Education
PhD. in Computer and Control Engineering, Politecnico di Torino, Italy
M.Sc. in Computer Engineering, Politecnico di Torino, Italy
B.Sc. in Computer Science, Virtual University, Pakistan
Research Interests
Information Security
Internet of Things Security and Reliability
Security Auditing and vulnerability analysis of IoT devices
Digital Forensics
Differential Privacy
Artificial Intelligence
Mobility modeling for UAVs
Software Defined Networking
Large Language Models
Selected Publications
1. Faheem H. B., Wali, A., Muzammil S., Mahmood, H., Sarfraz S, Asif H. (2026), "Advancing Text Summarization with Specialized Datasets: Computer Science and Geography Domains". Digital Scholarship in the Humanities, Accepted for Publication (2026)
2. Mahmood, H., Alamgir, Z., Javed, S.T., Karim, S., Awais, M. (2026), "Federated Generative Models in Medical Imaging:Current Advances, Challenges, and Future Directions", IEEE Access, 2026, Volume: 14, ISSN: 21693536, DOI: 10.1109/ACCESS.2026.3650810
3. Mahmood, H., Ather, S., Wali, A., A Ali, Malik, T.G., Kafeel W. (2025), "A Novel Fusion Approach with a Robust ParallelNet Model for Diabetic Retinopathy Diagnosis, PATTERN ANALYSIS AND APPLICATIONS", 2025, Volume: 28, Issue: 2, ISSN: 14337541, DOI: 10.1007/s10044-025-01448-3
4. Wang, Q. , Mahmmod, H. , Wali, A. , Saqib, M., Aziz, A. (2025), "Utilizing Augmented Reality to Bridge the Gap between Reading and Technology Usage in Children: A User Study", Interactive Learning Environments, 2025, ISSN: 10494820, DOI: 10.1080/10494820.2025.2597530
5. Mahmood H., Arshad M., Ahmed I., Fatima S.,ur Rehman H. (2024), "Comparative study of IoT forensic frameworks", Forensic Science International: Digital Investigation, 2024, Volume 49, 301748, ISSN 26662825, DOI:10.1016/j.fsidi.2024.301748
6. Mahmood, H., Iftikhar, M., Wali, A., Ali, A., Gulzar, M. (2024), "A Novel Cascaded Approach for Classification of Tuberculosis Using Cough Audio in Real-Time Environment", IEEE Access, 2024, Volume: 12, ISSN: 21693536, DOI: 10.1109/ACCESS.2024.3519296
7. Nadir, I., H. Mahmood, G. A. Shah (2022), “A taxonomy of IoT firmware security and principal firmware analysis techniques”, International Journal of Critical Infrastructure Protection, 2022, Volume 38, 100552, ISSN 1874-5482, DOI:10.1016/j.ijcip.2022.100552
8. Ahmed, F. , Mahmood, H. , Niaz, Y. (2021) "Mobility modelling for urban traffic surveillance by a team of unmanned aerial vehicles", International Journal of Ad Hoc and Ubiquitous Computing, 2021, ISSN: 17438225, DOI: 10.1504/IJAHUC.2021.113382
9. HU Rehman, I Bari, A Ali, H Mahmood (2017) "A Bayesian approach for estimating protein–protein interactions by integrating structural and non-structural biological data", Molecular BioSystems 13 (12), 2017, p. 2592-2602, DOI: 10.1039/c7mb00484b
10. H. Mahmood, M. Loghi, E. Macii, M. Poncino (2013) "Energy/Lifetime Cooptimization by Cache Partitioning With Graceful Performance Degradation", TVLSI: Transactions on Very Large Scale Integration Systems, September 2013, DOI 10.1109/TVLSI.2013.2278549
Teaching Courses
o Secure Programming
o Computer Forensics
o Risk Assessment and Management
o Emerging Technologies: IoT and Cloud
o Servers, Data Centers and Clouds
o Object Oriented Programming
o Discrete Structures
Expertise related to UN Sustainable Goals
In 2015, UN member states agreed to 17 Global Sustainable Goals (SDGs) to end poverty, protect the planet and ensure prosperity for all.
This person's work contributes towards the following SDGs.
Federated Generative Models in Medical Imaging: Current Advances, Challenges, and Future Directions
Published in: IEEE Access
Jan 05, 2026
Haroon Mahmood
Zareen Alamgir
Sobia Tariq Javed
Saira Karim
Muhammad Awais
The fusion of Federated Learning (FL) and deep generative models is transforming medical imaging by enabling privacy-preserving and data-efficient machine learning. Training large-scale deep models on radiological imaging data remains challenging due to data scarcity, heterogeneity, and strict privacy constraints that limit data sharing across institutions. FL addresses these challenges by employing collaborative model training that does not expose raw patient data, while generative models synthesize realistic medical images to alleviate scarcity and imbalance. The convergence of these two fields has given rise to federated generative models (FGMs), which extend Generative Adversarial Networks (GANs), Variational Autoencoders (VAE), and Diffusion models into distributed and privacy-preserving environments. FGMs have demonstrated significant potential in various medical imaging tasks, including data augmentation, image reconstruction, cross-modality conversion, and enhancing classification and segmentation models. Despite promising progress, the field remains in its early stages, facing open challenges in communication efficiency, scalability, data heterogeneity, and clinical reliability. This survey provides a comprehensive review of FGMs in medical imaging, analyzing their architectures, applications, and benchmark datasets, while highlighting key challenges and outlining future research directions to advance privacy-preserving generative modeling in healthcare.
Utilizing augmented reality to bridge the gap between reading and technology usage in children: a user study
Published in: Interactive Learning Environments
Dec 14, 2025
Qing Wang
Haroon Mahmood
Aamir Wali
Mirah Saqib
Ayishm Aziz
There has been a gradual decline in the number of children reading storybooks, as many now favor electronic gadgets for entertainment. This shift is reflected in the performance of school-going children such as their English composition and comprehension. Research indicates that around 80% of Pakistani students struggle with basic reading comprehension skills. As technology advances and changes how we interact with it, there is an opportunity to incorporate reading into children's screen time. This study explores the potential of augmented reality (AR) to make reading more engaging and interactive, encouraging children to read more. It investigates the effects of AR on reading motivation, engagement, and attitudes, while also identifying effective AR design elements that sustain children's interests. The ultimate goal is to develop an AR app that not only enhances the reading experience but also fosters regular reading habits in children, addressing the current issue of declining reading practices. This paper introduces “StoriesAR,” an Augmented Reality application designed to combat the growing concern over children's decreasing interest in reading. This study demonstrates that integrating narrative-based learning with AI-driven scaffolding not only improves learners' engagement and comprehension but also provides a replicable framework for incorporating digital storytelling into language pedagogy and learning technologies.
Advancing text summarization with specialized datasets: computer science and geography domains
Published in: Digital Scholarship in the Humanities
Nov 14, 2025
Huzaifa Bin Faheem
Aamir Wali
Saleha Muzammil
Haroon Mahmood
Summaira Sarfraz
Haya Asif
Automatic text summarization (ATS) has seen considerable development, but state-of-the-art models depend on vast training data, limiting their applicability in scholarly communication and digital knowledge management for technical and domain-specific texts where annotated datasets are scarce. This article addresses this critical gap. First, to create a controlled, data-constrained setting for study, we created two specialized datasets: CS-Summ and Geog-Summ, comprising 794 and 438 manually summarized excerpts from the Computer Science and Geography domains, respectively. We then propose and evaluate a simple, computationally inexpensive method of keyword-guided fine-tuning to enhance performance. By prepending extracted keywords, identified using spaCy’s NER model, to the source document as an explicit signal of salience, we guide the model’s attention during generation. Using these datasets, we conduct baseline experiments by keyword-guided fine-tuning on T5, BART, and BERT models and comparing their performance against zero-shot GPT-4. Evaluation using standard lexical metrics demonstrates that our keyword-guided approach consistently improves performance over the fine-tuned baselines. More critically, using factual consistency metrics such as BERTScore and NLI-based metrics for abstractive summarization, we show that keyword guidance reduces model hallucination and produces more reliable and factually accurate summaries. Our findings advance computational methods for technical content summarization and underscore the practical value of this method in the development of more robust and accessible summarization systems in resource-constrained environments, contributing to broader efforts to develop digital tools that support specialized scholarly analysis.
Optimizing Fetal Health Diagnosis: An Active Learning Framework with LightGBM
Published in: 2025 Sixth International Conference on Intelligent Data Science Technologies and Applications (IDSTA)
Sep 01, 2025
Sana Fatima
Mamoona Akbar
Haroon Mahmood
Anosha Khan
Fetal health classification is crucial for the timely identification of abnormalities and the improvement of neonatal care. Early prediction of fetal health is necessary to ensure a healthy pregnancy and lower rates of maternal and newborn mortality. Machine learning algorithms improve fetal health monitoring by enabling early detection of abnormalities and facilitating timely medical interventions. However, traditional machine learning models rely on large labeled datasets, which are often costly and time-consuming to obtain in medical applications. Active learning (AL) mitigates this challenge by strategically selecting the most informative samples for annotation, significantly reducing the labeling effort. In this paper, we present an active learning approach coupled with LightGBM (Gradient Boosting Machine) that achieves more than 99% classification accuracy with merely 16% of training data thus significantly reducing data annotation cost and effort.
A novel fusion approach with a robust ParallelNet model for diabetic retinopathy diagnosis: H. Mahmood et al.
Published in: Pattern Analysis and Applications
Mar 21, 2025
Haroon Mahmood
Saad Ather
Aamir Wali
Arshad Ali
Tayyaba Gul Malik
Wardah Kafeel
Diabetic Retinopathy (DR) is a serious diabetes-related complication that can lead to significant retinal damage and irreversible vision loss if not detected and treated early. While numerous deep learning algorithms have recently been developed for DR diagnosis, however they often focus on specific symptoms like exudates, vessels, or hemorrhages, overlooking a comprehensive analysis of all relevant indicators. Though, previous studies have shown high performance on benchmark public datasets but have struggled with real-time data. This paper introduces a diagnostic system that systematically incorporates all detectable symptoms of diabetic retinopathy and has demonstrated reliable performance on 108 test images from Lahore General Hospital, showcasing its robustness in real-world scenarios. Additionally, a novel algorithm for extracting retinal exudates is proposed, outperforming existing methods. The study categorizes retinal fundus images into both 2-class and multi-class diabetic retinopathy. Evaluation of current models on a local hospital dataset shows significant accuracy improvements. We also present ParallelNet, a model for classifying Diabetic Retinopathy stages: No DR, NPDR, PDR. ParallelNet outperforms established models, achieving 96% accuracy on the APTOS dataset and 90.16% on the local dataset for binary classification. For multi-classification, it achieves 90% accuracy on the APTOS dataset and 87.05% on the local dataset. These results highlight the improved performance achieved by combining our extraction algorithms with the ParallelNet model, demonstrating robustness across both public and local real-time hospital datasets.
Automated Authentication of Quranic Verses Using BERT (Bidirectional Encoder Representations from Transformers) based Language Models
Published in: Proceedings of the New Horizons in Computational Linguistics for Religious Texts
Mar 01, 2025
Khubaib Amjad Alam
Maryam Khalid
Syed Ahmed Ali
Haroon Mahmood
Qaisar Shafi
Muhammad Haroon
Zulqarnain Haider
The proliferation of Quranic content on digital platforms, including websites and social media, has brought about significant challenges in verifying the authenticity of Quranic verses. The inherent complexity of the Arabic language, with its rich morphology, syntax, and semantics, makes traditional text-processing techniques inadequate for robust authentication. This paper addresses this problem by leveraging state-of-the-art transformer-based Language models tailored for Arabic text processing. Our approach involves fine-tuning three transformer architectures BERT-Base-Arabic, AraBERT, and MarBERT on a curated dataset containing both authentic and non-authentic verses. Non-authentic examples were created using sentence-BERT, which applies cosine similarity to introduce subtle modifications. Comprehensive experiments were conducted to evaluate the performance of the models. Among the three candidate models, MarBERT, which is specifically designed for handling Arabic dialects demonstrated superior performance, achieving an F1-score of 93.80%. BERT-Base-Arabic also showed competitive F1 score of 92.90% reflecting its robust understanding of Arabic text. The findings underscore the potential of transformer-based models in addressing linguistic complexities inherent in Quranic text and pave the way for developing automated, reliable tools for Quranic verse authentication in the digital era.
A novel cascaded approach for classification of tuberculosis using cough audio in real-time environment
Published in: IEEE Access
Dec 17, 2024
Haroon Mahmood
Manal Iftikhar
Aamir Wali
Arshad Ali
Maryam Gulzar
Tuberculosis (TB) is an infectious disease primarily impacting the lungs. It spreads through the air when an infected person coughs, sneezes, or talks. Diagnosing TB involves clinical examinations and specialized tests performed by medical professionals. Coughing is a common symptom. The diagnosis of TB involves clinical examinations and specialized tests. However, studies have shown that medical doctors can distinguish between cough sounds associated with different respiratory conditions. Therefore, using artificial intelligence to analyze cough recordings of patients to diagnose TB is a promising research direction. In this study, we propose a customized cascaded approach for diagnosing TB using cough audio. This approach involves a series of models arranged in a sequence, where the output of one model serves as the input for the next. In the first phase, we distinguish between bursts in audio signal as noise or cough. In the second phase, we classify cough as TB or non-TB. Non-TB cough includes both voluntary and non-TB reflex cough. For this study we collected a dataset consisting of cough audio recordings from TB and non-TB patients at Mayo Hospital in Lahore, Pakistan. The recordings were obtained using the AI4LYF DCT application, a fully automated phone-based system, with no manual annotation. We apply statistical classifiers based on spectral and time domain features, both with and without clinical metadata. Through a stratified grouped cross-validation approach, our results show that using cough sounds along with demographic and clinical factors yielded an accuracy of 97% when the random forest was used. Similarly, for all other classifiers, the accuracies were ≥90 % when demographic and clinical data was included (from ≤80 )^. Our findings suggest that our model based on patient data and cough auido could support community health workers and health programs in identifying TB cases more effectively and cost-efficiently.
Optimized Coverage of Urban Territory for Traffic Surveillance using Multiple UAVs
Published in: Global Congress on Emerging Technologies (GCET-2024)
Dec 09, 2024
Saba Tariq
Haroon Mahmood
Farooq Ahmad
Aamir Wali
With ever increasing vehicles on the roads, traffic congestion is one of the major concerns of the modern world. Cities frequently experience heavy traffic, especially in developing countries due to lack of technology and automatic control systems. To address this issue, numerous models of traffic monitoring have been proposed by researchers. However, due to the static nature of these approaches, they are either inefficient or require a lot of resources and in turn a huge investment. Use of low-cost Unmanned aerial vehicles (UAVs) for urban traffic Surveillance is gaining increased attention in the research community in order to overcome existing limitations and provide a cost-effective and sustainable solution. A key challenge is to optimize the continuous coverage of large urban territories using a minimal number of UAVs. This research provides a comprehensive methodology to determine optimal resource requirements as well as positioning for improved coverage. Relying on a structural analysis of road network (represented as a graph) and using a combination of edge clustering and Silhouette-Score metric, the proposed method provides full coverage with a sufficient number of UAVs. Experimental results show that operational performance for continuous coverage is improved with reduced idle time.
Comparative study of IoT forensic frameworks
Published in: Forensic Science International: Digital Investigation
Jun 01, 2024
Haroon Mahmood
Maliha Arshad
Irfan Ahmed
Sana Fatima
Hafeez ur Rehman
Internet of Things (IoT) systems often consist of heterogeneous, resource-constrained devices that generate massive amounts of data. This data is important for assessments, behaviour analysis, and decision-making. However, IoT devices are also susceptible to cyber-attacks, such as information theft, personal device intervention, and privacy invasion. In case of an incident, these devices are subject to digital forensic investigation to identify and analyze crimes and misuse. Over the years, several forensic frameworks and techniques have been proposed to facilitate the investigation of IoT networks and devices, but finding a perfect solution that covers the diversity of IoT devices and networks is still a research challenge. In this study, we present a comparative analysis of existing forensic investigation frameworks and identify their strengths and weaknesses in handling forensic challenges of IoT devices. The study uses evaluation metrics of ten important parameters, including heterogeneity, scalability, and chain of custody, to thoroughly audit the effectiveness of these models. Our analysis concludes that the existing investigation frameworks do not cater to all requirements and aspects of IoT forensics. It further highlights the need for standard mechanisms to acquire and analyze digital artifacts in IoT devices.
An ensemble approach for iot firmware strength analysis using stride threat modeling and reverse engineering
Published in: 2022 24th International Multitopic Conference (INMIC)
Oct 21, 2022
Muhammad Shaharyar Yaqub
Haroon Mahmood
Ibrahim Nadir
Ghalib Asadullah Shah
Internet of Things (IoT) market is growing exponentially and automated smart solutions are revolutionizing a diverse range of areas with innovative technologies. The most critical and vital part of an IoT system that cannot be overlooked at any cost is its security. The security standards for IoT devices are not mature enough to provide foolproof security and there is still a long journey for manufacturers to incorporate stealth in devices. The most vulnerable component of an IoT system is the firmware which controls all the functionality of the device. If subverted by an attacker, the firmware of the IoT device can prove to be a critical attack surface for obtaining enough information to annihilate an IoT device. In this paper, we propose a twofold strategy to critically analyze the security of an IoT firmware. We will first use the STRIDE threat model to identify the security parameters that attackers could exploit to launch attacks. We will then use reverse engineering to examine and evaluate the security of a wide range of firmware being used in the latest and most commonly used IoT devices based on the identified security parameters. The same parameters can then derive security expectations for a secure IoT firmware. The proposed approach provides a powerful strategy to comprehensively analyze an IoT system's security. Our experimental results show that more than 50 percent of the firmware are exposing critical information that can be used to launch attacks. We believe that our findings will also help establish recommendations for developing secure and resilient firmware.
An efficient fault-prediction mechanism for improving yield in industry 5.0
Published in: 2022 24th International Multitopic Conference (INMIC)
Oct 21, 2022
Fariha Maqbool
Haroon Mahmood
Hasan Ali Khattak
Industrial sectors are constantly under pressure to produce higher-quality goods while maximizing yield. Machine maintenance is a critical component of manufacturing, accounting for a significant portion of total production costs. Corrective, preventive, and conditional maintenance strategies only make a negligible contribution to cost and downtime reduction. With the fifth industrial revolution, industrialists can now use sensors and cyber-physical systems to perform predictive maintenance on manufacturing operations. This strategy eliminates unnecessary maintenance and minimizes downtime by continuously collecting and analyzing data to predict time to failure. Numerous approaches to fault prediction have been proposed for predictive maintenance, but most of them are prohibitively expensive due to the massive number of features in manufacturing machines. The purpose of this work is to develop a technique for reliably predicting machine problems with the fewest possible features. To select features, we used SVR-based Recursive Feature Elimination (SVR-RFE) and Random Forest Regressor (RFR), while to predict, we used Long Short-Term Memory (LSTM) and Convolutional Neural Network (CNN). Our experiments on the 2018 PHM Challenge Dataset demonstrate that the proposed strategy outperforms prior approaches and reduces the mean absolute percentage error (SMAPE).
A taxonomy of IoT firmware security and principal firmware analysis techniques
Published in: International Journal of Critical Infrastructure Protection
Sep 01, 2022
Ibrahim Nadir
Haroon Mahmood
Ghalib Asadullah
Internet of Things (IoT) has come a long way since its inception. However, the standardization process in IoT systems for a secure IoT solution is still in its early days. Numerous quality review articles have been contributed by researchers on existing frameworks, architectures, as well as the threats to IoT on different layers. However, most of the existing work neglects the security aspects of firmware in the IoT ecosystem. As such, there is a lack of comprehensive survey on IoT firmware security that highlights critical reasons for firmware insecurity in IoT, lists vulnerabilities, and perform an in-depth review of the principal analysis techniques. This article aims to fill that gap by delivering, to the best of our knowledge, the first comprehensive review article of the firmware (in)security of IoT devices. Starting by highlighting the importance of firmware security, this research work recognizes critical reasons behind the insecurity of firmware by discussing technical, commercial, standardization, and researching aspects. In particular, the scope, evolution, and internals of IoT firmware along with their security implications are discussed. Furthermore, a taxonomic classification of IoT firmware vulnerabilities has been presented. We also discuss complications that hinder the detection of firmware vulnerabilities before doing a detailed analysis of existing vulnerability assessment tools and techniques. A comparative analysis of the principal analysis techniques is provided in terms of the vulnerabilities they discover, the methodology they employ, and the platform and/or architectures they support. Towards the end, some key research issues have been identified to encourage and facilitate research in the firmware security domain of IoT. Finally, some recommendations have been provided for the IoT device vendors, developers, and integrators.
Mobility modelling for urban traffic surveillance by a team of unmanned aerial vehicles
Published in: International Journal of Ad Hoc and Ubiquitous Computing
Feb 18, 2021
Farooq Ahmed
Haroon Mahmood
Yasir Niaz
Use of unmanned aerial vehicles (UAVs) for road traffic surveillance is an exciting idea for improving surveillance quality, as a component of intelligent transportation systems and smart cities. Calibrated mobility models help study and analyse several mobility related issues, for their successful deployment in large urban environments. This paper discusses an energy-aware and scalable mobility model for a team of cooperative UAVs monitoring urban road traffic. It also accompanies an extensible framework for territory distribution, to optimise the number of UAVs required for maximising coverage, without compromising operational performance. Since the problem of territory distribution in general is NP-hard, a genetic algorithm-based strategy, considering edge-disjoint paths, is recommended. Simulation results show that the proposed model considerably improves area coverage in comparison with other mobility models. A published dataset simulating vehicular traffic for a mid-size urban city is used to ascertain scalability to realistic urban territories.
Identifying mirai-exploitable vulnerabilities in iot firmware through static analysis
Published in: 2020 International Conference on Cyber Warfare and Security (ICCWS)
Oct 20, 2020
Zafeer Ahmed
Ibrahim Nadir
Haroon Mahmood
Ali Hammad Akbar
Ghalib Asadullah Shah
The prevalent use of IoT has raised numerous security concerns in recent times. One particular vulnerability in IoT ecosystem is weak authentication credentials. A large number of IoT attacks exploit such vulnerabilities. Emerged in 2016, the famous Mirai malware conducts attacks that benefits from poorly chosen username and passwords. Since its advent, Mirai attacks have only increased with time. Although multiple solutions have been suggested in literature based on dynamic packet analysis but existing solutions are expensive and are mostly based on reactionary measures. In this research work, we propose a scalable solution to audit the security of IoT firmware against the Mirai attack. Furthermore, we test our system by statically analyzing more than 1200 recent firmware images to inspect their resistance against Mirai botnet. Our results show that 193 out of 1200+ firmware images are susceptible to Mirai malware. To get effective results, we tested our solution against a variety of IoT devices' firmware images. We conclude that our solution is more scalable, less expensive and proactive as compared to other solutions.
An Auditing Framework for Vulnerability Analysis of IoT System
Published in: 2019 IEEE European Symposium on Security and Privacy Workshops (EuroS&PW)
Jun 17, 2019
Ibrahim Nadir
Zafeer Ahmad
Haroon Mahmood
Ghalib Asadullah Shah
Introduction of IoT is a big step towards the convergence of physical and virtual world as everyday objects are connected to the internet nowadays. But due to its diversity and resource constraint nature, the security of these devices in the real world has become a major challenge. Although a number of security frameworks have been suggested to ensure the security of IoT devices, frameworks for auditing this security are rare. We propose an open-source framework to audit the security of IoT devices covering hardware, firmware and communication vulnerabilities. Using existing open-source tools, we formulate a modular approach towards the implementation of the proposed framework. Standout features in the suggested framework are its modular design, extensibility, scalability, tools integration and primarily autonomous nature. The principal focus of the framework is to automate the process of auditing. The paper further mentions some tools that can be incorporated in different modules of the framework. Finally, we validate the feasibility of our framework by auditing an IoT device using proposed toolchain.
A Bayesian approach for estimating protein–protein interactions by integrating structural and non-structural biological data
Published in: Molecular BioSystems
Dec 01, 2017
Hafeez Ur Rehman
Inam Bari
Anwar Ali
Haroon Mahmood
Accurate elucidation of genome wide protein–protein interactions is crucial for understanding the regulatory processes of the cell. High-throughput techniques, such as the yeast-2-hybrid (Y2H) assay, co-immunoprecipitation (co-IP), mass spectrometric (MS) protein complex identification, affinity purification (AP) etc., are generally relied upon to determine protein interactions. Unfortunately, each type of method is inherently subject to different types of noise and results in false positive interactions. On the other hand, precise understanding of proteins, especially knowledge of their functional associations is necessary for understanding how complex molecular machines function. To solve this problem, computational techniques are generally relied upon to precisely predict protein interactions. In this work, we present a novel method that combines structural and non-structural biological data to precisely predict protein interactions. The conceptual novelty of our approach lies in identifying and precisely associating biological information that provides substantial interaction clues. Our model combines structural and non-structural information using Bayesian statistics to calculate the likelihood of each interaction. The proposed model is tested on Saccharomyces cerevisiae's interactions extracted from the DIP and IntAct databases and provides substantial improvements in terms of accuracy, precision, recall and F1 score, as compared with the most widely used related state-of-the-art techniques.
Cache aging reduction with improved performance using dynamically re-sizable cache
Published in: 2014 Design, Automation & Test in Europe Conference & Exhibition (DATE)
Mar 24, 2014
Haroon Mahmood
Massimo Poncino
Enrico Macii
Aging of transistors is a limiting factor for long term reliability of devices in sub-100nm technologies. It's a worst-case metric where the lifetime of a device is determined by the earliest failing component. Impact is more serious on memory arrays, where failure of a single SRAM cell would cause the failure of the whole system. Previous works have shown that partitioning based strategies based on power management techniques can effectively control aging effects and can extend lifetime of the cache significantly. However, such a benefit comes as a tradeoff with performance which reduces proportionally as the time elapses. To address this problem and provide a single solution to concurrently improve aging, energy and performance of the cache, we propose an architectural solution based on the dynamically re-sizable cache [5] and cache partitioning approaches. By this strategy, cache is dynamically re-sized and reconfigured whenever a cache block becomes unreliable. Coupling such aging mitigation technique along with dynamically re-sizable cache approach provides on average 30% lifetime improvement with less than 0.4x degradation in performance whereas, in previous solutions, performance degradation sometimes goes upto 10x.
Energy/lifetime cooptimization by cache partitioning with graceful performance degradation
Published in: IEEE Transactions on Very Large Scale Integration (VLSI) Systems
Sep 10, 2013
Haroon Mahmood
Mirko Loghi
Massimo Poncino
Enrico Macii
Aging of transistors can adversely impact the long-term reliability of devices in subnanometric technologies. Without any countermeasure, the first component that becomes unreliable will determine the life span of an entire device. The effect is more susceptible in memory arrays, where failure of a single SRAM cell would cause the failure of the whole system. In this paper, we propose a reliability management technique based on the idea of cache partitioning, which deals with cell failures by gracefully degrading its performance. By this partitioning-based strategy, various subblocks will become unreliable at different times, and the cache will keep functioning with reduced efficiency. A coarse-grain implementation of this approach, with the use of a smart aging-driven partitioning algorithm, provides a lifetime extension of more than 2X . On the other hand, a fine-grain strategy with a single cache line as a unit of power management, stretch the lifetime to its maximum limits with an addition of small hardware overhead.
Aging-aware caches with graceful degradation of performance
Published in: 2012 IEEE/IFIP 20th International Conference on VLSI and System-on-Chip (VLSI-SoC)
Oct 07, 2012
Haroon Mahmood
Massimo Poncino
Mirko Loghi
Enrico Macii
Aging of transistors can substantially shorten the lifetime of devices in sub-nanometric technologies. Without any countermeasure, the first component which becomes unreliable will determine the life span of an entire device. This problem is even more relevant for memory arrays, where failure of a single SRAM cell would cause the failure of the whole system. Traditional implementation of power management by turning idle cache lines into a low-energy state can also mitigate the aging effects caused by Negative Bias Temperature Instability (NBTI) provided that idleness is correctly exploited. In this work, we propose a cache structure which deals with cell failures by gracefully degrading its performance. By this partitioning-based strategy, various sub-blocks will become unreliable at different times, and the cache will keep functioning with reduced efficiency. Coupling such aging mitigation with the resulting energy reduction techniques we can obtain up to 2.5x lifetime extension and 40% energy savings with respect to a power managed cache.
Energy-optimal caches with guaranteed lifetime
Published in: 2012 ACM/IEEE international symposium on Low power electronics and design
Jul 30, 2012
Mirko Loghi
Haroon Mahmood
Andrea Calimera
Massimo Poncino
Enrico Macii
This work addresses the aging of the memory sub-system due to NBTI (Negative Bias Temperature Instability) in systems that have to provide a guaranteed level of service, and specifically, a guaranteed lifetime. Our approach leverages a novel cache architecture in which a smart joint use of redundancy and power management allows us to obtain caches that meet a desired lifetime target with minimal energy consumption. This is made possible by exploiting the possibility of putting the cache sub-block used for redundancy into a deep low-power state, thus allowing more energy saving than a regular architecture. Sacrificing a portion of the cache for aging mitigation only marginally affects performance thanks to the non-linear dependency of miss rate versus cache size, which allows to find the best cache size that maximizes the objective. Simulation results show that it is possible to meet the target lifetime by achieving energy reductions (measured over the lifetime of the system) ranging from 3X to 10X (2X to 8X) for a lifetime target of 15 (25) years, with marginal miss rate overhead.
Application-specific memory partitioning for joint energy and lifetime optimization
Published in: 2012 Design, Automation & Test in Europe Conference & Exhibition (DATE)
Mar 12, 2012
Haroon Mahmood
Massimo Poncino
Mirko Loghi
Enrico Macii
Power management of caches based on turning idle cache lines into a low-energy state is also beneficial for the aging effects caused by Negative Bias Temperature Instability (NBTI), provided that idleness is correctly exploited; unlike energy, aging, being a measure of delay, is in fact a worst-case metric.