Analog rf computing enables edge ai with minimal impact on data rates
Analog RF Computing for Integrated Data and AI Flows Over Wireless Networks
Information TheoryEmerging TechnologiesMachine Learning
Summary
Many small devices don't have enough power or memory to run AI programs themselves. The paper shows a way to use radio waves to send AI program parts from a base station to devices, letting the devices do AI calculations with very little power. The challenge is doing this while still sending normal data to users. The authors studied how mixing AI and data signals affects overall data speed and found it reduces data speed by less than 5%. This means networks can offer AI services alongside regular data without much interference.
What this means in practice
- •For telecom network operators: Offer AI inference services to connected edge devices using analog RF computing without significantly reducing data throughput.$Commercial implications: Enables telecom operators to sell integrated AI and data services over existing wireless networks with minimal performance trade-offs.
- •For wireless system designers: Design joint beamforming solutions that support simultaneous AI and data flows while respecting power, accuracy, and hardware constraints.
Authors
Wentao Yu, Vincent W. S. Wong
Abstract
Many resource-constrained edge devices require artificial intelligence (AI) services but are not powerful enough to store or execute neural networks (NNs). Analog radio frequency (RF) computing addresses this limitation by letting a base station (BS) multicast a waveform encoding NN weights to different clients. Each client then multiplies the received weight-encoded waveform with its locally generated input-encoded waveform through a passive mixer in its RF frontend, thereby realizing the matrix-vector multiplications (MVMs) needed for AI inference with an extremely low power and memory requirement. This can enable telecom operators to provide AI inference services to the edge devices connected to their networks. In practice, however, this multicast AI flow must share the spectrum, power, and spatial resources with other data flows. This paper considers a multi-user multiple-input multiple-output (MU-MIMO) downlink scenario and studies the resulting relative aggregate data rate reduction from integrating the multicast AI flow with the unicast data flows. We formulate a joint beamforming problem for the data and AI flows to maximize the aggregate data rate subject to computing accuracy, client energy, total power, and hardware-related constraints. We develop a weighted minimum mean-square error (WMMSE)-based algorithm to solve the problem. Numerical results based on 3rd Generation Partnership Project (3GPP) channels show that the relative aggregate data rate reduction from integrating an AI flow with data flows is below 5% in representative settings. This indicates that telecom operators can use analog RF computing to provide edge AI services while providing data communication services simultaneously.