SMSuhib Bani Melhem

Suhib Bani Melhem

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Suhib Bani Melhem has a total of 19 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Tiny Large Language Models for IoT networks: Potentials and challenges

    by Muhammed Golec, Suhib Bani Melhem, Yaser Khamayseh, Abdulmalik Alwarafy, Naofal Al-Dhahir


  • Open Access

    Tiny Large Language Models for IoT Networks: Potentials and Challenges

    by Muhammed Golec, Suhib Melhem, Yaser Khamayseh, Abdulmalik Alwarafy, Naofal Al-Dhahir

    Large Language Models (LLM), which have gained great momentum in recent years, have revolutionized the field of Artificial Intelligence (AI); while their applicability for hardware-constrained Internet of Things (IoT) environments has begun to be questioned. This has led to the emergence of compact architecture and resource-efficient Tiny LLM models. This survey paper systematically examines Tiny LLMs for IoT networks and classifies existing approaches in five basic dimensions: model architectures, optimization strategies, transfer learning methods, deployment paradigms, and explainability-security integration. By applying the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) method, 139 related studies published between 2020 and 2025 are analyzed to evaluate architectural adaptation, resource adaptation, and collaborative learning approaches. The findings reveal the most current fundamental technical gaps in the field and discuss the research directions of Tiny LLM in terms of generalizability, interpretability, and reliability in mission-critical IoT ecosystems.

  • Open Access

    EdgeAI-Powered Hybrid ESN-GRU Model for High-Accuracy and Efficient Short-Term Load Forecasting in Smart Grids

    by Suhib Bani Melhem, Muhammed Golec, Saed Alrabaee, Muath Alshaikh, Murat Uyar

    With the widespread use of renewable energy sources (RES) in the smart grid, the next generation power system, short-term load forecasting (STLF) is of critical importance in grid stability and energy optimization. Traditional STLF models include issues such as high computational cost, dependency on cloud infrastructure, and latency issues, which are undesirable for real-time energy management. To solve these issues, the EdgeAI paradigm, which combines edge computing and artificial intelligence (AI), can be a promising solution. EdgeAI reduces the dependency on cloud-based systems by processing data close to the data source, offering advantages such as lowlatency and lowbandwidth. Thus, it increases the response speed by processing data in real time, making it suitable for STLF applications. In order to benefit from all these advantages, the EdgeAI-driven Hybrid Echo State Network and Gated Recurrent Unit (ESN-GRU) model is introduced for real-time and efficient STLF in this paper. ESN-GRU combines the fast training capabilities of ESN and sequential learning capabilities of GRU, and offers the advantages of high prediction performance and low latency inference in STLF. Experimental results show that the proposed model improves the R² score by 2.5% and significantly reduces the MAPE value from 129 to 0.101 compared to existing models. Benchmark results in edge environment prove that ESN-GRU provides up to 79% and 92% faster inference compared to state-of-the-art methods.


  • Open Access

    LENS: Lightweight and Explainable LLM-Based APT Detection at the Edge for 6G Security

    by Suhib Bani Melhem, Muhammed Golec, Abdulmalik Alwarafy, Yaser Khamayseh

    Expected to be deployed in the early 2030s, sixth-generation (6G) wireless networks, with their high speed and integration with cutting-edge technology such as intelligent edge computing, expand the attack surface and face serious cyber threat risks such as Advanced Persistent Threats (APTs). This type of cyber attack can imitate benign network traffic and operate for long periods of time without being detected by traditional detection systems. This paper introduces LENS, a lightweight and explainable LLM-based network security framework designed to address this cybersecurity threat for 6G environments. LENS uses a fine-tuned DistilBERT model to convert raw network streams into natural language commands using contextual metadata and is trained on the CICAPT-IIoT (2024) dataset generated using real-time network traffic data. To evaluate the proposed model, adapted versions of DeepLog and EarlyCrow are compared using F1-score, false positive rate, and explainability metrics for binary APT classification on the CICAPT-IIoT dataset. All models are trained using a high-performance GPU (Nvidia A10) and validated by deploying on a real-world resource-constrained edge node (Raspberry Pi 4). The results confirm that LENS has higher performance in APT detection with 0.82 accuracy and 0.82 recall despite consuming higher energy compared to the other two baselines, and is applicable for edge-enabled 6G environments.


  • Open Access

    LLM-Driven APT Detection for 6G Wireless Networks: A Systematic Review and Taxonomy

    by Muhammed Golec, Yaser Khamayseh, Suhib Bani Melhem, Abdulmalik Alwarafy

    Sixth Generation (6G) wireless networks, which are expected to be deployed in the 2030s, have already created great excitement in academia and the private sector with their extremely high communication speed and low latency rates. However, despite the ultra-low latency, high throughput, and AI-assisted orchestration capabilities they promise, they are vulnerable to stealthy and long-term Advanced Persistent Threats (APTs). Large Language Models (LLMs) stand out as an ideal candidate to fill this gap with their high success in semantic reasoning and threat intelligence. In this paper, we present a comprehensive systematic review and taxonomy study for LLM-assisted APT detection in 6G networks. We address five research questions, namely, semantic merging of fragmented logs, encrypted traffic analysis, edge distribution constraints, dataset/modeling techniques, and reproducibility trends, by leveraging most recent studies on the intersection of LLMs, APTs, and 6G wireless networks. We identify open challenges such as explainability gaps, data scarcity, edge hardware limitations, and the need for real-time slicing-aware adaptation by presenting various taxonomies such as granularity, deployment models, and kill chain stages. We then conclude the paper by providing several research gaps in 6G infrastructures for future researchers. To the best of our knowledge, this paper is the first comprehensive systematic review and classification study on LLM-based APT detection in 6G networks.


  • Regression-Based Dynamic Provisioning and Monitoring for Responsive Resources in Cloud Infrastructure Networks

    by Mustafa Daraghmeh, Suhib Bani Melhem, Anjali Agarwal, Nishith Goel, Marzia Zaman

    Cloud computing model is the most complex computing model that requires implementing effective techniques to manage infrastructure resources of datacenters. Unproductive tasks scheduling can lead to an increase in the operational cost of cloud provider side, which in turn increases the cloud services cost at cloud consumer side. One of the effective techniques to address these issues in cloud datacenters is the elasticity by allowing dynamic resource provisioning based on the current demand and varying workload running upon virtual machines (VMs) over time. This leads to an increase in the resource utilization, and reduced power consumption by turning off the idle physical machines. However, the dynamic resource provisioning due to the growing service demand and higher quality of service requirements of the users can cause a violation of service level agreement. In this paper, we propose a model based on linear regression to manage and reformulate cloud users requests and dynamically generating rules based on historical data of their requests in order to update association functions to address and adapt the changes of different types of workloads running on the cloud provider datacenter. The experiments and simulation results based on dynamic workloads show the proposed algorithm significantly increases the resource utilization on cloud datacenter.


  • Open Access

    Markov Prediction Model for Host Load Detection and VM Placement in Live Migration

    by Suhib Bani Melhem, Anjali Agarwal, Nishith Goel, Marzia Zaman

    The design of good host overload/underload detection and VM placement algorithms play a vital role in assuring the smoothness of VM live migration. The presence of the dynamic environment that leads to a changing load on the VMs motivates us to propose a Markov prediction model to forecast the future load state of the host. We propose a host load detection algorithm to find the future overutilized/underutilized hosts state to avoid immediate VMs migration. Moreover, we propose a VM placement algorithm determine the set of candidates hosts to receive the migrated VMs in a way to reduce their VM migrations in near future. We evaluate our proposed algorithms through CloudSim simulation on different types of PlanetLab real and random workloads. The experimental results show that our proposed algorithms have a significant reduction in terms of SLA violation, number of VM migrations, and other metrics than the other competitive algorithms.


  • Minimizing biased VM selection in live VM migration

    by Suhib Bani Melhem, Anjali Agarwal, Nishith Goel, Marzia Zaman

    VM selection algorithm selects one or more VMs from the full set of VMs running on a given overload host, once a decision to migrate VMs from that host is made to achieve host/server consolidation and load balancing in cloud data centers while satisfying the QoS constraints. Presently, VM selection is a crucial decision for resource management in the cloud data center management, specially with high dynamic environment. In this paper, we propose two new VM selection algorithms, namely Minimum VM Migrated Count and Minimum migration time Minimum VM Migrated Count to avoid frequent SLA violation on the same VM. We propose new metrics to compare with other VM selection algorithms. We evaluate our proposed algorithms through CloudSim simulation on different types of PlanetLab real and random workloads. The experimental results demonstrate that the proposed algorithms show significant reduction in the Maximum number of VM migrated count and the degree of load balancing of VMs migrated count with the other state of the art algorithms.


  • Live VM Migration Across Cloud Data Centers

    by Suhib Bani Melhem, Anjali Agarwal, Mustafa Daraghmeh, Nishith Goel, Marzia Zaman

    Live VM migration is a technique that consists of a selection process and a migration process to migrate a VM from one host to another in the same data center without changing the IP address, or in a different data center with necessity to the VM to get a new IP address. The changing of IP address results into a mobility problem, which may render the service unreachable. In this paper, we propose a system model that selects data center randomly for VM placement while reducing this IP address reconfiguration. A new metric is proposed to indicate number of users that need IP reconfiguration. We extended CloudSim to simulate our work to identify the number of IP reconfigurations required for VM migration across the data centers on random workload.

  • A Markov-Based Prediction Model for Host Load Detection in Live VM Migration

    by Suhib Bani Melhem, Anjali Agarwal, Nishith Goel, Marzia Zaman

    Host load detection algorithm determines if a given host is overload or underloaded then the decision can be made to migrate VMs to achieve host/server consolidation and load balancing in cloud data centers while satisfying the QoS constraints. Presently, host load detection is a challenging problem in the cloud data center management specially with high dynamic environment for the host load. In this paper, we propose a novel Markov-based prediction algorithm to forecast the future load state of the host. The experimental results demonstrate that the proposed algorithm has better performance than the other competitive algorithms. The results for different types of PlanetLab real and random workloads show significant reduction of the SLA violation and the number of VM migrations.