Approximate Proximal Operators for Analog Compressed Sensing Using PN-junction Diode

2025-10-13

AI Summary

This work addresses the hardware implementation challenge of sparse signal reconstruction in analog compressive sensing. We propose an analog proximal operator implementation leveraging the forward voltage–current nonlinearity of PN-junction diodes. By modeling diode conduction characteristics, we construct, for the first time, hardware-embeddable approximate proximal operators for both ℓ₁ and minimax concave penalty (MCP) regularization functions, integrated directly into an analog proximal gradient descent framework. The design eliminates digital quantization and iterative control logic, significantly reducing power consumption and latency. Circuit-level simulations demonstrate that the proposed analog reconstructor maintains high accuracy and robustness under device noise and process variations, achieving reconstruction performance close to that of ideal digital implementations. This work establishes a novel, monolithically integrable, low-overhead paradigm for analog-domain sparse optimization, advancing the paradigm shift of compressive sensing from digital to analog hardware.

Technology Category

• Search and Optimization: Non-convex Optimization
• Machine Learning: Hardware-aware ML
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Security and Privacy: Large-scale security measurements

Abstract

In order to realize analog compressed sensing, the paper considers approximate proximal operators of the $ell_1$ and minimax concave penalty (MCP) regularization functions. Specifically, we propose to realize the approximate functions by an electric analog circuit using forward voltage-current (V-I) characteristics of the PN-junction diodes. To confirm the validity of the proposed approach, we employ the proposed approximate proximal operators for the $ell_1$ and MCP regularization functions in compressed sensing with the proximal gradient method. The sparse reconstruction performance of the algorithms using the proposed approximate proximal operators is demonstrated via computer simulations taking into account the impact of additive noise introduced by analog devices.

Problem

• Realizing analog compressed sensing using PN-junction diodes
• Approximating proximal operators for L1 and MCP regularization
• Validating sparse reconstruction via simulations with analog noise

Innovation

• Analog circuit using PN-junction diodes
• Approximate proximal operators for regularization
• Sparse reconstruction with proximal gradient method