Multi-Scale computingsystems aim at bringing the computing as close as possible to the data sources, to optimize both computation and networking. These systems are composed of at least three computing layers: the ter...
详细信息
Quantum Key Distribution (QKD) is a secure communication method that uses quantum mechanics to exchange secret keys. Currently, the QKD process uses a static error correction method, which can lead to inaccuracies and...
详细信息
Log anomaly detection on edge devices is one of the key components to ensure system security and diagnose system faults. Existing log anomaly detection methods usually contain a large number of parameters and computat...
详细信息
ISBN:
(纸本)9798350359329;9798350359312
Log anomaly detection on edge devices is one of the key components to ensure system security and diagnose system faults. Existing log anomaly detection methods usually contain a large number of parameters and computations. However, processing large-scale logs on edge devices is still a bottleneck due to the limited computational power of edge devices, which can hardly afford the complexity of these detection models. Hence, we propose a lightweight and efficient log anomaly detection method called TinyLog, which aims to facilitate the identification of anomalous log patterns in largescale IoT systems. Specifically, TinyLog uses Word2Vec and TF-IDF to construct weighted log semantic vector. Further, a lightweight temporal convolutional network based on depthwise separable convolution and triplet attention is designed for anomaly detection in edge devices. Experimental results show that TinyLog outperforms other state-of-the-art benchmark methods in terms of performance and efficiency, while its computational overhead is extremely low. Thus, our approach can be effectively applied to resource-constrained edge devices.
This work proposed a vacancy-modulated analog resistive random-access memory (ReRAM) for neuromorphic computing, utilizing a structure composed of Ag/Zn@ZnO/ZnO/FTO. The resistive switching behavior can be modulated b...
详细信息
The adaptability of devices can be significant for a customer that inserts them in an industrial production line. The ability to modify an object bought along with a machine that can be personalized with its features ...
详细信息
ISBN:
(纸本)9798350316971
The adaptability of devices can be significant for a customer that inserts them in an industrial production line. The ability to modify an object bought along with a machine that can be personalized with its features can change how they want to do measurements for different reasons, like predictive maintenance. Fog computing local centers already exist in the market, but they are usually on-the-shelf products with no margin of change for any user. However, with the usage of Docker and containers, this can change. This paper describes a fog computing local central called Concentrator, which can not only execute its essential functions built-in by the producer but also be customized by the user to add in the elaborations on other external sensors, expanding its capabilities and usage. We wanted to improve the device already tested on a Linux PC on a Raspberry Pi and try its performance and characteristics, seeing if it could be transformed into an embedded architecture and an industrial feature.
E-voting system is a digital mechanism that enables people to vote electronically, ensuring confidentiality, security, and transparency. Visual cryptography (VC) is a cryptographic technique where a secret image is di...
详细信息
Being dangerous threats and attacks in this cyber era, botnets require proper handling. Nevertheless, some bot detection models that have been proposed are centralized and can only detect at one point of attack, even ...
详细信息
ISBN:
(纸本)9781665499736;9781665499729
Being dangerous threats and attacks in this cyber era, botnets require proper handling. Nevertheless, some bot detection models that have been proposed are centralized and can only detect at one point of attack, even though there are two known types of botnet activity: single and bot group. In fact, attacks from grouped bots can have a series of attacks with the same pattern at several different attack targets. So, it requires a distributed detection model that can detect bot attacks on some detection sensors and assemble them in the form of correlation analysis. This paper proposes a prototype distributed botnet detection model that can synchronize detection in each detection sensor and analyze a series of bot attack activities. It aims to obtain information on the series of attacks that occur at several attack points and state as a correlated botnet attack scenario. With the existence of a distributed botnet activity detection prototype, it will be able to facilitate the analysis and anticipation process from the system and network security administrators.
In post-disaster scenarios, field robots and unmanned aerial vehicles (UAVs) are effective tools to provide an ondemand response and guarantee the safety of humans. Sensors, field robots, and the devices equipped by U...
详细信息
Over the years, Cloud computing has significantly influenced various sectors, including Web learning, as education is considered crucial for individual and national growth. The primary objective is to develop an Appli...
详细信息
As social media usage continues to expand daily, so too does the spread of hate speech. To swiftly remove inflammatory content, every social media platform needs to implement a dependable algorithm for identifying hat...
详细信息
暂无评论