In an era dominated by the Internet of Things (IoT), security and privacy remain paramount concerns. However, implementing robust cryptographic protocols on resource-constrained IoT devices presents significant challenges. This study introduces a novel approach that leverages quantum computing to generate post-quantum cryptographic keys securely. We evaluated the feasibility of using a third-party quantum computer to generate cryptographic keys for IoT devices, ensuring strong security without overburdening device resources. Using the unique advantages of quantum computing, our solution mitigates the threats posed by quantum attacks on classical cryptographic systems. We validate our approach using CRYSTALS-Kyber and McEliece Key Encapsulation Mechanisms (KEMs). To the best of our knowledge, this method has not been previously implemented. Our findings suggest promising avenues for future research in integrating quantum computing for secure IoT communications.

5G is expected to support numerous Internet of Things (IoT) devices. However, the inherent vulnerabilities and limited resources of IoT devices make them susceptible to compromise and exploitation, potentially leading to Distributed Denial of Service (DDoS) attacks on 5G infrastructure from within. While conventional Intrusion Detection Systems (IDS) can assist, 5G’s unique protocols, such as the General Packet Radio Service (GPRS) Tunneling Protocol User Plane (GTP-U), pose challenges due to the inability to analyze packet headers. Therefore, we propose using Software-Defined Networking (SDN), Machine Learning (ML), and programmable switches that utilize the Programming Protocol-independent Packet Processors (P4) language to analyze GTP traffic on the fly for DDoS attack detection. Our framework enhances the efficiency of DDoS attack detection and mitigation, as demonstrated through evaluations on an actual 5G testbed using real datasets. Compared to an alternative solution that forwards GTP packets to the SDN controller, our method significantly reduces attack detection time while enhancing throughput on the SDN switch.

Stacking of multiple Machine Learning (ML) classifiers have gained popularity in addressing anomalous data classification along with Deep Learning (DL) algorithms. This study compares traditional ML classifiers, multi-layer stacking ML classifiers, and DL classifiers using an open-source malware dataset-containing equal numbers of benign and malware samples. The results on the realistic dataset indicate that the DL classifier, utilizing a Bidirectional Long Short-Term Memory (BiLSTM) model, outperformed the stacked classifiers with Logistic Regression (LR) and Support Vector Machine (SVM) as Meta learners by 36.78% and 39.69%, respectively, in terms of classification accuracy and performance. The research work was extended to study the impact of Generative Adversarial Network (GAN) based synthetic dataset of relatively smaller size on deep learning models. It was observed that the Deep Learning Multi-Layer Perceptron (DLMLP) Model had relatively superior performance as compared to complex deep learning models like Long Short-Term Memory LSTM and BiLSTM

Developing lightweight algorithms to implement DoS attack mitigation on edge devices is a growing interest in edge cybersecurity. Various types of micro-controller boards can be programmed to capture network traffic and implement lightweight machine learning models to analyze the supplied traffic data for signs of intrusion and attacks. This study experimented with building Support Vector Machine and Logistic Regression models on real-time DoS attack scenario data and the CICIoT2023 dataset. The main contribution of this study is to propose a framework for data capturing, processing, and analysis to produce edge machine learning models for DoS attack mitigation,

5G networks denote a revolutionary improvement in wireless communication by introducing three service grades: Enhanced Mobile Broadband (eMBB), Ultra-Reliable Low Latency Communications (URLLC), and Massive Machine Type Communications (mMTC). These three service grades represent a cost-efficient solution and enhanced user experience with higher data rates and lower latency. However, at the same time, these aspects can benefit attackers (e.g., by leveraging the support for mMTC) to launch various attacks effectively. mMTC comes with a massive number of unattended Internet of Things (IoT) devices known for having low-security capabilities. One of the biggest security concerns related to IoT is that it increases the chances of internal DDoS attacks, which can disrupt 5G core network services. In this paper, we propose our ongoing work on monitoring the GPRS Tunneling Protocol User Plane (GTP-U) traffic, which is used to transport user data from User Equipment (UE) devices. We offer internal traffic filtering mechanisms using Software Defined Networks (SDN) to block the IoT traffic that appears to be malicious. The proposed approach is implemented in a 5G testbed to evaluate the performance and efficiency of factual scenarios.

In the field of machine learning, Stochastic Gradient Descent has been proven an effective method to shorten the time spent on minimizing the output cost. Due to the fact that the data pattern of each mini-batch could be somewhat varied from the full dataset, most existing optimization algorithms attempt to alleviate this variance via computing certain calibration terms associated with the previous gradient. Since the previous gradient is computed from the past Stochastic Gradient Descent state, it can adversely affect the calibration terms introducing deviation that result in an inaccurate new gradient. To resolve this problem, we propose a method that reduces the aforementioned deviation via applying a preprocessing technique to the previous gradient prior to its usage. The technique uses the angle between the previous and the current gradients to improve the precision of the calibration terms, reducing the effects of the deviations. Empirical results are obtained from incorporating the proposed method with a fully-connected vanilla neural network. The proposed technique is evaluated using the MNIST dataset against 10 previously proposed optimization algorithms. The experiment shows the benefits in reducing the cost result from adopting the proposed preprocessing technique to improve the gradient derivation.

The rapid evolution of technology in our sociality has brought great advantages, but at the same time it has increased the cybersecurity threats associated with a constantly expanded surface exposure. At the forefront of these threats is the proliferation of malware from traditional computing platforms to the rapidly expanding Internet-of-things. Our premise is that malware needs to be detected as soon as possible to prevent extensive consequence linked to the fulfillment of its malicious activity. Our proposed framework analyzes task structure features as well as the system calls and memory access patterns made by a process to determine its validity and integrity. The proposed scheme would use all three approaches applying an ensemble technique to detect malware.

In today's software and systems environments, security frameworks and models are evolving exponentially. Many traditional host-based frameworks are currently available to detect cyber threats in Linux environment. But there have been many challenges in detecting rootkits that modify the Linux Operating System (OS) kernel to avoid detection. These limitations have lead us to design a virtualization framework for detection of cyber threats in Linux environment. Instead of relying on the Linux Operating System kernel which is now a common victim of cyber-attacks, this virtualization framework will rely on the virtual machine hypervisor which is a more secure software layer that runs the OS kernel and the hardware. The paper proposed a virtualization framework based on well-known hypervisors, to detect cyber threats. The proposed work allowed for a more robust cyber threat detection method than traditional host-based frameworks. It can also possess self-healing properties since it will not only detect compromised servers but also suspend their operation by replacing them with uncompromised versions. This innovative framework promises to secure large scale IT infrastructure with minimum maintenance cost.

Software Defined Networking (SDN) is a networking model that allows for greater dynamic control of a networking environment. With today's increasingly complex networking environment, SDN networks allow for a greater degree of control and flexibility of a network. This is accomplished through the separation of the control and data planes, as well as the implementation of a global programmable controller. A Network Intrusion Detection Systems (NIDS) can work very well with SDN networks as it can help monitor the overall security of a network by analyzing the network as a whole and making choices to defend the network based on data from the entire network. Using a Hidden Markov Model (HMM), a NIDS could monitor a network and learn from the evolving network activity of the present and react accordingly. This machine-learning NIDS could improve the efficiency of security applications and increases the range of activities that they are able to accomplish. In this paper we plan to demonstrate the possibility of using Hidden Markov models to develop an adaptive NIDS for use in the new emerging technology of SDN.

The emerging technology of Software-Defined Networking (SDN) affords a platform and architecture which is dynamic, manageable, cost-effective, and adaptable, making it ideal for many applications that are high-bandwidth and dynamic in nature. As this technology grows and matures, there is a need for cybersecurity applications to be designed, developed and evaluated. In this paper, we propose a development environment configuration to build security applications targeting the SDN-controller in an effort to explore the technology and the resources available to teach and research the platform from a security application development perspective.

The technology exists for the migration of healthcare data from its archaic paper-based system to an electronic one, and, once in digital form, to be transported anywhere in the world in a matter of seconds. The advent of universally accessible healthcare data has benefited all participants, but one of the outstanding problems that must be addressed is how the creation of a standardized nationwide electronic healthcare record system in the United States would uniquely identify and match a composite of an individual's recorded healthcare information to an identified individual patients out of approximately 300 million people to a 1:1 match. To date, a few solutions to this problem have been proposed that are limited in their effectiveness. We propose the use of biometric technology within our fingerprint, iris, retina scan, and DNA (FIRD) framework, which is a multiphase system whose primary phase is a multilayer consisting of these four types of biometric identifiers: 1) fingerprint; 2) iris; 3) retina scan; and 4) DNA. In addition, it also consists of additional phases of integration, consolidation, and data discrepancy functions to solve the unique association of a patient to their medical data distinctively. This would allow a patient to have real-time access to all of their recorded healthcare information electronically whenever it is necessary, securely with minimal effort, greater effectiveness, and ease.

Abstract As the Internet revolution moves into full swing, those countries that have not embraced e‐commerce technology will face new hurdles as they seek to develop their economies. Standing in the path of these countries' attempts to adapt e‐commerce technologies are several key issues that can be broadly defined as trust barriers. Rather than think of the trust issues as barriers one must think of them as assets. Presents a conceptual model and framework that highlight the key factors in business trust relationships within developing countries; information security, technical and industrial infrastructure, education, government, and socio‐cultural factors. These factors are considered in the light of different types of e‐commerce business transactions taking place within and across borders such as business‐to‐business (B2B), business‐to‐consumer (B2C), consumer‐to‐business (C2B), and consumer‐to‐consumer (C2C).

In the past decade, the business community has embraced the capabilities of the Internet for a multitude of services that involve access to data and information. Of particular concern to these businesses have been the protection and authentication of digital data as it is distributed electronically. This paper proposes a novel approach that combines the reactive rule-based scheme of an active database management system (ADBMS) with the technology of digital watermarking to automatically protect digital data. The ADMBS technology facilitates the establishment of Event-Condition-Action (ECA) rules that define the actions to be triggered by events under certain conditions. These actions consist of the generation of unique watermarks and the tagging of digital data with unique signatures. Watermarking is a technology that embeds, within the digital data’s context, information identifying its owner and/or creator. The integration of these two technologies is a powerful mechanism for protecting digital data in a consistent and formal manner with applications in e-business in establishing and authenticating the ownership of images, audio, video, and other digital materials.

The design and development of real-time systems is often a difficult and time-consuming task. System realization has become increasingly difficult due to the proliferation of larger and more complex applications. To offset some of these difficulties, real-time developers have turned to object-oriented methodology. The success of object-oriented concepts in the development of non-real-time programs motivates the relevance of these concepts to achieve similar gains from encapsulation and code reuse in the real-time domain. This article presents an approach of integrating real-time constraint specifications within the constructs of an object-oriented language, affording these constraints a status equivalent to other language elements. This has led to the definition of such novel concepts as temporal abstract classes, virtual temporal constraints, and temporal specification inheritance, which extends inheritance mechanisms to accommodate real-time constraint specifications. These extensions provide real-time developers with the ability to manage and maintain the temporal behavior of a real-time program in a comparable manner to its functional behavior.