arXiv subjects

Moayad Aloqaily

Publications and source records attributed to Moayad Aloqaily.

At least 19 recordsLinked to original sources

arXiv2026

ChronosAudio: A Comprehensive Long-Audio Benchmark for Evaluating Audio-Large Language Models

Although Audio Large Language Models (ALLMs) have witnessed substantial advancements, their long audio understanding capabilities remain unexplored. A plethora of benchmarks have been proposed for general audio tasks, they predominantly focus on short-form clips, leaving without a consensus on evaluating ALLMs over extended durations. This paper proposes ChronosAudio, the first multi-task benchmark tailored for long-audio understanding in ALLMs. It encompasses six major task categories and comprises 36,000 test instances totaling over 200 hours audio, stratified into short, middle, and long-form categories to comprehensively evaluate length generalization. Extensive experiments on 16 state-of-the-art models using ChronosAudio yield three critical findings: 1.Precipitous Long-Context Collapse: ALLMs exhibit a severe inability to sustain performance, with the transition from short to long contexts triggering a staggering performance degradation of over 90% in specific tasks. 2.Structural Attention Dilution: Performance degradation stems from a fundamental failure in maintaining temporal locality; attention mechanisms suffer from significant diffusion in later sequences. 3.Restorative Ceiling of Mitigation: Current strategies only offer 50% recovery. These findings reveal significant challenges in long-audio, underscoring the urgent need for approaches to achieve robust, document-level audio reasoning.

cs.SD↗

arXiv2026

Backdoor Collapse: Eliminating Unknown Threats via Known Backdoor Aggregation in Language Models

Backdoor attacks are a significant threat to large language models (LLMs), often embedded via public checkpoints, yet existing defenses rely on impractical assumptions about trigger settings. To address this challenge, we propose \ourmethod, a defense framework that requires no prior knowledge of trigger settings. \ourmethod is based on the key observation that when deliberately injecting known backdoors into an already-compromised model, both existing unknown and newly injected backdoors aggregate in the representation space. \ourmethod leverages this through a two-stage process: \textbf{first}, aggregating backdoor representations by injecting known triggers, and \textbf{then}, performing recovery fine-tuning to restore benign outputs. Extensive experiments across multiple LLM architectures demonstrate that: (I) \ourmethod reduces the average Attack Success Rate to 4.41\% across multiple benchmarks, outperforming existing baselines by 28.1\%$\sim$69.3\%$\uparrow$. (II) Clean accuracy and utility are preserved within 0.5\% of the original model, ensuring negligible impact on legitimate tasks. (III) The defense generalizes across different types of backdoors, confirming its robustness in practical deployment scenarios.

cs.CL↗

arXiv2026

SafeSeek: Universal Attribution of Safety Circuits in Language Models

Mechanistic interpretability reveals that safety-critical behaviors (e.g., alignment, jailbreak, backdoor) in Large Language Models (LLMs) are grounded in specialized functional components. However, existing safety attribution methods struggle with generalization and reliability due to their reliance on heuristic, domain-specific metrics and search algorithms. To address this, we propose \ourmethod, a unified safety interpretability framework that identifies functionally complete safety circuits in LLMs via optimization. Unlike methods focusing on isolated heads or neurons, \ourmethod introduces differentiable binary masks to extract multi-granular circuits through gradient descent on safety datasets, while integrates Safety Circuit Tuning to utilize these sparse circuits for efficient safety fine-tuning. We validate \ourmethod in two key scenarios in LLM safety: \textbf{(1) backdoor attacks}, identifying a backdoor circuit with 0.42\% sparsity, whose ablation eradicates the Attack Success Rate (ASR) from 100\% $\to$ 0.4\% while retaining over 99\% general utility; \textbf{(2) safety alignment}, localizing an alignment circuit with 3.03\% heads and 0.79\% neurons, whose removal spikes ASR from 0.8\% $\to$ 96.9\%, whereas excluding this circuit during helpfulness fine-tuning maintains 96.5\% safety retention.

cs.LG↗

arXiv2026

MCPShield: A Security Cognition Layer for Adaptive Trust Calibration in Model Context Protocol Agents

The Model Context Protocol (MCP) standardizes tool use for LLM-based agents and enable third-party servers. This openness introduces a security misalignment: agents implicitly trust tools exposed by potentially untrusted MCP servers. However, despite its excellent utility, existing agents typically offer limited validation for third-party MCP servers. As a result, agents remain vulnerable to MCP-based attacks that exploit the misalignment between agents and servers throughout the tool invocation lifecycle. In this paper, we propose MCPShield as a plug-in security cognition layer that mitigates this misalignment and ensures agent security when invoking MCP-based tools. Drawing inspiration from human experience-driven tool validation, MCPShield assists agent forms security cognition with metadata-guided probing before invocation. Our method constrains execution within controlled boundaries while cognizing runtime events, and subsequently updates security cognition by reasoning over historical traces after invocation, building on human post-use reflection on tool behavior. Experiments demonstrate that MCPShield exhibits strong generalization in defending against six novel MCP-based attack scenarios across six widely used agentic LLMs, while avoiding false positives on benign servers and incurring low deployment overhead. Overall, our work provides a practical and robust security safeguard for MCP-based tool invocation in open agent ecosystems.

cs.CR↗

arXiv2026

HearSay Benchmark: Do Audio LLMs Leak What They Hear?

While Audio Large Language Models (ALLMs) have achieved remarkable progress in understanding and generation, their potential privacy implications remain largely unexplored. This paper takes the first step to investigate whether ALLMs inadvertently leak user privacy solely through acoustic voiceprints and introduces $\textit{HearSay}$, a comprehensive benchmark constructed from over 22,000 real-world audio clips. To ensure data quality, the benchmark is meticulously curated through a rigorous pipeline involving automated profiling and human verification, guaranteeing that all privacy labels are grounded in factual records. Extensive experiments on $\textit{HearSay}$ yield three critical findings: $\textbf{Significant Privacy Leakage}$: ALLMs inherently extract private attributes from voiceprints, reaching 92.89% accuracy on gender and effectively profiling social attributes. $\textbf{Insufficient Safety Mechanisms}$: Alarmingly, existing safeguards are severely inadequate; most models fail to refuse privacy-intruding requests, exhibiting near-zero refusal rates for physiological traits. $\textbf{Reasoning Amplifies Risk}$: Chain-of-Thought (CoT) reasoning exacerbates privacy risks in capable models by uncovering deeper acoustic correlations. These findings expose critical vulnerabilities in ALLMs, underscoring the urgent need for targeted privacy alignment. The codes and dataset are available at https://github.com/JinWang79/HearSay_Benchmark

cs.CL↗

arXiv2025

Generative AI for Networking

Generative Artificial Intelligence (GenAI) and Large Language Models (LLMs) are revolutionizing network management systems, paving the way towards fully autonomous and self-optimizing communication systems. These models enable networks to address complex decision-making tasks across both short-term operational scenarios and long-term strategic planning. Through natural language understanding, LLMs can analyze customer inquiries, predict network congestion patterns, and automate troubleshooting processes, leading to more efficient customer support and network maintenance. GenAI can optimize content delivery by generating personalized recommendations, improving user engagement, and dynamically adjusting network resources based on real-time demands, ultimately enhancing overall performance and user experience in telecommunication services. In this paper, we discuss the pivotal role of GenAI in advancing network performance and achieving the ultimate objective of self-adaptive networks. Moreover, we present a use case that leverages the self-attention mechanism of transformers to perform long-term traffic prediction. Harnessing these cutting-edge technologies demonstrates the transformative power of LLM and GenAI in revolutionizing telecommunication networks, elevating resilience and adaptability to unprecedented levels.

cs.NI↗

arXiv2025

Backdoor Attribution: Elucidating and Controlling Backdoor in Language Models

Fine-tuned Large Language Models (LLMs) are vulnerable to backdoor attacks through data poisoning, yet the internal mechanisms governing these attacks remain a black box. Previous research on interpretability for LLM safety tends to focus on alignment, jailbreak, and hallucination, but overlooks backdoor mechanisms, making it difficult to understand and fully eliminate the backdoor threat. In this paper, aiming to bridge this gap, we explore the interpretable mechanisms of LLM backdoors through Backdoor Attribution (BkdAttr), a tripartite causal analysis framework. We first introduce the Backdoor Probe that proves the existence of learnable backdoor features encoded within the representations. Building on this insight, we further develop Backdoor Attention Head Attribution (BAHA), efficiently pinpointing the specific attention heads responsible for processing these features. Our primary experiments reveals these heads are relatively sparse; ablating a minimal \textbf{$\sim$ 3%} of total heads is sufficient to reduce the Attack Success Rate (ASR) by \textbf{over 90%}. More importantly, we further employ these findings to construct the Backdoor Vector derived from these attributed heads as a master controller for the backdoor. Through only \textbf{1-point} intervention on \textbf{single} representation, the vector can either boost ASR up to \textbf{$\sim$ 100% ($\uparrow$)} on clean inputs, or completely neutralize backdoor, suppressing ASR down to \textbf{$\sim$ 0% ($\downarrow$)} on triggered inputs. In conclusion, our work pioneers the exploration of mechanistic interpretability in LLM backdoors, demonstrating a powerful method for backdoor control and revealing actionable insights for the community.

cs.CR↗

arXiv2025

ED-DAO: Energy Donation Algorithms based on Decentralized Autonomous Organization

Energy is a fundamental component of modern life, driving nearly all aspects of daily activities. As such, the inability to access energy when needed is a significant issue that requires innovative solutions. In this paper, we propose ED-DAO, a novel fully transparent and community-driven decentralized autonomous organization (DAO) designed to facilitate energy donations. We analyze the energy donation process by exploring various approaches and categorizing them based on both the source of donated energy and funding origins. We propose a novel Hybrid Energy Donation (HED) algorithm, which enables contributions from both external and internal donors. External donations are payments sourced from entities such as charities and organizations, where energy is sourced from the utility grid and prosumers. Internal donations, on the other hand, come from peer contributors with surplus energy. HED prioritizes donations in the following sequence: peer-sourced energy (P2D), utilitygrid-sourced energy (UG2D), and direct energy donations by peers (P2PD). By merging these donation approaches, the HED algorithm increases the volume of donated energy, providing a more effective means to address energy poverty. Experiments were conducted on a dataset to evaluate the effectiveness of the proposed method. The results showed that HED increased the total donated energy by at least 0.43% (64 megawatts) compared to the other algorithms (UG2D, P2D, and P2PD).

cs.DC↗

arXiv2024

Analysis of Blockchain Assisted Energy Sharing Algorithms with Realistic Data Across Microgrids

With escalating energy demands, innovative solutions have emerged to supply energy affordably and sustainably. Energy sharing has also been proposed as a solution, addressing affordability issues while reducing consumers' greed. In this paper, we analyse the feasibility of two energy sharing algorithms, centralized and peer-to-peer, within two scenarios, between microgrids within a county, and between microgrids across counties. In addition, we propose a new sharing algorithm named Selfish Sharing, where prosumers take advantage of consumers' batteries in return for letting them consume part of the shared energy. The results for sharing between microgrids across counties show that the dependency on the grid could be reduced by approximately 5.72%, 6.12%, and 5.93% using the centralized, peer-to-peer and selfish sharing algorithms respectively, compared to trading only. The scenario of sharing between microgrids within a county has an average decrease in dependency on the grid by 5.66%, 6.0%, and 5.80% using the centralized, peer-to-peer and selfish algorithms respectively, compared to trading without sharing. We found that trading with batteries and the proposed sharing algorithms prove to be beneficial in the sharing between microgrids case. More specifically, the case of trading and sharing energy between microgrids across counties outperforms sharing within a county, with P2P sharing appearing to be superior.

cs.DC↗

arXiv2024

Blockchain-enabled Energy Trading and Battery-based Sharing in Microgrids

Carbon footprint reduction can be achieved through various methods, including the adoption of renewable energy sources. The installation of such sources, like photovoltaic panels, while environmentally beneficial, is cost-prohibitive for many. Those lacking photovoltaic solutions typically resort to purchasing energy from utility grids that often rely on fossil fuels. Moreover, when users produce their own energy, they may generate excess that goes unused, leading to inefficiencies. To address these challenges, this paper proposes innovative blockchain-enabled energy-sharing algorithms that allow consumers -- without financial means -- to access energy through the use of their own energy storage units. We explore two sharing models: a centralized method and a peer-to-peer (P2P) one. Our analysis reveals that the P2P model is more effective, enhancing the sharing process significantly compared to the centralized method. We also demonstrate that, when contrasted with traditional battery-supported trading algorithm, the P2P sharing algorithm substantially reduces wasted energy and energy purchases from the grid by 73.6%, and 12.3% respectively. The proposed system utilizes smart contracts to decentralize its structure, address the single point of failure concern, improve overall system transparency, and facilitate peer-to-peer payments.

cs.DC↗

arXiv2023

Self-Sovereign Identity for Trust and Interoperability in the Metaverse

With the advancement in computing power and speed, the Internet is being transformed from screen-based information to immersive and extremely low latency communication environments in web 3.0 and the Metaverse. With the emergence of the Metaverse technology, more stringent demands are required in terms of connectivity such as secure access and data privacy. Future technologies such as 6G, Blockchain, and Artificial Intelligence (AI) can mitigate some of these challenges. The Metaverse is now on the verge where security and privacy concerns are crucial for the successful adaptation of such disruptive technology. The Metaverse and web 3.0 are to be decentralized, anonymous, and interoperable. Metaverse is the virtual world of Digital Twins and non-fungible tokens (NFTs). The control and possession of users' data on centralized servers are the cause of numerous security and privacy concerns. This paper proposes a solution for the security and interoperability challenges using Self-Sovereign Identity (SSI) integrated with blockchain. The philosophy of Self-Sovereign Identity, where the users are the only holders and owners of their identity, comes in handy to solve the questions of decentralization, trust, and interoperability in the Metaverse. This work also discusses the vision of a single, open standard, trustworthy, and interoperable Metaverse with initial design and implementation of SSI concepts.

cs.CR↗

arXiv2023

Harris Hawks Feature Selection in Distributed Machine Learning for Secure IoT Environments

The development of the Internet of Things (IoT) has dramatically expanded our daily lives, playing a pivotal role in the enablement of smart cities, healthcare, and buildings. Emerging technologies, such as IoT, seek to improve the quality of service in cognitive cities. Although IoT applications are helpful in smart building applications, they present a real risk as the large number of interconnected devices in those buildings, using heterogeneous networks, increases the number of potential IoT attacks. IoT applications can collect and transfer sensitive data. Therefore, it is necessary to develop new methods to detect hacked IoT devices. This paper proposes a Feature Selection (FS) model based on Harris Hawks Optimization (HHO) and Random Weight Network (RWN) to detect IoT botnet attacks launched from compromised IoT devices. Distributed Machine Learning (DML) aims to train models locally on edge devices without sharing data to a central server. Therefore, we apply the proposed approach using centralized and distributed ML models. Both learning models are evaluated under two benchmark datasets for IoT botnet attacks and compared with other well-known classification techniques using different evaluation indicators. The experimental results show an improvement in terms of accuracy, precision, recall, and F-measure in most cases. The proposed method achieves an average F-measure up to 99.9\%. The results show that the DML model achieves competitive performance against centralized ML while maintaining the data locally.

cs.LG↗

arXiv2023

An Incremental Gray-box Physical Adversarial Attack on Neural Network Training

Neural networks have demonstrated remarkable success in learning and solving complex tasks in a variety of fields. Nevertheless, the rise of those networks in modern computing has been accompanied by concerns regarding their vulnerability to adversarial attacks. In this work, we propose a novel gradient-free, gray box, incremental attack that targets the training process of neural networks. The proposed attack, which implicitly poisons the intermediate data structures that retain the training instances between training epochs acquires its high-risk property from attacking data structures that are typically unobserved by professionals. Hence, the attack goes unnoticed despite the damage it can cause. Moreover, the attack can be executed without the attackers' knowledge of the neural network structure or training data making it more dangerous. The attack was tested under a sensitive application of secure cognitive cities, namely, biometric authentication. The conducted experiments showed that the proposed attack is effective and stealthy. Finally, the attack effectiveness property was concluded from the fact that it was able to flip the sign of the loss gradient in the conducted experiments to become positive, which indicated noisy and unstable training. Moreover, the attack was able to decrease the inference probability in the poisoned networks compared to their unpoisoned counterparts by 15.37%, 14.68%, and 24.88% for the Densenet, VGG, and Xception, respectively. Finally, the attack retained its stealthiness despite its high effectiveness. This was demonstrated by the fact that the attack did not cause a notable increase in the training time, in addition, the Fscore values only dropped by an average of 1.2%, 1.9%, and 1.5% for the poisoned Densenet, VGG, and Xception, respectively.

cs.CR↗

arXiv2023

Decentralized Zero-Trust Framework for Digital Twin-based 6G

The Sixth Generation (6G) network is a platform for the fusion of the physical and virtual worlds. It will integrate processing, communication, intelligence, sensing, and storage of things. All devices and their virtual counterparts will become part of the service-provisioning process. In essence, 6G is a purposefully cooperative network that heavily depends on the capabilities of edge and end-devices. Digital Twin (DT) will become an essential part of 6G, not only in terms of providing a virtual representation of the physical elements and their dynamics and functionalities but rather DT will become a catalyst in the realization of the cooperative 6G environment. DT will play a main role in realizing the full potential of the 6G network by utilizing the collected data at the cyber twin and then implementing using the physical twin to ensure optimal levels of accuracy and efficiency. With that said, such a cooperative non-conventional network infrastructure cannot rely on conventional centralized intrusion detection and prevention systems. Zero-trust is a new security framework that aims at protecting distributed data, devices, components and users. This article presents a new framework that integrates the zero-trust architecture in DT-enabled 6G networks. Unlike conventional zero-trust solutions, the proposed framework adapts a decentralized mechanism to ensure the security, privacy and authenticity of both the physical devices and their DT counterparts. Blockchain plays an integral part in the authentication of DTs and the communicated data. Artificial Intelligence (AI) is integrated into all cooperating nodes using meta, generalized and federated learning solutions. The article also discusses current solutions and future outlooks, with challenges and some technology enablers.

cs.CY↗

arXiv2022

Energy Pricing in P2P Energy Systems Using Reinforcement Learning

The increase in renewable energy on the consumer side gives place to new dynamics in the energy grids. Participants in a microgrid can produce energy and trade it with their peers (peer-to-peer) with the permission of the energy provider. In such a scenario, the stochastic nature of distributed renewable energy generators and energy consumption increases the complexity of defining fair prices for buying and selling energy. In this study, we introduce a reinforcement learning framework to help solve this issue by training an agent to set the prices that maximize the profit of all components in the microgrid, aiming to facilitate the implementation of P2P grids in real-life scenarios. The microgrid considers consumers, prosumers, the service provider, and a community battery. Experimental results on the \textit{Pymgrid} dataset show a successful approach to price optimization for all components in the microgrid. The proposed framework ensures flexibility to account for the interest of these components, as well as the ratio of consumers and prosumers in the microgrid. The results also examine the effect of changing the capacity of the community battery on the profit of the system. The implementation code is available \href{https://github.com/Artifitialleap-MBZUAI/rl-p2p-price-prediction}{here}.

cs.LG↗

arXiv2022

Integrating Digital Twin and Advanced Intelligent Technologies to Realize the Metaverse

The advances in Artificial Intelligence (AI) have led to technological advancements in a plethora of domains. Healthcare, education, and smart city services are now enriched with AI capabilities. These technological advancements would not have been realized without the assistance of fast, secure, and fault-tolerant communication media. Traditional processing, communication and storage technologies cannot maintain high levels of scalability and user experience for immersive services. The metaverse is an immersive three-dimensional (3D) virtual world that integrates fantasy and reality into a virtual environment using advanced virtual reality (VR) and augmented reality (AR) devices. Such an environment is still being developed and requires extensive research in order for it to be realized to its highest attainable levels. In this article, we discuss some of the key issues required in order to attain realization of metaverse services. We propose a framework that integrates digital twin (DT) with other advanced technologies such as the sixth generation (6G) communication network, blockchain, and AI, to maintain continuous end-to-end metaverse services. This article also outlines requirements for an integrated, DT-enabled metaverse framework and provides a look ahead into the evolving topic.

cs.HC↗

arXiv2022

Internet of Behavior (IoB) and Explainable AI Systems for Influencing IoT Behavior

Pandemics and natural disasters over the years have changed the behavior of people, which has had a tremendous impact on all life aspects. With the technologies available in each era, governments, organizations, and companies have used these technologies to track, control, and influence the behavior of individuals for a benefit. Nowadays, the use of the Internet of Things (IoT), cloud computing, and artificial intelligence (AI) have made it easier to track and change the behavior of users through changing IoT behavior. This article introduces and discusses the concept of the Internet of Behavior (IoB) and its integration with Explainable AI (XAI) techniques to provide trusted and evident experience in the process of changing IoT behavior to ultimately improving users' behavior. Therefore, a system based on IoB and XAI has been proposed in a use case scenario of electrical power consumption that aims to influence user consuming behavior to reduce power consumption and cost. The scenario results showed a decrease of 522.2 kW of active power when compared to original consumption over a 200-hours period. It also showed a total power cost saving of 95.04 Euro for the same period. Moreover, decreasing the global active power will reduce the power intensity through the positive correlation.

cs.DC↗

arXiv2022

SFCaaS: Service Function Chains as a Service in NFV Environments

With the emergence of network softwarization trend, traditional networking services offered by Internet providers are expected to evolve by fully leveraging new recent technologies like network function virtualization and software defined networking. In this paper, we investigate offering Service Function Chains as a Service (SFCaaS) in NFV Environments. We first describe the potential business model to offer such a service. We then conduct a detailed study of the costs of virtual machine instances offered by Amazon EC2 with respect to the location, instance size, and performance in order to guide service chain provisioning and resource allocation. Afterwards, we address the resource allocation problem for service chain functions from the SFC provider's perspective while leveraging the performed cost study. We hence formulate the problem as an Integer Linear Program (ILP) aiming at reducing the SFC provider's operational costs of virtual machine instances and links as well as the synchronization costs among the instances. We also propose a new heuristic algorithm to solve the mapping problem with the same aforementioned goals taking into account the conducted study of the costs of Amazon EC2 instances. We show through extensive simulations that the proposed heuristic significantly reduce operational costs compared to a Baseline algorithm inspired by the existing literature.

cs.NI↗