Phishing has been one of the most successful attacks in recent years. Criminals are motivated by increasing financial gain and constantly improving their email phishing methods. A key goal, therefore, is to develop effective detection methods to cope with huge volumes of email data. In this paper, a solution using BLSTM neural network and FastText word embeddings has been proposed. The solution uses preprocessing techniques like stop-word removal, tokenization, and padding. Two datasets were used in three experiments: balanced and imbalanced, whereas in the imbalanced dataset, the effect of maximum token size was investigated. Evaluation of the model indicated the best metrics: 99.12% accuracy, 98.43% precision, 99.49% recall, and 98.96% f1-score on the imbalanced dataset. It was compared to an existing solution that uses the DL model and word embeddings. Finally, the model and solution architecture were implemented as a browser plug-in.
A methodology for development for distributed computer network (DCN) information security system (IS) for an informatization object (OBI) was proposed. It was proposed to use mathematical modeling at the first stage of the methodology. In particular, a mathematical model was presented based on the use of the apparatus of probability theory to calculate the vulnerability coefficient. This coefficient allows one to assess the level of information security of the OBI network. Criteria for assessing the acceptable and critical level of risks for information security were proposed as well. At the second stage of the methodology development of the IS DCN system, methods of simulation and virtualization of the components of the IS DCN were used. In the course of experimental studies, a model of a protected DCN has been built. In the experimental model, network devices and DCN IS components were emulated on virtual machines (VMs). The DCN resources were reproduced using the Proxmox VE virtualization system. IPS Suricata was deployed on RCS hosts running PVE. Splunk was used as SIEM. It has been shown that the proposed methodology for the formation of the IS system for DCN and the model of the vulnerability coefficient makes it possible to obtain a quantitative assessment of the levels of vulnerability of DCN OBI.
With rapid population increases, people are facing the challenge to maintain healthy conditions. One of the challenges is air pollution. Due to industrial development and vehicle usage air pollution is becoming a high threat to human life. This air pollution forms through various toxic contaminants. This toxic contamination levels increase and cause severe damage to the living things in the environment. To identify the toxic level present in the polluted air various methods were proposed by the authors, But failed to detect the tolerance level of toxic gases. This article discusses the methods to detect toxic gasses and classify the tolerance level of gasses present in polluted air. Various sensors and different algorithms are used for classifying the tolerance level. For this purpose “Artificial Sensing Methodology” (ASM), commonly known as e-nose, is a technique for detecting harmful gases. SO2-D4, NO2-D4, MQ-135, MQ136, MQ-7, and other sensors are used in artificial sensing methods (e-nose). “Carbon monoxide, Sulfur dioxide, nitrogen dioxide, and carbon dioxide” are all detected by these sensors. The data collected by sensors is sent to the data register from there it is sent to the Machine learning Training module (ML) and the comparison is done with real-time data and trained data. If the values increase beyond the tolerance level the system will give the alarm and release the oxygen.
Growing popularity of distributed generation is drawing special attention to communication technologies in smart power grids. This paper provides a detailed overview of the communication protocols utilized in the modern distributed grid laboratory. It describes both wired and wireless technologies used in Smart Grid and presents the remote operation of switching the subsystem from grid mode to island mode operating under nominal conditions. It shows the duration of power outages during a transfer to island mode with diesel generator running on idle - which simulates planned islanding and diesel generator stationary, which simulates unplanned islanding. Latency between registration of disturbance and executing control command is measured. The results obtained are compared with current legislation. The consequences to the power system that are possible in both scenarios are highlighted. Obtained results and description of the communication technologies can be useful for the design of distributed power grids, island-mode power grids, and Smart Grids, as well as for further research in the area of using combustion fuel generators as a primary power supply in the microgrid.
Drinking fresh water, turning the lights on, travelling
by tram, calling our family, or getting a medical treatment are
usual activities, but the underlying SCADA (Supervisory Control
and Data Acquisition) systems like CIS (Critical Infrastructure
Systems), ICS (Industrial Control Systems) or DCS (Distributed
Control Systems) were always the target of many types of attacks,
endangered the above mentioned simple activities. During the last
decades because of the fast spread of the internet based services
and the continuous technical development these systems become
more vulnerable than ever. Full reconstruction and innovative
changes in older SCADA systems has high cost, and it is not always
rewarding. Communication protocols as Modbus (1979) serve as a
main basis for SCADA systems, so security of Modbus has a major
impact of the security of SCADA systems. Our paper raises and
answers questions about the security of the Modbus RTU protocol.
We focus on the serial Modbus protocol, because in that method
we found many unsolved problems, like lack of authentication of
the participants, lack of secure channel and so on. The aim of this
paper to propose a secure communication alternative for Modbus
RTU @ RS485 wire. The main advantage of the proposed method
is the coexistence with traditional slaves and bus systems and only
software update is necessary
Although self-modifying code has been shyed away
from due to its complexity and discouragement due to safety
issues, it nevertheless provides for a very unique obfuscation
method and a different perspective on the relationship between
data and code. The generality of the von Neumann architecture is
hardly realized by today’s processor models. A code-only model is
shown where every instruction merely modifies other instructions
yet achieves the ability to compute and Turing machine operation
is easily possible.
Games are among problems that can be reduced to
optimization, for which one of the most universal and productive
solving method is a heuristic approach. In this article we present
results of benchmark tests on using 5 heuristic methods to solve
a physical model of the darts game. Discussion of the scores and
conclusions from the research have shown that application of
heuristic methods can simulate artificial intelligence as a regular
player with very good results.
This paper tries to get a response to the following question: When can a narrowband power amplifier (PA) be
considered to be memoryless and when can it not be considered memoryless? To this end, a thorough and consistent analysis of
the notions and definitions related with the above topic is carried out. In the considerations presented, two models of the
narrowband PA are exploited interchangeably: the black box model widely used in the literature and a model developed here,
which is based on the Volterra series. These two models complement each other. In this paper, the conditions for a linear
or nonlinear narrowband PA to be memoryless or approximately memoryless or possessing memory are derived and illustrated.
They are formulated in terms of the signal delay as well as in terms of the amplitude-to-phase (AM/PM) conversion of the
amplifier. Furthermore, the two possible interpretations of the amplitude-to-amplitude (AM/AM) and AM/PM conversions are
given a mathematical framework. That is these conversions are presented through some operations. One set of these operations
allows to treat the AM/AM and AM/PM conversions as distortions of the modulating signals. Or equivalently as
distortions of a given signal constellation when it passes through the PA. Finally, it is proved that the Saleh’s and Ghorbani’s
models of the AM/AM and AM/PM conversions occurring in the PAs, which were published in the literature, are not memoryless
ones.
In the paper, we demonstrate the feasibility of interdigital electrodes
fabrication with the usage of inkjet printing technology. The emphasis
was put to obtain better shape quality and lower spacing between
electrodes with respect to typical printing process. The paper presents
an analysis of the main factors that have an influence on the dimension
and quality of printed structures and proposes two methods that allow
eliminating the main problems. The first proposed method is based on
controlling the time between patterning of successive drops. While the
second method is based on changing the design methods considering
printing orientation. Both methods do not require any additional
technological processes or the use of any special surface preparation
methods. Finally, the obtained results and conclusions were presented
and discussed.