RadMamba: Efficient Human Activity Recognition through Radar-based Micro-Doppler-Oriented Mamba State-Space Model

2025-04-16

AI Summary

To address the high parameter count and computational overhead of radar micro-Doppler signal-based human activity recognition (HAR) models in edge deployment, this paper introduces the first adaptation of the Mamba state-space model for radar time-series modeling. We propose a lightweight, efficient paradigm through customized design of micro-Doppler time-frequency representations and a tailored Mamba architecture, achieving an unprecedented balance between ultra-low parameter count (as few as 6.7K parameters) and strong temporal modeling capability. Evaluated on three benchmark radar datasets—DIAT, CI4R, and UoG2020—the proposed method attains accuracies of 99.8%, 92.0%, and state-of-the-art performance, respectively, while reducing model size to only 0.25%–10% of competing approaches. This breakthrough significantly alleviates the constraints on memory, latency, and energy consumption inherent to edge intelligence deployment.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Planning, Routing, and Scheduling: Activity and Plan Recognition
• Intelligent Robots: State Estimation

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Radar-based HAR has emerged as a promising alternative to conventional monitoring approaches, such as wearable devices and camera-based systems, due to its unique privacy preservation and robustness advantages. However, existing solutions based on convolutional and recurrent neural networks, although effective, are computationally demanding during deployment. This limits their applicability in scenarios with constrained resources or those requiring multiple sensors. Advanced architectures, such as ViT and SSM architectures, offer improved modeling capabilities and have made efforts toward lightweight designs. However, their computational complexity remains relatively high. To leverage the strengths of transformer architectures while simultaneously enhancing accuracy and reducing computational complexity, this paper introduces RadMamba, a parameter-efficient, radar micro-Doppler-oriented Mamba SSM specifically tailored for radar-based HAR. Across three diverse datasets, RadMamba matches the top-performing previous model's 99.8% classification accuracy on Dataset DIAT with only 1/400 of its parameters and equals the leading models' 92.0% accuracy on Dataset CI4R with merely 1/10 of their parameters. In scenarios with continuous sequences of actions evaluated on Dataset UoG2020, RadMamba surpasses other models with significantly higher parameter counts by at least 3%, achieving this with only 6.7k parameters. Our code is available at: https://github.com/lab-emi/AIRHAR.

Problem

• Radar-based HAR needs efficient, lightweight computational models
• Existing CNN and RNN solutions are resource-intensive for deployment
• Current advanced architectures like ViT and SSM still have high complexity

Innovation

• Radar-based Mamba SSM for HAR
• Parameter-efficient micro-Doppler modeling
• High accuracy with minimal parameters