Millimeter-wave radar holds significant potential for dynamic gesture recognition in contactless human-computer interaction, particularly in the Internet of Things and consumer electronics applications. However, a considerable challenge persists in filtering vast amounts of extraneous data from millimeter-wave radar echoes to isolate meaningful gesture features. We present a novel approach based on sparse representation principles to address this. We first generate a range-Doppler map of gestures using a two-dimensional (2-D) fast Fourier transform, then construct a Doppler-Time trajectory from aggregated data across multiple frames. Capitalizing on the intrinsic sparsity in the Doppler-time domain, we employ the orthogonal matching pursuit algorithm to refine a multidimensional feature sequence across time, Doppler, and range dimensions. Central to our approach is a dual-stream Transformer network that explores complex 2-D correlations in feature sequences via multihead self-attention mechanisms. This technique significantly improves gesture feature extraction efficiency and reduces data redundancy. The experimental results show that our model has an average recognition accuracy of 99.17% and a size of 0.17M, which is very suitable for application in embedded devices.

Electrode-based electrocardiogram (ECG) measurement devices often lead to user discomfort and raise privacy concerns. In contrast, millimeter-wave radar offers a non-invasive alternative by enabling real-time heartbeat monitoring without direct physical contact. Nevertheless, radar outputs differ significantly from traditional ECG signals, complicating medical diagnosis. To address this challenge, we propose a method for accurately reconstructing ECG-like signals from millimeter-wave radar data using deep convolutional generative adversarial networks enhanced with an attention mechanism (Attention-DCGAN). Based on the MMECG public dataset, we first apply an improved K-means algorithm to cluster heartbeat data from various heart locations. We then train the Attention-DCGAN, consisting of a generator with a 4-layer deconvolution module and a 1-layer attention module, alongside a discriminator with a 4-layer convolution module, to convert radar-derived heartbeat data into ECG-like signals. Experimental results demonstrate that our method achieves higher reconstruction accuracy than traditional methods, with a root-mean-square error of 0.0329 mV and a Pearson correlation coefficient of 0.9853.

Decoupling multiple acoustic relaxations in multi-component gas mixtures is an effective method to acquire acoustic relaxation characteristic of gas molecules, which can be used to determine gas composition. However, decoupling multiple rotational relaxations of hydrogen still remains challenge. In this paper, we present the decoupling model of hydrogen and decouple the multiple rotational relaxation into the group of single-relaxation ones. Based on this model, the whole rotational relaxation process of hydrogen can be simplified as a few decoupled single rotational relaxation processes. The decoupled results of the effective specific heat, the absorption spectrum and the acoustic velocity dispersion agree with the experimental data, which validates our decoupling model. Furthermore, the proposed decoupling rotational relaxation model can be combined with the existing decoupling vibration model to detect gas mixture. Applications in the detection of gas mixtures including N ₂ , CO ₂ , O ₂ and H ₂ are provided. Compared with existing gas sensing method ignoring rotational relaxation, our detection result of gas mixtures H ₂ - O ₂ - H ₂ O is more convincing since the proposed decoupling model contains significant rotational relaxation of hydrogen. Therefore, the proposed decoupling model not only reveals the mechanism of multiple rotational relaxations of hydrogen, but also provides the new method to detect hydrogen gas mixtures.

Gesture recognition with millimeter-wave radar has broad application prospects in human-computer interaction. However, traditional recognition methods generate overly redundant features and construct large-scale networks, rendering them unsuitable for embedded devices with limited memory. To address this challenge, we propose a sparse-driven dynamic gesture recognition network in millimeter-wave radar, named TCNFormer (Time Convolutional Network and transFormer). Firstly, we employ a two-dimensional fast Fourier transform (2D-FFT) to obtain the range-Doppler maps (RDM). These maps are then processed through incoherent integration of multiple frames to produce Doppler-time maps (DTM). We subsequently use the orthogonal matching pursuit algorithm to achieve a sparse representation of the Doppler-time trajectories and integrate the RDM to extract the range features of gestures, obtaining the multi-dimensional sparse sequences encompassing the range-Doppler-time feature. We then design a TCNFormer network tailored to these multi-dimensional sparse sequences. This network leverages a shallow TCN to learn local features, a Transformer network to capture global features and an adaptive weighting method to fuse these local and global features effectively. Experimental results demonstrate that our network fully exploits the sparse multi-dimensional sequences, achieving a recognition accuracy of 99.17% on a self-built dataset. The parameter size of the network is only 0.13M, significantly outperforming existing state-of-the-art models in relevant metrics, thereby proving its suitability for embedded applications in human-computer interaction.

Reducing energy consumption while maintaining tracking accuracy is a significant challenge in the context of underwater target tracking using wireless sensor networks. The allocation of digitalizing bits for each sensor node is pivotal in influencing both tracking accuracy and energy efficiency. However, traditional optimization methods for bit allocation in sensing quantization are constrained by environmental dependencies, inflexibility, and prolonged computation times. To address these issues, we propose an innovative bit allocation algorithm based on deep reinforcement learning for target tracking in underwater sensor networks. This algorithm focuses on adaptively quantizing and transmitting sensor data to optimize energy usage. Initially, we establish a quantitative relationship between bit allocation, tracking accuracy, and energy consumption within these sensor nodes. Building on this foundation, we develop a bit allocation model to minimize energy consumption while adhering to accuracy constraints. We present a novel Transformer-based deep reinforcement learning network, the Transformer-Double Dueling Deep Q-Network (T-D3QN), to optimize bit allocation strategies. This network facilitates the offline-trained reinforcement learning agent to derive optimal sensor quantization bit allocations in real time. Our experimental results demonstrate that the proposed algorithm effectively achieves a balance between target tracking precision and digital bit consumption, significantly enhancing real-time computational efficiency.