Performance Analysis of Wearable Dual-Band Patch Antenna Based on EBG and SRR Surfaces.

This paper presents the performance comparison of a dual-band conventional antenna with a split-ring resonator (SRR)- and electromagnetic bandgap (EBG)-based dual-band design operating at 2.4 GHz and 5.4 GHz. The compactness and dual-frequency operation in the legacy Wi-Fi range of this design make it highly favorable for wearable sensor network-based Internet of Things (IoT) applications. Considering the current need for wearable antennas, wash cotton (with a relative permittivity of 1.51) is used as a substrate material for both conventional and metamaterial-based antennas. The radiation characteristics of the conventional antenna are compared with the EBG and SRR ground planes-based antennas in terms of return loss, gain, and efficiency. It is found that the SRR-based antenna is more efficient in terms of gain and surface wave suppression as well as more compact in comparison with its two counterparts. The compared results are found to be based on two distinct frequency ranges, namely, 2.4 GHz and 5.4 GHz. The suggested SRR-based antenna exhibits improved performance at 5.4 GHz, with gains of 7.39 dbi, bandwidths of 374 MHz, total efficiencies of 64.7%, and HPBWs of 43.2 degrees. The measurements made in bent condition are 6.22 db, 313 MHz, 52.45%, and 22.3 degrees, respectively. The three considered antennas (conventional, EBG-based, and SRR-based) are designed with a compact size to be well-suited for biomedical sensors, and specific absorption rate (SAR) analysis is performed to ensure user safety. In addition, the performance of the proposed antenna under bending conditions is also considered to present a realistic approach for a practical antenna design.

JLAN: medical code prediction via joint learning attention networks and denoising mechanism.

BACKGROUND: Clinical notes are documents that contain detailed information about the health status of patients. Medical codes generally accompany them. However, the manual diagnosis is costly and error-prone. Moreover, large datasets in clinical diagnosis are susceptible to noise labels because of erroneous manual annotation. Therefore, machine learning has been utilized to perform automatic diagnoses. Previous state-of-the-art (SOTA) models used convolutional neural networks to build document representations for predicting medical codes. However, the clinical notes are usually long-tailed. Moreover, most models fail to deal with the noise during code allocation. Therefore, denoising mechanism and long-tailed classification are the keys to automated coding at scale. RESULTS: In this paper, a new joint learning model is proposed to extend our attention model for predicting medical codes from clinical notes. On the MIMIC-III-50 dataset, our model outperforms all the baselines and SOTA models in all quantitative metrics. On the MIMIC-III-full dataset, our model outperforms in the macro-F1, micro-F1, macro-AUC, and precision at eight compared to the most advanced models. In addition, after introducing the denoising mechanism, the convergence speed of the model becomes faster, and the loss of the model is reduced overall. CONCLUSIONS: The innovations of our model are threefold: firstly, the code-specific representation can be identified by adopted the self-attention mechanism and the label attention mechanism. Secondly, the performance of the long-tailed distributions can be boosted by introducing the joint learning mechanism. Thirdly, the denoising mechanism is suitable for reducing the noise effects in medical code prediction. Finally, we evaluate the effectiveness of our model on the widely-used MIMIC-III datasets and achieve new SOTA results.

Identifying VoIP traffic in VPN tunnel via Flow Spatio-Temporal Features.

The persistent emergence of new network applications, along with encrypted network communication, has make traffic analysis become a challenging issue in network management and cyberspace security. Currently, virtual private network (VPNs) has become one of the most popular encrypted communication services for bypassing censorship and guarantee remote access to geographically locked services. In this paper, a novel identification scheme of VoIP traffic tunneled through VPN is proposed. We employed a set of Flow Spatio-Temporal Features (FSTF) to six well-known classifiers, including decision trees, K-Nearest Neighbor (KNN), Bagging and Boosting via C4.5, and Multi-Layer perceptron (MLP). The overall accuracy, precision, sensitivity, and F-measure verify that the proposed scheme can effectively distinguish between the VoIP flows and Non-VoIP ones in VPN traffic.