Using deep learning networks, anomaly detection systems have seen better performance and precision. However, adversarial examples render deep learning-based anomaly detection systems insecure since attackers can fool them, increasing the attack success rate. Therefore, improving anomaly systems' robustness against adversarial attacks is imperative. This paper tests adversarial examples against three anomaly detection models based on Convolutional Neural Network (CNN), Long Short-term Memory (LSTM), and Deep Belief Network (DBN). It assesses the susceptibility of current datasets (in particular, UNSW-NB15 and Bot-IoT datasets) that represent the contemporary network environment. The result demonstrates the viability of the attacks for both datasets where adversarial samples diminished the overall performance of detection. The result of DL Algorithms gave different results against the adversarial samples in both our datasets. The DBN gave the best performance on the UNSW dataset.

Due to the increased strain each new Internet-connected item puts on the IPv4 infrastructure, the emergence of additional Internet-connected places and devices has accelerated IPv4 exhaustion. Service providers have been obliged to invest in infrastructure to handle greater traffic due to unexpected growth in subscribers and linked IoT devices. Service providers are struggling to maintain growth and business continuity due to the expiration of IPv4 globally and the adoption of IPv6. There is strongly a need to address both a short-term solution for the maintenance of their current IPv4 address allocation and a long-term solution for a seamless transition to an IPv6 infrastructure, various service providers will need to design an address translation strategy. This paper presents a solution using CGNAT towards the migration of IPv6 networks. A general overview of the various parts needed to manage the depletion of IPv4 addressing and the engagement of full carrier grade network address translation solution is also discussed in this paper along with the different types of NAT.

Quantum cryptography is the most convenient resolution for information security systems that presents an ultimate approach for key distribution. Today, the most viable key distribution resolutions for information security systems are those based on quantum cryptography. It is based on the quantum rules of physics rather than the assumed computational complexity of mathematical problems. But, the initial BB84 quantum key distribution protocol which is the raw key exchange of S13 quantum key distribution protocol has weakness of disclosure of large portion of secrete key or eavesdropping. Also, it cannot make use of most of the generated random bit. This paper enhanced S13 quantum key distribution protocol by employing polarization, secrete key disclosure and non-repudiation. The use of biometric or MAC address ensures non-repudiation. The row key exchange part of the S13 quantum key distribution which is the same as BB84 is enhanced by employing polarization techniques to make use of most of the generated random bit. Then, the tentative final key generated at the end of error estimation phase should be divided into blocks, padding, inverting the last bit of each block and XORing the block to generate a totally different key from the tentative one. Also, the random bits will be from biometric or serve MAC address respectively. The enhanced S13 quantum key is evaluated using cryptanalysis which shows that the enhanced protocol ensures disclosures of large portion of secrete key to prevent eavesdropping, utilization of most of the chosen binary strings to generate strong key and safeguarding against impersonation attack.

In this paper, in order to improve the coverage and the Quality of Service of end-users on the edge of a cellular network, the use of unmanned aerial vehicles (UAVs) is employed to give them direct line of sight. In addition to that, to improve the performance of the said UAVs, reconfigurable intelligent surfaces (RIS) are introduced to the model, in such a way that will enhance the connection of the UAVs with the base station. An RIS will receive a signal from the base station, modulate it and then the RIS will act as the transmitter, sending the signal towards the UAV. By simulating our proposed approach using MATLAB, we have demonstrated that utilizing RIS-assisted communication maximizes coverage between the Base Station and the UAV, outperforming the simulation results of coverage as a function of height without the use of RIS. The significance of this work lies in its ability to enhance the signal quality and coverage at cell edges by leveraging UAVs as intermediate relays. These UAVs serve the purpose of connecting users with weak or no links, effectively bridging the gap. In our simulation results, we employed RIS to strengthen the backhaul link quality between the UAVs and base stations. While our work successfully addresses the challenges of connectivity and coverage, it is important to note that we have not specifically focused on the cost aspect of these factors.

Dimensionality reduction is an essential ingredient of machine learning modelling that seeks to improve the performance of such models by extracting better quality features from data while removing irrelevant and redundant ones. The technique aids reduce computational load, avoiding data over-fitting, and increasing model interpretability. Recent studies have revealed that dimensionality reduction can benefit from labeled information, through joint approximation of predictors and target variables from a low-rank representation. A multiplicity of linear and non-linear dimensionality reduction techniques are proposed in the literature contingent on the nature of the domain of interest. This paper presents an evaluation of the performance of a hybrid deep learning model using feature extraction techniques while being applied to a benchmark network intrusion detection dataset. We compare the performance of linear and non-linear feature extraction methods namely, the Principal Component Analysis and Isometric Feature Mapping respectively. The Principal Component Analysis is a non-parametric classical method normally used to extract a smaller representative dataset from high-dimensional data and classifies data that is linear in nature while preserving spatial characteristics. In contrast, Isometric Feature Mapping is a representative method in manifold learning that maps high-dimensional information into a lower feature space while endeavouring to maintain the neighborhood for each data point as well as the geodesic distances present among all pairs of data points. These two approaches were applied to the CICIDS 2017 network intrusion detection benchmark dataset to extract features. The extracted features were then utilized in the training of a hybrid deep learning-based intrusion detection model based on convolutional and a bi-direction long short term memory architecture and the model performance results were compared. The empirical results demonstrated the dominance of the Principal Component Analysis as compared to Isometric Feature Mapping in improving the performance of the hybrid deep learning model in classifying network intrusions. The suggested model attained 96.97% and 96.81% in overall accuracy and F1-score, respectively, when the PCA method was used for dimensionality reduction. The hybrid model further achieved a detection rate of 97.91% whereas the false alarm rate was reduced to 0.012 with the discriminative features reduced to 48. Thus the model based on the principal component analysis extracted salient features that improved detection rate and reduced the false alarm rate.