Hussain M. J. Almohri
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Hussain M. J. Almohri has a total of 15 co-authors. Here are the top co-authors by number of shared publications:
- David Evans
University of Virginia University of Virginia University of Virginia
3 shared publications - Layne T. Watson
Virginia Polytechnic Institute and State University
3 shared publications - Karim Elish
Department of Computer Science, Florida Polytechnic University, Lakeland, USA
2 shared publications Layne T. Watson
Virginia Polytechnic Institute and State University, Blacksburg, VA, USA
2 shared publications- Layne T. Watson
Departments of Computer Science, Mathematics, and Aerospace and Ocean Engineering, Virginia Polytechnic Institute & State University, Blacksburg, VA 24061
2 shared publications Mohammad Almutawa
Kuwait University, Kuwait
2 shared publications
Latest Publications
- Open Access
Alohomora: Workflow-Aware Authentication and Authorization in Heterogeneous Systems
Current federated identity management systems lack contextual awareness of workflows across independent systems, creating security gaps and workflow integrity challenges. This article details the design and implementation of Alohomora, a distributed workflow-aware authentication system that maintains cross-system workflow context through path-bound tokens. Alohomora complements existing identity providers such as OAuth and SAML by adding workflow orchestration capabilities while leveraging standard authentication protocols for initial user verification. The system introduces workflow graphs as a formal model for representing dependencies between functions across heterogeneous systems and employs a distributed caching architecture with collaboration groups for scalable session management. In a typical deployment scenario, an employee onboarding workflow across human resources services, account provisioning, and benefits systems forms a trust group where Alohomora enforces ordered step execution, validates prerequisite completion at each transition, and generates cryptographic completion assertions upon workflow finalization. Extensive performance evaluation under concurrent user requests demonstrates polynomial performance characteristics with superior scalability compared to centralized OAuth introspection. The results show that Alohomora maintains high throughput under heavy load while providing strong, secure access control through workflow path binding and distributed trust orchestration. The prototype implementation is available as open source.
Open AccessAlohomora: Workflow-Aware Authentication and Authorization in Heterogeneous Systems
Current federated identity management systems lack contextual awareness of workflows across independent systems, creating security gaps and workflow integrity challenges. This article details the design and implementation of Alohomora, a distributed workflow-aware authentication system that maintains cross-system workflow context through path-bound tokens. Alohomora complements existing identity providers such as OAuth and SAML by adding workflow orchestration capabilities while leveraging standard authentication protocols for initial user verification. The system introduces workflow graphs as a formal model for representing dependencies between functions across heterogeneous systems and employs a distributed caching architecture with collaboration groups for scalable session management. In a typical deployment scenario, an employee onboarding workflow across human resources services, account provisioning, and benefits systems forms a trust group where Alohomora enforces ordered step execution, validates prerequisite completion at each transition, and generates cryptographic completion assertions upon workflow finalization. Extensive performance evaluation under concurrent user requests demonstrates polynomial performance characteristics with superior scalability compared to centralized OAuth introspection. The results show that Alohomora maintains high throughput under heavy load while providing strong, secure access control through workflow path binding and distributed trust orchestration. The prototype implementation is available as open source.
- Open Access
- 6 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 92019-92035
FedQP: Large-Scale Private and Flexible Federated Query Processing
State-of-the-art federated learning methods orchestrate iterations of the stochastic gradient descent algorithm among a network of clients to refine a unified set of model parameters, all while safeguarding individual client datasets. Although promising, the success of existing approaches hinges on using a uniform data model with the same optimization and learning algorithms by the participants. In addition, existing methods can be subject to privacy attacks, such as model-inversion attacks. A pivotal challenge lies in crafting an algorithm that can efficiently aggregate client inputs, ensuring minimal client dataset exposure and maximum flexibility in model adaptation. We propose a new scheme for aggregating local, diverse, and independent data models, addressing data inconsistency and model-inversion attacks. The proposed scheme, federated query processing FedQP, enables clients to build local models independently, without coordinating with the server or other clients. The server communicates with selected clients only to predict future values. Our scheme only requires one communication round between each client and the server for model initialization without the need for a training phase. Clients do not share data or model parameters. We evaluated FedQP by assessing the impact of our new aggregation mechanism on model accuracy and the communication efficacy of our architecture. We show that the caching algorithm in FedQP reduces redundant client communications during query processing. In addition, we present a privacy analysis that shows that our method outperforms prominent gradient-based approaches to federated learning. Our experiments show that FedQP consistently achieves higher classification accuracy in non-IID settings and demonstrates stronger resilience to reconstruction attacks compared to gradient-based methods such as federated averaging (FedAvg).
- 6 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 92019-92035
Open AccessFedQP: Large-Scale Private and Flexible Federated Query Processing
State-of-the-art federated learning methods orchestrate iterations of the stochastic gradient descent algorithm among a network of clients to refine a unified set of model parameters, all while safeguarding individual client datasets. Although promising, the success of existing approaches hinges on using a uniform data model with the same optimization and learning algorithms by the participants. In addition, existing methods can be subject to privacy attacks, such as model-inversion attacks. A pivotal challenge lies in crafting an algorithm that can efficiently aggregate client inputs, ensuring minimal client dataset exposure and maximum flexibility in model adaptation. We propose a new scheme for aggregating local, diverse, and independent data models, addressing data inconsistency and model-inversion attacks. The proposed scheme, federated query processing FedQP, enables clients to build local models independently, without coordinating with the server or other clients. The server communicates with selected clients only to predict future values. Our scheme only requires one communication round between each client and the server for model initialization without the need for a training phase. Clients do not share data or model parameters. We evaluated FedQP by assessing the impact of our new aggregation mechanism on model accuracy and the communication efficacy of our architecture. We show that the caching algorithm in FedQP reduces redundant client communications during query processing. In addition, we present a privacy analysis that shows that our method outperforms prominent gradient-based approaches to federated learning. Our experiments show that FedQP consistently achieves higher classification accuracy in non-IID settings and demonstrates stronger resilience to reconstruction attacks compared to gradient-based methods such as federated averaging (FedAvg).
- Open Access
- 28 June 2024
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Mathematical Software
- Vol. 50 (2) , 1-8
Remark on Algorithm 1012: Computing Projections with Large Datasets
In ACM TOMS Algorithm 1012, the DELAUNAYSPARSE software is given for performing Delaunay interpolation in medium to high dimensions. When extrapolating outside the convex hull of the training set, DELAUNAYSPARSE calls the nonnegative least squares solver DWNNLS to compute projections onto the convex hull. However, DWNNLS and many other available sum-of-squares optimization solvers were not intended for usage with many variable problems, which result from the large training sets that are typical in machine learning applications. Thus, a new PROJECT subroutine is given, based on the highly customizable quadratic program solver BQPD . This solution is shown to be as robust as DELAUNAYSPARSE for projection onto both synthetic and real-world datasets, where other available solvers frequently fail. Although it is intended as an update for DELAUNAYSPARSE , due to the difficulty and prevalence of the problem, this solution is likely to be of external interest as well.
- 28 June 2024
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Mathematical Software
- Vol. 50 (2) , 1-8
Open AccessRemark on Algorithm 1012: Computing Projections with Large Datasets
In ACM TOMS Algorithm 1012, the DELAUNAYSPARSE software is given for performing Delaunay interpolation in medium to high dimensions. When extrapolating outside the convex hull of the training set, DELAUNAYSPARSE calls the nonnegative least squares solver DWNNLS to compute projections onto the convex hull. However, DWNNLS and many other available sum-of-squares optimization solvers were not intended for usage with many variable problems, which result from the large training sets that are typical in machine learning applications. Thus, a new PROJECT subroutine is given, based on the highly customizable quadratic program solver BQPD . This solution is shown to be as robust as DELAUNAYSPARSE for projection onto both synthetic and real-world datasets, where other available solvers frequently fail. Although it is intended as an update for DELAUNAYSPARSE , due to the difficulty and prevalence of the problem, this solution is likely to be of external interest as well.
- Open Access
- 30 June 2022
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Autonomous and Adaptive Systems
- Vol. 17 (1-2) , 1-23
Dynamic System Diversification for Securing Cloud-based IoT Subnetworks
Remote exploitation attacks use software vulnerabilities to penetrate through a network of Internet of Things (IoT) devices. This work addresses defending against remote exploitation attacks on vulnerable IoT devices. As an attack mitigation strategy, we assume it is not possible to fix all the vulnerabilities and propose to diversify the open-source software used to manage IoT devices. Our approach is to deploy dynamic cloud-based virtual machine proxies for physical IoT devices. Our architecture leverages virtual machine proxies with diverse software configurations to mitigate vulnerable and static software configurations on physical devices. We develop an algorithm for selecting new configurations based on network anomaly detection signals to learn vulnerable software configurations on IoT devices, automatically shifting towards more secure configurations. Cloud-based proxy machines mediate requests between application clients and vulnerable IoT devices, facilitating a dynamic diversification system. We report on simulation experiments to evaluate the dynamic system. Two models of powerful adversaries are introduced and simulated against the diversified defense strategy. Our experiments show that a dynamically diversified IoT architecture can be invulnerable to large classes of attacks that would succeed against a static architecture.
- 30 June 2022
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Autonomous and Adaptive Systems
- Vol. 17 (1-2) , 1-23
Open AccessDynamic System Diversification for Securing Cloud-based IoT Subnetworks
Remote exploitation attacks use software vulnerabilities to penetrate through a network of Internet of Things (IoT) devices. This work addresses defending against remote exploitation attacks on vulnerable IoT devices. As an attack mitigation strategy, we assume it is not possible to fix all the vulnerabilities and propose to diversify the open-source software used to manage IoT devices. Our approach is to deploy dynamic cloud-based virtual machine proxies for physical IoT devices. Our architecture leverages virtual machine proxies with diverse software configurations to mitigate vulnerable and static software configurations on physical devices. We develop an algorithm for selecting new configurations based on network anomaly detection signals to learn vulnerable software configurations on IoT devices, automatically shifting towards more secure configurations. Cloud-based proxy machines mediate requests between application clients and vulnerable IoT devices, facilitating a dynamic diversification system. We report on simulation experiments to evaluate the dynamic system. Two models of powerful adversaries are introduced and simulated against the diversified defense strategy. Our experiments show that a dynamically diversified IoT architecture can be invulnerable to large classes of attacks that would succeed against a static architecture.
- 14 January 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computers
- Vol. 71 (11) , 2982-2995
Lightweight, Effective Detection and Characterization of Mobile Malware Families
Android malware is an ongoing threat to billions of smart devices’ security, ranging from mobile phones to car infotainment systems. Despite numerous approaches and previous studies to develop solutions for detecting and preventing Android malware, the rapid continuous development of new malware variants requires a careful reconsideration and the development of effective methods to identify malware families given a meager number of malware instances. In this paper, we present DroidMalVet, a novel Android malware family classification and detection approach that does not require to perform complex program analyses or utilize large feature sets. DroidMalVet is the first to use a promising, diverse, and small set of software metrics as features in a supervised learning platform to classify and detect various Android malware families. Our extensive empirical evaluations on two large public malware datasets show that DroidMalVet accurately detects both small and large malware families with F-Score accuracy of 94.4% and 96%, and AUC equal to 99.5% and 99.7% on the malware families in Drebin and AMD datasets, respectively. Moreover, our results demonstrate the superior performance of DroidMalVet in detecting small families (i.e., families with few samples). DroidMalVet complements existing approaches and presents an early warning tool for detecting known and emerging malware families.
- 14 January 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computers
- Vol. 71 (11) , 2982-2995
Lightweight, Effective Detection and Characterization of Mobile Malware Families
Android malware is an ongoing threat to billions of smart devices’ security, ranging from mobile phones to car infotainment systems. Despite numerous approaches and previous studies to develop solutions for detecting and preventing Android malware, the rapid continuous development of new malware variants requires a careful reconsideration and the development of effective methods to identify malware families given a meager number of malware instances. In this paper, we present DroidMalVet, a novel Android malware family classification and detection approach that does not require to perform complex program analyses or utilize large feature sets. DroidMalVet is the first to use a promising, diverse, and small set of software metrics as features in a supervised learning platform to classify and detect various Android malware families. Our extensive empirical evaluations on two large public malware datasets show that DroidMalVet accurately detects both small and large malware families with F-Score accuracy of 94.4% and 96%, and AUC equal to 99.5% and 99.7% on the malware families in Drebin and AMD datasets, respectively. Moreover, our results demonstrate the superior performance of DroidMalVet in detecting small families (i.e., families with few samples). DroidMalVet complements existing approaches and presents an early warning tool for detecting known and emerging malware families.
- Open Access
- 13 April 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 9, 58824-58837
On Parallel Real-Time Security Improvement Using Mixed-Integer Programming
Network security defenses evolve, responding to real-time attack incidents, modifying the underlying topology, or reallocating defense systems across the network. The present work emphasizes reducing the time to compute new optimal reallocations of defense systems, responding to emerging real-time remote attacks. The proposed heuristic method utilizes parallel processing by slicing the underlying graphical model representing the network topology, solving in parallel multiple mixed-integer programming problems corresponding to the created subgraphs, and producing an estimate of the optimal defense. The parallelized method to compute a new defense enables producing a response, in real-time, before remote attackers compromise a target machine in the network. Our prototype tool to compute a new defense, the high-performance security analyzer, has a speedup of at least 20 over solving the original problem using a serial algorithm, and with an insignificant difference between the performance of the (computed in parallel) approximately optimal defense and the (serially computed) optimal defense. A major conclusion is that further speedups will come from parallel integer programming algorithms rather than from graph partitioning.
- 13 April 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 9, 58824-58837
Open AccessOn Parallel Real-Time Security Improvement Using Mixed-Integer Programming
Network security defenses evolve, responding to real-time attack incidents, modifying the underlying topology, or reallocating defense systems across the network. The present work emphasizes reducing the time to compute new optimal reallocations of defense systems, responding to emerging real-time remote attacks. The proposed heuristic method utilizes parallel processing by slicing the underlying graphical model representing the network topology, solving in parallel multiple mixed-integer programming problems corresponding to the created subgraphs, and producing an estimate of the optimal defense. The parallelized method to compute a new defense enables producing a response, in real-time, before remote attackers compromise a target machine in the network. Our prototype tool to compute a new defense, the high-performance security analyzer, has a speedup of at least 20 over solving the original problem using a serial algorithm, and with an insignificant difference between the performance of the (computed in parallel) approximately optimal defense and the (serially computed) optimal defense. A major conclusion is that further speedups will come from parallel integer programming algorithms rather than from graph partitioning.
- Open Access
- 23 July 2020
- Published by Hindawi Limited in Security and Communication Networks
- Vol. 2020, 1-12
A Client Bootstrapping Protocol for DoS Attack Mitigation on Entry Point Services in the Cloud
This paper presents a client bootstrapping protocol for proxy-based moving target defense system for the cloud. The protocol establishes the identity of prospective clients who intend to connect to web services behind obscure proxy servers in a cloud-based network. In client bootstrapping, a set of initial line of defense services receive new client requests, execute an algorithm to assign them to a proxy server, and reply back with the address of the chosen proxy server. The bootstrapping protocol only reveals one proxy address to each client, maintaining the obscurity of the addresses for other proxy servers. Hiding the addresses of proxy servers aims to lower the likelihood that a proxy server becomes the victim of a denial-of-service (DoS) attack. Existing works address this problem by requiring the solution of computationally intensive puzzles from prospective clients. This solution slows the progression of attacks as well as new clients. This paper presents an alternative idea by observing that limited capacity of handling initial network requests is the primary cause of denial-of-service attacks. Thus, the suggested alternative is to utilize cost-effective high-capacity networks to handle client bootstrapping, thus thwarting attacks on the initial line of defense. The prototype implementation of the protocol using Google's firebase demonstrates the proof of concept for web services that receive network requests from clients on mobile devices.
- 23 July 2020
- Published by Hindawi Limited in Security and Communication Networks
- Vol. 2020, 1-12
Open AccessA Client Bootstrapping Protocol for DoS Attack Mitigation on Entry Point Services in the Cloud
This paper presents a client bootstrapping protocol for proxy-based moving target defense system for the cloud. The protocol establishes the identity of prospective clients who intend to connect to web services behind obscure proxy servers in a cloud-based network. In client bootstrapping, a set of initial line of defense services receive new client requests, execute an algorithm to assign them to a proxy server, and reply back with the address of the chosen proxy server. The bootstrapping protocol only reveals one proxy address to each client, maintaining the obscurity of the addresses for other proxy servers. Hiding the addresses of proxy servers aims to lower the likelihood that a proxy server becomes the victim of a denial-of-service (DoS) attack. Existing works address this problem by requiring the solution of computationally intensive puzzles from prospective clients. This solution slows the progression of attacks as well as new clients. This paper presents an alternative idea by observing that limited capacity of handling initial network requests is the primary cause of denial-of-service attacks. Thus, the suggested alternative is to utilize cost-effective high-capacity networks to handle client bootstrapping, thus thwarting attacks on the initial line of defense. The prototype implementation of the protocol using Google's firebase demonstrates the proof of concept for web services that receive network requests from clients on mobile devices.
- 14 May 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Information Forensics and Security
- Vol. PP (15566013) , 1
An Attack-Resilient Architecture for the Internet of Things
With current IoT architectures, once a single device in a network is compromised, it can be used to disrupt the behavior of other devices on the same network. Even though system administrators can secure critical devices in the network using best practices and state-of-the-art technology, a single vulnerable device can undermine the security of the entire network. The goal of this work is to limit the ability of an attacker to exploit a vulnerable device on an IoT network and fabricate deceitful messages to co-opt other devices. The approach is to limit attackers by using device proxies that are used to retransmit and control network communications. We present an architecture that prevents deceitful messages generated by compromised devices from affecting the rest of the network. The design assumes a centralized and trustworthy machine that can observe the behavior of all devices on the network. The central machine collects application layer data, as opposed to low-level network traffic, from each IoT device. The collected data is used to train models that capture the normal behavior of each individual IoT device. The normal behavioral data is then used to monitor the IoT devices and detect anomalous behavior. This paper reports on our experiments using both a binary classifier and a density-based clustering algorithm to model benign IoT device behavior with a realistic test-bed, designed to capture normal behavior in an IoT-monitored environment. Results from the IoT testbed show that both the classifier and the clustering algorithms are promising and encourage the use of applicationlevel data for detecting compromised IoT devices.
- 14 May 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Information Forensics and Security
- Vol. PP (15566013) , 1
An Attack-Resilient Architecture for the Internet of Things
With current IoT architectures, once a single device in a network is compromised, it can be used to disrupt the behavior of other devices on the same network. Even though system administrators can secure critical devices in the network using best practices and state-of-the-art technology, a single vulnerable device can undermine the security of the entire network. The goal of this work is to limit the ability of an attacker to exploit a vulnerable device on an IoT network and fabricate deceitful messages to co-opt other devices. The approach is to limit attackers by using device proxies that are used to retransmit and control network communications. We present an architecture that prevents deceitful messages generated by compromised devices from affecting the rest of the network. The design assumes a centralized and trustworthy machine that can observe the behavior of all devices on the network. The central machine collects application layer data, as opposed to low-level network traffic, from each IoT device. The collected data is used to train models that capture the normal behavior of each individual IoT device. The normal behavioral data is then used to monitor the IoT devices and detect anomalous behavior. This paper reports on our experiments using both a binary classifier and a density-based clustering algorithm to model benign IoT device behavior with a realistic test-bed, designed to capture normal behavior in an IoT-monitored environment. Results from the IoT testbed show that both the classifier and the clustering algorithms are promising and encourage the use of applicationlevel data for detecting compromised IoT devices.
- Open Access
An efficient deception architecture for cloud-based virtual networks
Emerging deceptive systems present a new promise for the uprising security problems in cloud-based virtual networks, especially those operated by small and medium enterprises. The main goal of deceptive systems is to form a layer of defensive nodes in an Internet-accessible cloud-based virtual network to distract and deceive malicious clients. While numerous approaches provide distinct models for developing decisive systems, misery digraphs present a promising decisive model for distracting powerful remote intrusions. Misery digraphs can delay access to targets deep in a cloud-based virtual network. A central challenge to the theory of misery digraphs is verifying their applicability in prominent cloud computing platforms as well as measuring the efficiency of networks that adapt them. Thus, architectural support is needed that can be realized with long-term support technologies and can be deployed for large networks. This work presents and analyzes a high-throughput architecture for misery digraphs, embarking on implementation details and performance analysis. Full implementation of the architecture in Amazon Web Services imposes modest performance delays in the request processing, while highly delaying stealth intrusions in the network.
Open AccessAn efficient deception architecture for cloud-based virtual networks
Emerging deceptive systems present a new promise for the uprising security problems in cloud-based virtual networks, especially those operated by small and medium enterprises. The main goal of deceptive systems is to form a layer of defensive nodes in an Internet-accessible cloud-based virtual network to distract and deceive malicious clients. While numerous approaches provide distinct models for developing decisive systems, misery digraphs present a promising decisive model for distracting powerful remote intrusions. Misery digraphs can delay access to targets deep in a cloud-based virtual network. A central challenge to the theory of misery digraphs is verifying their applicability in prominent cloud computing platforms as well as measuring the efficiency of networks that adapt them. Thus, architectural support is needed that can be realized with long-term support technologies and can be deployed for large networks. This work presents and analyzes a high-throughput architecture for misery digraphs, embarking on implementation details and performance analysis. Full implementation of the architecture in Amazon Web Services imposes modest performance delays in the request processing, while highly delaying stealth intrusions in the network.
- 24 June 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Information Forensics and Security
- Vol. 15 (15566013) , 500-511
Predictability of IP Address Allocations for Cloud Computing Platforms
One way to combat denial-of-service attacks on cloud-based virtual networks is to use unpredictable network addresses, aiming to increase attacker effort by requiring attackers to search a large IP address space to find a target host. IP address randomization is used by several moving target defenses, relying on the assumption that it is difficult for an attacker to predict newly allocated IP addresses. This paper analyzes whether IP addresses used by cloud providers are unpredictable enough in practice. We analyze the IP address allocation behaviors in two major cloud computing providers (Amazon Web Services and Google Cloud Platform) and find that the actual entropy provided by allocated IP addresses is limited. We evaluate several prediction models, including a simple frequency-based model as well as a Markov process model that produces an address prediction set from time series data of collected IP addresses. Our results show that simple models can reduce the search space for allocated IP addresses and diminish the effectiveness of randomization defenses.
- 24 June 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Information Forensics and Security
- Vol. 15 (15566013) , 500-511
Predictability of IP Address Allocations for Cloud Computing Platforms
One way to combat denial-of-service attacks on cloud-based virtual networks is to use unpredictable network addresses, aiming to increase attacker effort by requiring attackers to search a large IP address space to find a target host. IP address randomization is used by several moving target defenses, relying on the assumption that it is difficult for an attacker to predict newly allocated IP addresses. This paper analyzes whether IP addresses used by cloud providers are unpredictable enough in practice. We analyze the IP address allocation behaviors in two major cloud computing providers (Amazon Web Services and Google Cloud Platform) and find that the actual entropy provided by allocated IP addresses is limited. We evaluate several prediction models, including a simple frequency-based model as well as a Markov process model that produces an address prediction set from time series data of collected IP addresses. Our results show that simple models can reduce the search space for allocated IP addresses and diminish the effectiveness of randomization defenses.