AI Native Air Interface
- Daniel Ezekiel
- Jun 18
- 4 min read
In a previous article, I discussed the role of AI in telecommunications, focusing on areas like network planning, business intelligence, and self-organizing networks. Now, as we look toward 6G, AI's potential becomes even more transformative—particularly in the realm of an "AI Native Air Interface." This next-generation technology is poised to redefine how wireless systems operate, paving the way for dynamic spectrum management, adaptive signal processing, and intelligent resource allocation.
To set the context, the following table gives a gist of benefits of AI usage across the ‘Protocol Stack’
The 3GPP standardization deals with the following (as a part of REL 18, 5G Advanced), and paves the way forward for a AI Native Air Interface that is 6G Centric.
The 6 key technologies for 6G :
AI/ML-Air-Interface
New Spectrum Technologies
Network as a Sensor
RAN-Core Convergence & Specialisation
Extreme Connectivity
Security and Trust
AI offers significant potential for enhancing the "Air Interface," which traditionally relies on stochastic channel modelling. With the surge in devices, high-speed mobility, and increased spectrum usage, AI can better handle the complexities of evolving radio environments. An AI Native Air Interface transforms traditional Transceiver design, shifting from fixed protocol stacks to dynamic, AI-driven architectures that adapt more flexibly to changing channel conditions. This is especially valuable in urban environments with dense, high-mobility devices and the growing number of connected devices.
What does a AI Native Air Interface look like ?
This involves the following key aspects including dynamic spectrum management, adaptive modulation and coding, intelligent signal processing, and radio resource management (massive MIMO and Intelligent Beam Forming). From an implementation standpoint this implies all the implementation can be done within the Radio Transceiver.
Evolution of Network and Devices Chipset Architecture
‘Communications and Computation’ have always gone hand in hand for the past few decades. Both from communication systems requiring compute with specific needs, and from the closer integration of compute and communications systems.
From 2G to 5G, the Chipset topology/architecture at a devices level and network level have comprised of the following
Applications processor (that requires high performing General Compute). This used to be primarily x86 based systems, with ARM based systems in the nascent level, however has a long way to go in order be become a viable alternative. Some forms of RISC-V based systems, are making an initial foray, however they need many more years to mature
Baseband processor: That catered to most protocol layers activities, including parts of Layer 1 (usually beefed up by additional accelerators in case of real time critical operations). Initially this used to be ASIC, and DSP based world, that has increasing become a mix of ASICs, ARM and MIPs based systems with HW Accelerators.
RF ASIC: That initially handled RF centric operations viz., Mux/DeMuxing, modulation and demodulation, CRC, A2D and D2A functionaries and interface to the RF Front end related operations, and were most Analog, over the past couple of decades the RF ASICs have become more Digital in nature. The interfaces between RF ASIC and Baseband processors are completely digital in nature these days.
Transceiver changes with AI Native Air Interface
The rise of AI in modem technology, especially through the ‘AI Native Air Interface,’ is reshaping network architecture by integrating AI deeper into the air interface layer. This shift promises significant gains in energy efficiency, cost savings, and sustainability through component reuse and upgrade. Such advancements will help leading innovation-driven economies maintain an edge over low-cost markets by optimizing both performance and resource efficiency.
Traditional and AI Native Air Interface Implementation changes
In a traditional RF Transceiver architecture -
The traditional Transmitter performs Encoding, Symbol Mapping and Modulation/Waveform generation.
In the roadmap to an AI-Native-Air Interface Transmitter for 6G, each of the above functions are at first supplemented by MLm eventually a single AI/ML block replaces all three functions and AI takes over the functions of encoding, symbol mapping and modulation.
The traditional Receiver does the following functions - Sync, Channel Estimation, Channel Equalisation, Symbol Remapping and Decoding.
In the roadmap to an AI.Native-Air-Interface Receiver for 6G, each of the above functions are supplemented by ML, leading to the next logical phase of Channel Estimation, Equalisation and Symbol Mapping being replaced by an AI/ML block, and eventually the complete Receiver being replaced by a single self learning and functioning AI/ML block.
Summary Ecosystem Players in AI Native Air Interface
Leading telecom vendors like Qualcomm, Marvell, ARM, Huawei, Ericsson, and Nokia are heavily investing in AI Native Air Interface research. Companies are developing AI-driven modems, network architecture, and chipsets that can dynamically adapt to changing channel conditions, spectrum use, and device mobility.
By integrating AI across the air interface, 6G networks will operate more efficiently, reduce latency, and support a massive increase in connected devices—setting the stage for the next generation of wireless communications.
--
👉 If you'd like to learn more, how this applies to your wireless semiconductor business, and how to make optimal semiconductor choices — reach out and book time at : https://lnkd.in/eTk5pQxx
_
