期刊,Mobile networks & applications 2024年29卷6期_国家学术搜索

Intrusion Detection System Using Hybrid Convolutional Neural Network

Amani K. SamhaNidhi MalikDeepak SharmaKavitha S...

1719-1731页

查看更多>>摘要：The exponential development of the internet and electronic communications has led to a tremendous rise in the quantity of data exchanged. Intruders are always developing novel techniques to acquire or manipulate such data since they are so valuable. A rising number of such attacks pose a threat to the safety of the networks and is a formidable obstacle to intrusion detection. An IDS analyses network traffic to spot potential threats. There have been numerous studies and innovative IDS developed, however, IDS still requires improvements to have excellent detection performance with minimizing false rates. The challenges of classic IDS systems are as follows: Feature selection is crucial, and it is necessary for enhancing performance. Unbalanced data can limit efforts to improve accuracy. However, most IDS have trouble identifying zero-day attacks. In this paper, we take a hybrid Convolution Neural Network and Deep Watershed Auto-encoder (CNN-DWA) approach to address the above-mentioned challenges. The suggested network is trained and evaluated using the KDD CUP 1999 dataset. The benefits of the suggested model are demonstrated by comparing the results obtained using the CNN-DWA approach with the Convolution Neural Network (CNN) method. The results of the experiments indicate that the suggested model has a higher accuracy (98.05%) than CNN (94.54%).

原文链接:

NETL
NSTL
Springer Nature

Intrusion Detection System Using Hybrid Convolutional Neural Network

Amani K. SamhaNidhi MalikDeepak SharmaKavitha S...

1719-1731页

查看更多>>摘要：The exponential development of the internet and electronic communications has led to a tremendous rise in the quantity of data exchanged. Intruders are always developing novel techniques to acquire or manipulate such data since they are so valuable. A rising number of such attacks pose a threat to the safety of the networks and is a formidable obstacle to intrusion detection. An IDS analyses network traffic to spot potential threats. There have been numerous studies and innovative IDS developed, however, IDS still requires improvements to have excellent detection performance with minimizing false rates. The challenges of classic IDS systems are as follows: Feature selection is crucial, and it is necessary for enhancing performance. Unbalanced data can limit efforts to improve accuracy. However, most IDS have trouble identifying zero-day attacks. In this paper, we take a hybrid Convolution Neural Network and Deep Watershed Auto-encoder (CNN-DWA) approach to address the above-mentioned challenges. The suggested network is trained and evaluated using the KDD CUP 1999 dataset. The benefits of the suggested model are demonstrated by comparing the results obtained using the CNN-DWA approach with the Convolution Neural Network (CNN) method. The results of the experiments indicate that the suggested model has a higher accuracy (98.05%) than CNN (94.54%).

原文链接:

NETL
NSTL
Springer Nature

Efficient and Secure Graph-Based Trust-Enabled Routing in Vehicular Ad-Hoc Networks

Intyaz AlamManisha ManjulVinay PathakVajenti Mala...

1732-1752页

查看更多>>摘要：Vehicular Ad hoc Networks (VANETs) have gained significant recognition as a prospective technology for augmenting road safety and optimizing traffic efficiency through facilitating instantaneous communication between vehicles and roadside infrastructure. However, routing in VANETs faces significant challenges due to the dynamic network topology and security threats. In this context, trust-based routing offers an effective solution by improving reliability, security, and quality of service (QoS) in vehicle-to-infrastructure communication. However, trust-based routing in IOVs requires reliable trust evaluation mechanisms, privacy preservation, authentication, and access control. Challenges arise from the dynamic nature of IOVs, necessitating scalable and efficient trust computation algorithms. Moreover, ensuring the resilience of trust-based routing against malicious attacks, such as Sybil attacks or collusion among malicious vehicles, is an issue of great importance that necessitates attention and resolution. This research paper proposes a novel Graph-Based Trust-Enabled Routing (GBTR) scheme specifically designed for VANETs. The scheme incorporates direct trust, indirect trust, and contextual trust to evaluate the trustworthiness of participating nodes. Direct trust is determined based on factors such as frequency and consistency of successful communication, communication delay, and a mobility factor that incorporates punishment/reward parameters. Indirect trust is calculated using feedback trust value and link reliability, also considering the mobility factor. The contextual trust incorporates factors like location, time of day, weather conditions, and traffic density for each node pair. Routing decisions are made based on the final trust scores obtained from these trust evaluations. The route request/reply mechanism and route maintenance mechanism ensure the selection of the most reliable and trustworthy routes, thereby improving network performance. Additionally, a trust update algorithm with a concept of less reward and more penalty is employed to periodically update the trust values of participating vehicles. This approach enhances security, reliability, robustness, and efficiency of network resource usage, reducing congestion and enabling real-time trust evaluation while minimizing false positives. The simulation results substantiate that the GBTR scheme, as proposed, surpasses existing routing schemes across various performance metrics, including packet delivery ratio (PDR%), dropped packet ratio (DPR%), end-to-end delay (ms), throughput (Kbps), and normalized routing load (packets/sec). These outcomes underscore the efficacy of the proposed scheme in enhancing network performance and bolstering reliability. Overall, the graph-based trust-enabled routing scheme presented in this research contributes to enhancing the reliability and security of VANETs, thereby supporting the development of intelligent transportation systems.

原文链接:

NETL
NSTL
Springer Nature

Efficient and Secure Graph-Based Trust-Enabled Routing in Vehicular Ad-Hoc Networks

Intyaz AlamManisha ManjulVinay PathakVajenti Mala...

1732-1752页

查看更多>>摘要：Vehicular Ad hoc Networks (VANETs) have gained significant recognition as a prospective technology for augmenting road safety and optimizing traffic efficiency through facilitating instantaneous communication between vehicles and roadside infrastructure. However, routing in VANETs faces significant challenges due to the dynamic network topology and security threats. In this context, trust-based routing offers an effective solution by improving reliability, security, and quality of service (QoS) in vehicle-to-infrastructure communication. However, trust-based routing in IOVs requires reliable trust evaluation mechanisms, privacy preservation, authentication, and access control. Challenges arise from the dynamic nature of IOVs, necessitating scalable and efficient trust computation algorithms. Moreover, ensuring the resilience of trust-based routing against malicious attacks, such as Sybil attacks or collusion among malicious vehicles, is an issue of great importance that necessitates attention and resolution. This research paper proposes a novel Graph-Based Trust-Enabled Routing (GBTR) scheme specifically designed for VANETs. The scheme incorporates direct trust, indirect trust, and contextual trust to evaluate the trustworthiness of participating nodes. Direct trust is determined based on factors such as frequency and consistency of successful communication, communication delay, and a mobility factor that incorporates punishment/reward parameters. Indirect trust is calculated using feedback trust value and link reliability, also considering the mobility factor. The contextual trust incorporates factors like location, time of day, weather conditions, and traffic density for each node pair. Routing decisions are made based on the final trust scores obtained from these trust evaluations. The route request/reply mechanism and route maintenance mechanism ensure the selection of the most reliable and trustworthy routes, thereby improving network performance. Additionally, a trust update algorithm with a concept of less reward and more penalty is employed to periodically update the trust values of participating vehicles. This approach enhances security, reliability, robustness, and efficiency of network resource usage, reducing congestion and enabling real-time trust evaluation while minimizing false positives. The simulation results substantiate that the GBTR scheme, as proposed, surpasses existing routing schemes across various performance metrics, including packet delivery ratio (PDR%), dropped packet ratio (DPR%), end-to-end delay (ms), throughput (Kbps), and normalized routing load (packets/sec). These outcomes underscore the efficacy of the proposed scheme in enhancing network performance and bolstering reliability. Overall, the graph-based trust-enabled routing scheme presented in this research contributes to enhancing the reliability and security of VANETs, thereby supporting the development of intelligent transportation systems.

原文链接:

NETL
NSTL
Springer Nature

Applying an Improved Squirrel Search Algorithm (ISSA) for Clustering and Low-Energy Routing in Wireless Sensor Networks (WSNs)

Alaa A. Qaffas

1753-1768页

查看更多>>摘要：Wireless sensor networks (WSNs) plays an important role in the advancement of the industrial 4.0/5.0 revolution, as they are integrated into IoT and other cyber-physical systems. Various techniques have been developed to optimize WSNs, but there are still some limitations. Clustering has been a highly effective method for efficient communication with low power and battery usage in high-speed communication systems. In this work, an Improved Squirrel Search Algorithm (I-SSA) is developed for selecting the cluster head (CH) node in WSNs. The improved I-SSA algorithm introduces a sine chaotic mapping strategy to boost population diversity, a backward learning mechanism to constrict the selection of excellent solution sets, and a cross-learning mechanism to enhance the accuracy of the algorithm optimization procedure in order to confront the drawbacks of the SSA algorithm, including such easy having fallen into local optimal solution and inadequate variability. The fitness function, which is employed to assess the performance of the solutions generated by the optimization algorithm, plays a key role in this cluster head selection process. Four parameters including residual energy, average intra-cluster distance, average sink distance, and CH balance factor are used in the fitness function. Network density analysis was performed by changing the number of sensor nodes (SNs) from 100 to 500 and randomly distributing them in the simulation region. For the simulation study, we utilize the most recent and stable release of MATLAB, version 2021a. Results from the simulation indicate that the proposed I-SSA based approach improves network performance and uses more stable energy compared to existing techniques such as SSO-CHST, ACO-based, and GA-based methods.

原文链接:

NETL
NSTL
Springer Nature

Applying an Improved Squirrel Search Algorithm (ISSA) for Clustering and Low-Energy Routing in Wireless Sensor Networks (WSNs)

Alaa A. Qaffas

1753-1768页

查看更多>>摘要：Wireless sensor networks (WSNs) plays an important role in the advancement of the industrial 4.0/5.0 revolution, as they are integrated into IoT and other cyber-physical systems. Various techniques have been developed to optimize WSNs, but there are still some limitations. Clustering has been a highly effective method for efficient communication with low power and battery usage in high-speed communication systems. In this work, an Improved Squirrel Search Algorithm (I-SSA) is developed for selecting the cluster head (CH) node in WSNs. The improved I-SSA algorithm introduces a sine chaotic mapping strategy to boost population diversity, a backward learning mechanism to constrict the selection of excellent solution sets, and a cross-learning mechanism to enhance the accuracy of the algorithm optimization procedure in order to confront the drawbacks of the SSA algorithm, including such easy having fallen into local optimal solution and inadequate variability. The fitness function, which is employed to assess the performance of the solutions generated by the optimization algorithm, plays a key role in this cluster head selection process. Four parameters including residual energy, average intra-cluster distance, average sink distance, and CH balance factor are used in the fitness function. Network density analysis was performed by changing the number of sensor nodes (SNs) from 100 to 500 and randomly distributing them in the simulation region. For the simulation study, we utilize the most recent and stable release of MATLAB, version 2021a. Results from the simulation indicate that the proposed I-SSA based approach improves network performance and uses more stable energy compared to existing techniques such as SSO-CHST, ACO-based, and GA-based methods.

原文链接:

NETL
NSTL
Springer Nature

Applying Machine Learning Approach to Identifying Channels in MlMO Networks for Communications in 5G-Enabled Sustainable Smart Cities

Shalini StalinManish GuptaK. MakanyadeviAmit Agrawal...

1769-1781页

查看更多>>摘要：During the process of sending a signal across a transmission channel that has a broad bandwidth, the Multiple-Input Multiple-Output, or MIMO, system will frequently require a larger quantity of energy and power than it would under normal circumstances. This is because the transmission channel will have a greater capacity to carry more data. When there are so many different sets of signals that have been found, it is necessary to be able to differentiate between them using one of the many different methods that have been developed. One of the methods that is now one of the most extensively employed in the process of discovery is called maximum likelihood detection. This method is also frequently referred to as machine learning (ML) detection. This is as a result of the very high throughput that ML detection has. On the other hand, the exponential growth in ideal throughput is far smaller than the complexity of the framework and the amount of energy that it consumes. In order to better simplify communication between the transmitter and the receiver, the major goal of this endeavor is to locate the shortest viable routing route for the physical data transmission layer and use that information to design the routing path. Because of this, both the quantity of energy that is used and the level of complexity that the system has will both drop. Improved Iterative Based Dijkstra Algorithm (IIBDA) to Gauge the Most Limited Course of the Channel with a Restricted Maximum Likelihood-Detection (RMLD) Design was a method that was presented to solve the problem. This was done in order to address the concerns that had been raised and resolved in the prior discussion. This was done in an effort to discover answers to problems such as these. The Execution Examination illustrates how the Complexity and Control Utilization in the Proposed Strategy may be Decreased by showing how these factors might be Addressed. An FPGA Virtex-6 was used for putting the suggested plan into action in order to accomplish this. When the recommended IIBDA-RMLD were put through their paces in terms of power consumption, area, time delay, and complexity, all of these characteristics exhibited improvements of up to 95.4%, 84.23%, 84.21%, 87.23%,90.14% respectively. These percentages represent the maximum levels of improvement that were observed. MIMO Transmission System is able to achieve a significantly higher Signal to Noise Ratio (Snr) demonstrating with reduced power usage of 0.538mw and range optimization accomplished 11093.13 m for Quadrature Phase Shift Keying (QPSK) balancing with the ML Location System using Feed Forward Neural Network (FFNN). This is in addition to the fact that the MIMO Transmission System can reduce the amount of power it uses by 0.538mw. This is accomplished with a reduced quantity of usage of electrical power. In conclusion, RMLD was employed on MIMO networks in order to enhance the transmission of information in military and other applications by making it more transportable and safer. This was done for a variety of reasons. This action was taken for a number of different reasons.

原文链接:

NETL
NSTL
Springer Nature

Applying Machine Learning Approach to Identifying Channels in MlMO Networks for Communications in 5G-Enabled Sustainable Smart Cities

Shalini StalinManish GuptaK. MakanyadeviAmit Agrawal...

1769-1781页

查看更多>>摘要：During the process of sending a signal across a transmission channel that has a broad bandwidth, the Multiple-Input Multiple-Output, or MIMO, system will frequently require a larger quantity of energy and power than it would under normal circumstances. This is because the transmission channel will have a greater capacity to carry more data. When there are so many different sets of signals that have been found, it is necessary to be able to differentiate between them using one of the many different methods that have been developed. One of the methods that is now one of the most extensively employed in the process of discovery is called maximum likelihood detection. This method is also frequently referred to as machine learning (ML) detection. This is as a result of the very high throughput that ML detection has. On the other hand, the exponential growth in ideal throughput is far smaller than the complexity of the framework and the amount of energy that it consumes. In order to better simplify communication between the transmitter and the receiver, the major goal of this endeavor is to locate the shortest viable routing route for the physical data transmission layer and use that information to design the routing path. Because of this, both the quantity of energy that is used and the level of complexity that the system has will both drop. Improved Iterative Based Dijkstra Algorithm (IIBDA) to Gauge the Most Limited Course of the Channel with a Restricted Maximum Likelihood-Detection (RMLD) Design was a method that was presented to solve the problem. This was done in order to address the concerns that had been raised and resolved in the prior discussion. This was done in an effort to discover answers to problems such as these. The Execution Examination illustrates how the Complexity and Control Utilization in the Proposed Strategy may be Decreased by showing how these factors might be Addressed. An FPGA Virtex-6 was used for putting the suggested plan into action in order to accomplish this. When the recommended IIBDA-RMLD were put through their paces in terms of power consumption, area, time delay, and complexity, all of these characteristics exhibited improvements of up to 95.4%, 84.23%, 84.21%, 87.23%,90.14% respectively. These percentages represent the maximum levels of improvement that were observed. MIMO Transmission System is able to achieve a significantly higher Signal to Noise Ratio (Snr) demonstrating with reduced power usage of 0.538mw and range optimization accomplished 11093.13 m for Quadrature Phase Shift Keying (QPSK) balancing with the ML Location System using Feed Forward Neural Network (FFNN). This is in addition to the fact that the MIMO Transmission System can reduce the amount of power it uses by 0.538mw. This is accomplished with a reduced quantity of usage of electrical power. In conclusion, RMLD was employed on MIMO networks in order to enhance the transmission of information in military and other applications by making it more transportable and safer. This was done for a variety of reasons. This action was taken for a number of different reasons.

原文链接:

NETL
NSTL
Springer Nature

SupportNet: a Deep Learning Based Channel Equalization Network for Multi-type Multipath Fading

Yibo ChenHonglian LiShengbin ZhuangXing Wei...

1782-1795页

查看更多>>摘要：High-speed moving receivers generate Doppler shift superimposed on multipath effects to produce serious self-interference in the signal, direct channel equalization is more difficult, often requiring channel estimation, although neural networks can perform channel estimation and channel equalization, the neural network training process requires both the corresponding channel estimation and channel equalization results of the two labels as a loss function. It is more difficult to take labels for channel estimation in realistic scenarios, there are small errors in channel estimation by various methods, and the use of a large number of channel estimation labels causes an increase in data cost. This paper proposes a channel equalisation model called SupportNet, which simulates both channel estimation and channel equalisation processes by inducing a sub-network into a model collapse state so that a part of the network acts like channel estimation without using channel estimation labels, allowing features to be separated and processed separately, and using the channel estimation results for channel equalisation to reduce BER. The property of neural networks that rely on gradient descent for training to produce pattern collapse allows the network to separate features without the need to add labels to each feature. The experimental results show that the homogeneous network can effectively reduce the impact caused by time-selective fading under the fast-fading channel generated by the physical layer emulation parameters of three mobile environment provided by the IEEE 802.11p standard, resulting in a lower BER.

原文链接:

NETL
NSTL
Springer Nature

SupportNet: a Deep Learning Based Channel Equalization Network for Multi-type Multipath Fading

Yibo ChenHonglian LiShengbin ZhuangXing Wei...

1782-1795页

查看更多>>摘要：High-speed moving receivers generate Doppler shift superimposed on multipath effects to produce serious self-interference in the signal, direct channel equalization is more difficult, often requiring channel estimation, although neural networks can perform channel estimation and channel equalization, the neural network training process requires both the corresponding channel estimation and channel equalization results of the two labels as a loss function. It is more difficult to take labels for channel estimation in realistic scenarios, there are small errors in channel estimation by various methods, and the use of a large number of channel estimation labels causes an increase in data cost. This paper proposes a channel equalisation model called SupportNet, which simulates both channel estimation and channel equalisation processes by inducing a sub-network into a model collapse state so that a part of the network acts like channel estimation without using channel estimation labels, allowing features to be separated and processed separately, and using the channel estimation results for channel equalisation to reduce BER. The property of neural networks that rely on gradient descent for training to produce pattern collapse allows the network to separate features without the need to add labels to each feature. The experimental results show that the homogeneous network can effectively reduce the impact caused by time-selective fading under the fast-fading channel generated by the physical layer emulation parameters of three mobile environment provided by the IEEE 802.11p standard, resulting in a lower BER.

原文链接:

NETL
NSTL
Springer Nature