AI And RF: Regulations Coming Soon?
By John Oncea, Editor
AI in RF enhances performance but lacks comprehensive regulation. The U.S. has federal orders and state laws, while global efforts emerge. The industry must stay informed as regulations evolve.
Artificial intelligence (AI) is becoming commonplace in almost every industry, including RF. Its ability to solve complex problems, mine and analyze data, and optimize various aspects of RF systems is changing people’s attitude toward AI from being a luxury to a must-have.
Or, as Philip Green, chairperson of now-defunct Arcadia, “Good, bad or indifferent, if you are not investing in new technology, you are going to be left behind.”
Of course, Green is known as Britain’s most hated businessman, so maybe we should just move on. Then again, for his 50th birthday, he chartered an Airbus A200 and flew 200 guests to a hotel in Cyprus for a three-day toga party featuring performances by Tom Jones and Rod Stewart so he can’t be all bad.
But I digress.
AI, as it stands right now, is kind of like the wild, Wild West: a situation in which disorderly behavior prevails, especially due to a lack of regulatory oversight and an inadequate legal system. That’s not to say efforts aren’t being made to reel in AI, but very few regulations are currently on the books and that could spell trouble.
Here, we take a look at attempts to regulate AI, but first, let’s look at some of the ways RF is using this rapidly developing technology.
How AI Is Improving The Industry
First, AI isn’t coming for your job. Rather, it is going to enhance your performance by automating tasks and reducing some complexities. So, think of AI as a tool that can assist, not replace.
And the odds are, you’re already using AI in some way. Take Miller MMIC, for example. The company introduced RapidRF recently, an AI-driven solution that allows manufacturers to automate their design of RFICs for next-generation wireless communication while accelerating innovation for a smarter future in terms of speed of manufacturing, cost, and productivity.
Along with eliminating barriers that prohibit a fully automated development of RFICs, Miller MMIC expects the integration of AI with its platform will cut down up to 90% of development time for RFIC/MMIC design, as well as reduce cost in development by up to 50%.
More generally, AI techniques like machine learning and deep learning are being used to optimize the performance of RF systems, according to the Defense Advanced Research Projects Agency. This includes improving the sensitivity and performance of radios, reducing power consumption, and processing requirements, enabling real-time adaptation to dynamic environments, and optimizing control parameters like antennas, channels, bands, and beams.
AI is also enabling the development of cognitive RF systems that can perform situation assessments, make intelligent decisions, and learn and adapt in real-time during missions, according to IEEE. This applies to cognitive radio, cognitive radar, and cognitive electronic warfare applications.
According to The Fast Mode, AI-enabled RF awareness is providing benefits such as:
- More accurate RF-based sensing and positioning.
- Improved spectrum sensing and access.
- Better understanding of the RF environment.
- Enhanced beamforming and beam management.
AI is also assisting RF engineers in design tasks by optimizing RF component designs like filters and automating some modeling and optimization processes.
When it comes to network management, RCR Wireless writes AI is enabling smart signaling and interference management, improved mobility management, and enhanced positioning capabilities. The technology also enables applications like presence detection and sleep monitoring, touchless device control, indoor navigation, and Simultaneous Location and Mapping (SLAM).
While AI is transforming many aspects of RF engineering, human expertise is still critical for properly defining problems, interpreting results, and applying AI solutions effectively in real-world scenarios. AI is viewed as a powerful tool that enhances RF engineering rather than replacing human engineers entirely.
It is also, as already noted, mostly unregulated.
Legislating AI
AI is, in all industries, rapidly evolving, here to stay, and mostly unregulated. However, several broad AI regulatory frameworks are starting to emerge, including the European Union’s AI Act. According to Regulatory Focus, this act aims to improve the functioning of the internal market by prohibiting certain AI practices and adding specific requirements for high-risk AI systems. This could impact RF systems using AI, especially in medical devices and in-vitro diagnostics.
Closer to home, the U.S. National Institute of Standards and Technology (NIST) developed an AI Risk Management Framework to help manage risks associated with AI. This voluntary framework can be applied to AI systems in the RF industry.
AI, when applied to the defense industry, is poised to radically change military RF systems, applications, and concepts of operations (CONOPS). According to Military Embedded Systems, AI enables the learning of entirely new systems by processing sample data, potentially providing greater sensitivity, better performance, and reduced power consumption.
The complexity of wireless design has resulted in difficult trade-offs, which AI can help optimize, and deep learning techniques are being applied to address design complexities in RF systems, moving away from hand-engineered solutions.
While specific regulations are still developing, Research Features writes consideration must be paid to:
- Data Protection: AI’s reliance on data means RF systems using AI must comply with data protection laws.
- Safety and Risk Management: Highly regulated industries like transportation and healthcare, which often use RF technology, may face stricter AI oversight.
- Liability: There's an ongoing discussion about how to apportion liability for AI systems, which could affect RF industry players.
- Ethics and Trustworthiness: Organizations should consider the ethical implications and trustworthiness of AI in RF systems, aligning with frameworks like the WHO Guidance on Ethics & Governance of AI for Health.
Laws Currently On The Books
The U.S. does not have comprehensive federal legislation specifically regulating AI. However, there are several existing and proposed regulatory efforts at both the federal and state levels.
At a federal level, AI is governed by a mix of federal agencies, state governments, industry self-regulation, and the courts, according to Morgan Lewis. Some key developments include:
- Executive Order 14410: Issued in late 2023, this order directs federal agencies to develop AI guidance and regulations. Agencies like the Departments of State, Agriculture, Commerce, Education, Energy, and Justice have appointed AI leads in response.
- Export Controls: While AI is not broadly controlled as a category, some AI-related components fall under existing export control regulations.
- Executive Order 14117: Issued in February 2024, this order aims to prevent access to sensitive personal data by countries of concern, partly due to AI-related risks.
- Agency Actions: Over the coming months, regulations are expected to affect government procurement, technology development, and ethical use requirements for AI under specific agency authorities.
A sampling of state-level initiatives, courtesy of The Council of State Governments, include:
- Seventeen states have enacted 29 bills focused on regulating the design, development, and use of AI (as of June 2024).
- These state laws primarily address data privacy and accountability concerns.
- California, Colorado, and Virginia have been leaders in establishing regulatory frameworks for AI systems.
- Some states have created task forces or advisory councils to study AI impacts and recommend policies.
What To Expect
As AI becomes more prevalent, we can expect industry-specific regulations to emerge, particularly for high-risk applications. There will be an increased focus on regulatory compliance using AI itself, requiring expertise in both computer science and law.
In addition, the potential for the development of international standards for AI in autonomous vehicles and other RF-dependent systems exists, as does greater scrutiny of AI’s impact on RF spectrum management and allocation.
The most probable outcome, according to the Center for Strategic International Studies, is a bottom-up patchwork of executive branch actions, rather than a broad national AI law. This may include domain-specific agency actions in areas like healthcare, financial services, housing, workforce, and child safety.
In addition, there may be multiple executive orders addressing various aspects of AI, increased government spending on AI research, especially in defense and intelligence, potential trade friction with Europe over differing regulatory approaches, and continued private sector “responsible AI” initiatives.
The RF industry needs to stay informed about evolving AI regulations and proactively implement risk management strategies to ensure compliance and build trust in their AI-enabled RF systems. While this regulatory landscape is still evolving, companies operating in the RF industry – and any industry, for that matter – should stay informed about future developments to ensure compliance and manage AI-associated risks.