Electrocardiography (ECG) is a conventional method in arrhythmia diagnosis. In this paper, we proposed a novel neural network model which treats typical heartbeat classification task as ‘Translation’ problem. By introducing Transformer structure into model, and adding heartbeat-aware attention mechanism to enhance the alignment between encoded sequence and decoded sequence, after trained with ECG database, (which are collected from 200k patients in over 2000 hospitals for more than 10 years), the validation result of independent test dataset shows that this new heartbeat-aware Transformer model can outperform classic Transformer and other sequence to sequence methods. Finally, we show that the visualization of encoder-decoder attention weights provides more interpretable information about how a Transformer make a diagnosis based on raw ECG signals, which has guiding significance in clinical diagnosis.

Cryptography has been used as the first line of defense to ensure data integrity. However, it incurs a significant overhead for short packets which represent the most common form of traffic in emerging applications, such as the Internet of Things. In this paper, we propose an interdisciplinary framework for detecting false data in short packets that are collected from multiple sources through a relay. The proposed scheme performs a lightweight cryptographic integrity check that employs a short message authentication code, thereby reducing the overhead for data integrity, followed by the maximum a posteriori (MAP) test at the physical layer that detects any remaining false data that the cryptographic integrity check misses. This interdisciplinary approach interplays between the cryptographic integrity check and the physical-layer integrity check to assure the data integrity of short packets with reduced overhead and low latency. The proposed scheme can also provide a significant reduction in the average probability of detection error over the traditional cryptographic integrity check.

We present the physical-layer maximum a posteriori (MAP) test for detecting false data in multiple-source wireless relay networks. The MAP test, which minimizes the average probability of detection error, exploits the noisy (erroneous) information about the true packet that the destination receives (overhears) directly from the source nodes as a reference. We show that the average probability of detection error decreases exponentially as the number of source nodes increases, regardless of the reliability of the reference information. This indicates that the multiplicity of source nodes can be leveraged to construct a powerful integrity check mechanism for detecting false data at the destination. We also present a parallel MAP test that improves the detection accuracy when the number of source nodes is limited and overhearing channel disparities are high. Then, we present a packet recovery mechanism to recover false alarmed packets, i.e. unmodified packets that are mistakenly declared as modified, by the MAP test and thereby to improve the throughput.