub. “They propose AI’s functionality for enhanced positioning, beam management and Channel State Information (CSI) feedback. The Wireless Broadband Alliance (WBA) also touts AI for its ability to help wireless engineers in indoor positioning and beam management.”
There’s plenty of other editorial on our sister site, Electronic Specifier! Or you can always join in the conversation by commenting below or visiting our LinkedIn page.
