aiqbench analysis
This research introduces analog radio frequency (RF) computing as a specialized physical layer architecture designed to offload matrix-vector multiplications from edge devices to wireless base stations. By leveraging passive mixers at the client side, the proposed framework achieves ultra-low energy consumption for neural network inference while maintaining accuracy through joint beamforming and scaling algorithms. The study demonstrates that this approach can reduce client-side power usage by nearly two orders of magnitude compared to conventional digital computing methods.
Analog RF Computing: A New Paradigm for Energy-Efficient Edge AI Over MU-MIMO Systems
arXiv:2605.14331v2 Announce Type: replace-cross Abstract: Modern edge devices increasingly rely on neural networks for intelligent applications. However, conventional digital computing-based edge inference requires substantial memory and energy consumption. In analog radio frequency (RF) computing, a base station (BS) encodes the weights of the neural networks and broadcasts the RF waveforms to the clients. Each client reuses its passive mixer to multiply the received weight-encoded waveform with a locally generated input-encoded waveform. This enables wireless receivers to perform the matrix-vector multiplications (MVMs) that account for most of the computation burden in edge inference with ultra-low energy consumption. Unlike conventional downlink transmissions which are optimized for communications, analog RF computing requires a computing-centric physical layer that controls both the analog MVM accuracy and the energy consumption for inference. Motivated by this, in this paper, we propose a physical layer design framework for analog RF computing in MU-MIMO wireless systems. We derive tractable models for computing accuracy and energy consumption for inference, formulate a joint BS beamforming and client-side scaling problem subject to computing accuracy, transmit power, and hardware constraints, and develop a low-complexity algorithm to solve the non-convex problem. The proposed design provides client- and layer-specific accuracy control for both uniform- and mixed-precision inference. Simulations under 3GPP specifications show that analog RF computing can significantly reduce client-side energy consumption by nearly two orders of magnitude compared to digital computing, while mixed-precision inference requires even lower energy consumption than uniform-precision inference. Overall, these results establish analog RF computing over wireless networks as a promising paradigm for energy-efficient edge inference.
Abstract:Modern edge devices increasingly rely on neural networks for intelligent applications. However, conventional digital computing-based edge inference requires substantial memory and energy consumption. In analog radio frequency (RF) computing, a base station (BS) encodes the weights of the neural networks and broadcasts the RF waveforms to the clients. Each client reuses its passive mixer to multiply the received weight-encoded waveform with a locally generated input-encoded waveform. This enables wireless receivers to perform the matrix-vector multiplications (MVMs) that account for most of the computation burden in edge inference with ultra-low energy consumption. Unlike conventional downlink transmissions which are optimized for communications, analog RF computing requires a computing-centric physical layer that controls both the analog MVM accuracy and the energy consumption for inference. Motivated by this, in this paper, we propose a physical layer design framework for analog RF computing in MU-MIMO wireless systems. We derive tractable models for computing accuracy and energy consumption for inference, formulate a joint BS beamforming and client-side scaling problem subject to computing accuracy, transmit power, and hardware constraints, and develop a low-complexity algorithm to solve the non-convex problem. The proposed design provides client- and layer-specific accuracy control for both uniform- and mixed-precision inference. Simulations under 3GPP specifications show that analog RF computing can significantly reduce client-side energy consumption by nearly two orders of magnitude compared to digital computing, while mixed-precision inference requires even lower energy consumption than uniform-precision inference. Overall, these results establish analog RF computing over wireless networks as a promising paradigm for energy-efficient edge inference.
16 pages, 10 figures, 2 tables. This paper proposes analog RF computing as a new paradigm for energy-efficient edge inference over wireless networks and studies the corresponding physical layer design framework
Signal Processing (eess.SP); Artificial Intelligence (cs.AI); Emerging Technologies (cs.ET); Information Theory (cs.IT); Machine Learning (cs.LG)
arXiv:2605.14331 [eess.SP]
(or arXiv:2605.14331v2 [eess.SP] for this version)
Full text under the publisher's open licence · 305 words
Related coverage
Quantum Max d-Cut via qudit swap operators
Quantum Journal · Sep 3, 2026
How Photonic Chips Differ From Electronic Chips
Quantum Zeitgeist · Sep 19, 2026
Lightweight Deep Learning-Based Channel Estimation for RIS-Aided Extremely Large-Scale MIMO Systems on Resource-Limited Edge Devices
arXiv · Aug 19, 2026
[AINews] Opus 5.5 is good at explainer videos
Latent Space · Sep 29, 2026