AAAbderazek Ben Abdallah

Abderazek Ben Abdallah

*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


Abderazek Ben Abdallah has a total of 106 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Convergence in Neuromorphic Systems: From Circuit Innovation to Adaptive Cognition

    by Abderazek Ben Abdallah, Joshua Yang, Jason Eshraghian, Huajin Tang

  • Edge-Driven Dynamic Two-Tier Blockchain for Energy Trading in Vehicle-to-Grid Networks

    by Zhishang Wang, Yuxiao Liang, Achraf Ben Ahmed, Khanh N. Dang, Abderazek Ben Abdallah

    Recent developments in vehicle-to-grid (V2G) technology have positioned electric vehicles (EVs) as essential components for increasing the use of renewable energy and managing peak demand. However, while V2G systems utilize blockchain technology for secure energy transactions, they encounter notable inefficiencies due to the high volume of transactions when the number of energy participants increases. The communication overhead associated with processing energy supply and demand is significant. In addition, current energy distribution systems do not fully exploit the potential of EVs. They cannot schedule multiple discharge cycles for a single vehicle over specific periods. Furthermore, existing systems do not provide mechanisms that allow EVs to make smart autonomous decisions about their charging and discharging operations. Instead, they are usually dependent on fixed schedules or manual inputs, which further exacerbates the inefficiency of energy management and market integration. This paper proposes a novel edge-driven two-tier blockchain-based V2G method to optimize the energy management of distributed electric vehicles. The architecture incorporates local dual networks within an electric vehicle blockchain (EVBC) and an energy market blockchain (EMBC), coordinated by a high-level blockchain that manages unfulfilled requests. A dynamic segment-based energy allocation algorithm (DSBEA) is introduced, where the control system allocates energy requests by matching EV offers while continuously updating each EV's remaining energy and available time window after each assignment. The evaluation showed a significant reduction in total time cost compared to baseline methods, with reductions ranging from 42% to 80% in various energy trading scenarios. In addition, the proposed method achieved a 68.6% increase in energy fulfillment compared to the best existing approaches.

  • ApproxiMorph : Energy-Efficient Neuromorphic System With Layer-Wise Approximation of Spiking Neural Networks and 3-D-Stacked SRAM

    by Ryoji Kobayashi, Ngo-Doanh Nguyen, Abderazek Ben Abdallah, Nguyen Anh Vu Doan, Khanh N. Dang

    This article proposes ApproxiMorph, a comprehensive framework for both software and hardware co-design, targeting energy-efficient artificial intelligence (AI) applications using 3-D-IC-based neuromorphic systems. By leveraging parallel interconnections and high-bandwidth communications inherent to 3-D-ICs, and the noise-resilience characteristics of spiking neural networks (SNNs), ApproxiMorph achieves significant power savings by exploiting 1) approximate implementation of neuron cells; 2) layer-wise approximation of SNNs through the heuristic exploration algorithm; 3) reduced-voltage operation in the 3-D-stacked SRAM; and 4) incorporating a weight-tuning method. As a result, to search for the energy-optimal layer-wise approximation, ApproxiMorph explores only 0.44%–0.67% of all possible combinations, achieving a 28.06% power saving for additions with a 0.60% accuracy loss in comparison to the baseline SNN for MNIST. In the VGG16 for CIFAR-10, ApproxiMorph searches around $10^{3}$ combinations from over $10^{17}$ possible solutions, resulting in a 29.16% power saving with slight accuracy gain. Furthermore, integrating all methods enhances the accuracy of approximate implementations and demonstrates higher error resilience than accurate implementations.


  • NOMA: A Novel Reliability Improvement Methodology for 3D IC-Based Neuromorphic Systems

    by Ngo-Doanh Nguyen, Khanh N. Dang, Akram Ben Ahmed, Abderazek Ben Abdallah, Xuan-Tu Tran

    Hardware is the foundation of trust for all real-time applications and is especially crucial in Artificial Intelligence (AI). The growing reliance on AI for various tasks has raised concerns about the reliability of the hardware. Extensive research has shown that hardware faults arising from the manufacturing process or device variation can significantly impact the precision and accuracy of AI applications. This also applies to 3-D IC (Three-Dimensional Integrated Circuit)-based Spiking Neural Networks (SNNs), despite their noise resilience, low-power, and low memory footprint advantages. This is because, in addition to the mentioned hardware faults, thermal dissipation on 3-D ICs can negatively affect the memory stacked on upper dies. Hence, this article proposes a methodology called NOMA (Network-of-Memory Architecture) to enhance the reliability of 3-D IC-based SNNs by replacing the defective critical synaptic weights in high-priority layers with lower-priority ones. The method is assessed on multiple memory-on-logic architectures with various memory technologies such as SRAM, eDRAM, STT-RAM, and RRAM. As a result, for the 45nm SRAM library, the accuracy degradation improves by roughly 56% at a BER of 0.05 with a timing overhead of 3.538μs and a power overhead of 390.796 nJ .


  • Open Access

    BTSAM: Balanced Thermal-State-Aware Mapping Algorithms and Architecture for 3D-NoC-Based Neuromorphic Systems

    by Mohamed Maatar, Zhishang Wang, Khanh N. Dang, Abderazek Ben Abdallah

    Neuromorphic computing systems are biologically inspired approaches created from many highly connected neurons to model neuroscience theories and solve machine learning problems. They promise to drastically improve the efficiency of critical computational tasks such as decision-making and perception. Combining neuromorphic computing systems and 3D interconnect (3D-NoC) technology leads to an advanced architecture that inherits the benefits of both computing and interconnect paradigms. However, designing large-scale neuromorphic systems based on 3D-NoC faces several challenges, including thermal power, power distribution, increased power density at the heat sink interface, and fabrication requirements. This work tackles the thermal issues in designing large-scale neuromorphic systems by proposing a Balanced Thermal-State-Aware Mapping (BTSAM) for 3D-NoC-based neuromorphic systems. This includes a Periodic Activity Scoring (PAS), a Seesaw Neuron Clustering (SNC) method, and a thermal-aware genetic algorithm to eliminate hotspots, balance the thermal state, and lower the temperature while keeping the system’s accuracy acceptable. Evaluation results on various system configurations demonstrate a notable up to 12.4 K and 5.2 K temperature reduction compared to linear methods and HeterGenMap, respectively, and a 4× increase in Mean-Time-to-Failure (MTTF), with an acceptable power and area overheads and little degradation of the communication cost.


  • Cluster-Based Thermal-Aware Mapping for 3D-NoC-Based NASH Neuromorphic System

    by Mohamed Maatar, Khanh N. Dang, Abderazek Ben Abdallah

    Neuromorphic computing, inspired by the brain's neural architecture, offers a path towards energy-efficient and high-performance computing by mimicking spiking neurons and synaptic activities. The expansion of neuromorphic systems, how-ever, is hampered by issues related to system scalability, energy efficiency, and processing speed. To overcome these obstacles, the adoption of $3\mathrm{D}$ integrated circuits ( $3\mathrm{D}$ -ICs) and $3\mathrm{D}$ network-on-chip-based (3D-NoC) architectures has been suggested. These technologies facilitate enhanced system scalability, lower energy consumption, and minimized interconnect lengths, leading to improved performance. However, the integration of 3D-IC in neuromorphic systems introduces thermal challenges due to increased power density and uneven heat distribution across the stacked layers. This paper presents a cluster-based thermal-aware mapping method based on genetic algorithm that aims to minimize highest and the variance of the temperature across the system. The results of the proposed method show that our method has up to $39^{\mathrm{o}}\mathrm{C}$ lower peak temperature than linear mapping methods with a trade-off of 7. $5\mathrm{x}$ communication cost. Therefore, our method results in a lower temperature with a moderate thermal state across the layers.


  • Open Access

    A Hybrid Clustered Approach for Enhanced Communication and Model Performance in Blockchain-Based Collaborative Learning

    by Zhishang Wang, Masayuki Hisada, Abderazek Ben Abdallah

    Collaborative edge learning has emerged in various domains like vehicular networks and medical care, allowing local model training on edge devices while preserving privacy. The integration of blockchain technology further enhances security and privacy in the learning environment, although challenges such as communication overhead and vulnerability to poisoning attacks have yet to be resolved. This paper proposes a hybrid clustered blockchain method (HCB) for collaborative edge learning, which introduces a unique combination of cluster-based model updates and the use of delegate nodes for efficient model aggregation. We introduce the delegate-based adaptive model aggregation for robust collaborative learning, termed DAMA-RCL, and a novel disassembling-reassembling method for practical model transmission on the blockchain network. Experimental results demonstrate that HCB significantly enhances communication efficiency, reducing hop counts and transmission time by over 90%, while maintaining learning performance compared to traditional collaborative learning. Additionally, DAMA-RCL exhibits strong resilience in scenarios with up to 50% malicious clients. The HCB approach for collaborative edge learning, along with the DAMA-RCL algorithm and the model disassembling-reassembling method, provides a promising solution to the challenges of communication efficiency, privacy, and security, paving the way for effective and reliable collaborative learning in various application domains.


  • Open Access

    Robust Vehicle-to-Grid Energy Trading Method Based on Smart Forecast and Multi-Blockchain Network

    by Yuxiao Liang, Zhishang Wang, Abderazek Ben Abdallah

    In the present era, energy issues are a significant concern, and the energy trading market is the crucial sector to facilitate supply-demand balance and sustainable development. For better demand response and grid balancing, vehicle-to-grid (V2G) technology is rapidly gaining importance in energy markets. To narrow the gap between ideal V2G goals and actual applications needs, energy trading system has to overcome the challenges of over-centralized structure, inflexible timeline adaptation, limited market scale and energy efficiency, excessive feedback time costs, and low rate of economic return. To address these issues and ensure a secure energy market, we propose a decentralized intelligent V2G system called V2G Forecasting and Trading Network (V2GFTN) to achieve efficient and robust energy trading in campus EV networks. A multiple blockchain structure is proposed in V2GFTN to ensure trading security and data privacy between energy requests and offers. V2GFTN also integrates energy forecasting functions for EVs with a smart energy trading and EV allocation mechanism called SRET so that the EVs with driving tasks can supply their extra power back to the grid and achieve higher energy efficiency and economic profit. Through rigorous experimentation and compared with equivalent studies, V2GFTN system has demonstrated higher economic profit and energy demand fill rate by up to 1.6 times and 1.9 times than the state-of-the-art V2G approaches.


  • Open Access

    HeterGenMap: An Evolutionary Mapping Framework for Heterogeneous NoC-Based Neuromorphic Systems

    by Khanh N. Dang, Nguyen Anh Vu Doan, Ngo-Doanh Nguyen, Abderazek Ben Abdallah

    While task mapping for multi-core systems is known as an NP-hard problem, mapping for neuromorphic systems even scale it up due to a high number of neurons per core and a high number of core per system. Moreover, mapping for neuromorphic systems also has several challenges such as heterogeneous computing core or communication fabrics, and potential defects in neurons or routing units. Therefore, this paper presents a genetic algorithm framework named HeterGenMap which is a Genetic Algorithm framework for mapping multiple-layer Spiking Neural Network systems to solve the aforementioned problems. The results show that HeterGenMap improves the overall communication cost by 11.04-26.77% in comparison to the linear mapping. Moreover, under link faulty scenarios, neuron defects, or multi-chip designs, HeterGenMap can reduce the communication cost by 3.41-31.34%, 7.01%-41.51%, and 34.21-45.56% in comparison to the linear approach, respectively. The validation in hardware also demonstrated that HeterGenMap reduces the inference time by 63.10-77.87% from the linear mapping.