With the rapid development of WLAN technology, wireless device-free passive human detection becomes a newly-developing technique and holds more potential to worldwide and ubiquitous smart applications. Recently, indoor fine-grained device-free passive human motion detection based on the PHY layer information is rapidly developed. Previous wireless device-free passive human detection systems either rely on deploying specialized systems with dense transmitter-receiver links or elaborate off-line training process, which blocks rapid deployment and weakens system robustness. In the paper, we explore to research a novel fine-grained real-time calibration-free device-free passive human motion via physical layer information, which is independent of indoor scenarios and needs no prior-calibration and normal profile. We investigate sensitivities of amplitude and phase to human motion, and discover that phase feature is more sensitive to human motion, especially to slow human motion. Aiming at lightweight and robust device-free passive human motion detection, we develop two novel and practical schemes: short-term averaged variance ratio (SVR) and long-term averaged variance ratio (LVR). We realize system design with commercial WiFi devices and evaluate it in typical multipath-rich indoor scenarios. As demonstrated in the experiments, our approach can achieve a high detection rate and low false positive rate.

Android applications have emerged as a prime target for hackers. Android malware detection stands as a pivotal technology, crucial for safeguarding network security and thwarting anomalies. However, traditional static analysis makes it difficult to analyze new malicious applications, while dynamic analysis requires higher system resources. We propose a novel lightweight Android malware deep-learning detection framework based on attention temporal networks. This study delves into the Dalvik opcode sequences of Android malware, employing the N-gram algorithm to partition sequences and extract contextual information features. Then, LSTM and TCN algorithms are employed to capture long-term dependencies and local features, enabling comprehensive comprehension of temporal information within Dalvik opcode sequences. Especially, TCN facilitates feature extraction across various time scales, thereby enabling the detection of anomaly patterns across diverse temporal scales within Dalvik opcode sequences. Moreover, we introduce multi-head attention mechanisms and reinforced learning to direct the model’s focus toward behavioral cues within malicious software sequences. Finally, extensive experiment results show that our proposed methodology and model exhibit higher detection accuracy and robustness, achieving an accuracy rate of 98.69% on average, surpassing traditional machine learning methods such as random forest, and Pseudo-Label deep neural networks.

The extremely destructive nature of malware has become a major threat to Internet security. The research on malware detection techniques has been evolving. Deep learning-based malware detection methods have achieved good results by using large-scale, pre-trained models. However, these models are complex, have large parameters, and require a large amount of hardware resources and have a high inference time cost when applied. To address this challenge, this paper proposes DistillMal, a new method for lightweight malware detection based on knowledge distillation, which improves performance by using a student network to learn valuable cueing knowledge from a teacher network to achieve a lightweight model. We conducted extensive experiments on two new datasets and showed that the student network model’s performance is very close to that of the original model and the outperforms it on some metrics. Our approach helps address the resource constraints and computational challenges faced by traditional deep learning large models. Our research highlights the potential of using knowledge distillation to develop lightweight malware detection models.

Microblog has emerged as a popular medium for providing new sources of information and rapid communications, particularly during burst topics. Burst keywords detection from real-time microblog streams is important for burst topics detection. The exiting algorithms may detect fake burst keywords without taking into account the trustworthiness of the users and human's daily timetable. Our work is the first to combine the trustworthiness of the users with burst keywords detection. We propose a novel approach to detect burst keywords based on social trust and dynamics model. We adapt basic notions of dynamics from physics and model keywords bursts as momentum change of the keywords. On the analogy of physical dynamics model, this approach defines mass as the trustworthiness of user and position as the frequency of keywords. We compute each keyword's burst value by using Moving average convergence/divergence (MACD) and determine whether it is a burst keyword in a given time window. The experimental results on large-scale Sina microblog dataset show that the proposed approach can avoid detecting fake burst keywords.