Zahidur Talukder - Academia.edu (original) (raw)

Uploads

Papers by Zahidur Talukder

Performance evaluation review, Apr 26, 2023

Motivation. Since its inception [6], Federated Learning (FL) has been enjoying a strong interest ... more Motivation. Since its inception [6], Federated Learning (FL) has been enjoying a strong interest from the privacypreserving AI research community. FL also has been commercially implemented in popular applications such as Google's Gboard and Apple Siri. In FL, Machine Learning (ML) models are trained cooperatively by many clients (e.g., mobile phones) without explicitly sharing their identity and private data. The training is done through iterative communication rounds between a central parameter/model server and clients. In each communication round, the global model is supplied to a subset of clients who use their private data to do local training on their respective devices. The updated models from each participating client are anonymously sent back to the central server and aggregated to get the global model for the next communication round.

ACM SIGMETRICS Performance Evaluation Review

Motivation. Since its inception [6], Federated Learning (FL) has been enjoying a strong interest ... more Motivation. Since its inception [6], Federated Learning (FL) has been enjoying a strong interest from the privacypreserving AI research community. FL also has been commercially implemented in popular applications such as Google's Gboard and Apple Siri. In FL, Machine Learning (ML) models are trained cooperatively by many clients (e.g., mobile phones) without explicitly sharing their identity and private data. The training is done through iterative communication rounds between a central parameter/model server and clients. In each communication round, the global model is supplied to a subset of clients who use their private data to do local training on their respective devices. The updated models from each participating client are anonymously sent back to the central server and aggregated to get the global model for the next communication round.

Proceedings of the Twentieth ACM Conference on Embedded Networked Sensor Systems

Server-level power monitoring in data centers can significantly contribute to its efficient manag... more Server-level power monitoring in data centers can significantly contribute to its efficient management. Nevertheless, due to the cost of a dedicated power meter for each server, most data center power management only focuses on UPS or cluster-level power monitoring. In this paper, we propose a low-cost novel power monitoring approach that uses only one sensor to extract power consumption information of all servers. We utilize the conducted electromagnetic interference of server power supplies to measure its power consumption from non-intrusive single-point voltage measurement. Using a pair of commercial grade Dell PowerEdge servers, we demonstrate that our approach can estimate each server's power consumption with ∼3% mean absolute percentage error.

2022 IEEE International Conference on Edge Computing and Communications (EDGE)

Federated Learning (FL) offers a privacy-preserving massively distributed Machine Learning (ML) p... more Federated Learning (FL) offers a privacy-preserving massively distributed Machine Learning (ML) paradigm where many clients cooperatively work together towards training a shared machine learning model. FL, however, is susceptible to data heterogeneity problems as the FL clients have diverse data sources. Prior works employ auto-weighted model aggregation to mitigate the heterogeneity issue to minimize the impact of unfavorable model updates. However, existing approaches require extensive computation for statistical analysis of clients' model updates. To circumvent this, we propose, FedASL (Federated Learning with Auto-weighted Aggregation based on Standard Deviation of Training Loss) which uses only the local training loss of FL clients for auto-weighting the model aggregation. Our evaluation under three different datasets and various data corruption scenarios reveals that FedASL can effectively thwart data corruption from bad clients while causing as little as onetenth of the computation cost of existing approaches.

2017 International Conference on Current Trends in Computer, Electrical, Electronics and Communication (CTCEEC), 2017

This paper presents the design and prototype implementation of USenSewer, an automated sewerage m... more This paper presents the design and prototype implementation of USenSewer, an automated sewerage management system that uses Arduino microcontroller coupled with an ultrasonic sensor, a NRF module and a GSM module to automate the routine checkup and removal of drain blockage vital for the continuous waste water flow in the big cities. The proposed system consists of three main components; the blockage detection system which is placed inside the drain, the blockage removal system which is placed above the ground level and a control center which controls and coordinates the collection of waste materials placed by the first two components. Several pilot installations of the system in the city of Dhaka have shown that the proposed system significantly outperforms the state-of-theart manual systems prevailing in the city from the scalability, flexibility and economic point of view.

2017 International Conference on Current Trends in Computer, Electrical, Electronics and Communication (CTCEEC), 2017

The urging need for seamless connectivity in mobile environment has contributed to the rapid expa... more The urging need for seamless connectivity in mobile environment has contributed to the rapid expansion of Mobile IP. Mobile IP uses wireless transmission medium, thereby making it subject to many security threats during various phases of route optimization. Modeling Mobile IP attacks reasonably and efficiently is the basis for defending against those attacks, which requires quantitative analysis and modeling approaches for expressing threat propagation in Mobile IP. In this Paper, we present four well-known Mobile IP attacks, such as Denialof-Service (DoS) attack, bombing attack, redirection attack and replay attack and model them with Stochastic Game Petri Net (SGPN). Furthermore, we propose mixed strategy based defense strategies for the aforementioned attacks and model them with SGPN. Finally, we calculate the Nash Equilibrium of the attacker-defender game and thereby obtain the steady state probability of the vulnerable attack states. We show that, under the optimal strategy, an IDS needs to remain active 72.4%, 70%, 68.4% and 66.6% of the time to restrict the attacker's success rate to 8.5%, 6.4%, 7.2% and 8.3% respectively for the aforementioned attacks, thus performing better than the stateof-the-art approach.

2019 1st International Conference on Advances in Science, Engineering and Robotics Technology (ICASERT), 2019

Phasors are significant to power system analysis in both steady-state and dynamic conditions. IEE... more Phasors are significant to power system analysis in both steady-state and dynamic conditions. IEEE Standard C37.118.1 2011 along with its recent amendment C37.118.1a2014 define synchrophasor measurements with the focus on the behavior of phasor measurement units (PMUs) under steady-state and dynamic operating conditions. The Synchrophasors can find many applications in modern power systems including real-time status monitoring, precise state estimation, validating simulation models, protection, taking centralized control actions using wide-area measurements and integrating renewable and intermittent resources. In this paper, various PMU algorithms are compared to find out the best algorithm which gives best result in specific conditions. As different research papers suggested and focused on different criterions, it was very difficult and cumbersome task to accumulate all data in a common platform and compare them. After vigorous research a unique way is developed to present overall comparison result in a well-defined and more understandable manner.

ArXiv, 2020

While the COVID-19 pandemic continues to be as complex as ever, the collection and exchange of da... more While the COVID-19 pandemic continues to be as complex as ever, the collection and exchange of data in the light of fighting coronavirus poses a major challenge for privacy systems around the globe. The disease's size and magnitude is not uncommon but it appears to be at the point of hysteria surrounding it. Consequently, in a very short time, extreme measures for dealing with the situation appear to have become the norm. Any such actions affect the privacy of individuals in particular. For some cases, there is intensive monitoring of the whole population while the medical data of those diagnosed with the virus is commonly circulated through institutions and nations. This may well be in the interest of saving the world from a deadly disease, but is it really appropriate and right? Although creative solutions have been implemented in many countries to address the issue, proponents of privacy are concerned that technologies will eventually erode privacy, while regulators and priva...

International Journal of Network Security & Its Applications, 2020

The huge amounts of data and information that need to be analyzed for possible malicious intent a... more The huge amounts of data and information that need to be analyzed for possible malicious intent are one of the big and significant challenges that the Web faces today. Malicious software, also referred to as malware developed by attackers, is polymorphic and metamorphic in nature which can modify the code as it spreads. In addition, the diversity and volume of their variants severely undermine the effectiveness of traditional defenses that typically use signature-based techniques and are unable to detect malicious executables previously unknown. Malware family variants share typical patterns of behavior that indicate their origin and purpose. The behavioral trends observed either statically or dynamically can be manipulated by using machine learning techniques to identify and classify unknown malware into their established families. This survey paper gives an overview of the malware detection and analysis techniques and tools.

International Journal of Network Security & Its Applications (IJNSA), 2020

The huge amounts of data and information that need to be analyzed for possible malicious intent a... more The huge amounts of data and information that need to be analyzed for possible malicious intent are one of the big and significant challenges that the Web faces today. Malicious software, also referred to as malware developed by attackers, is polymorphic and metamorphic in nature which can modify the code as it spreads. In addition, the diversity and volume of their variants severely undermines the effectiveness of traditional defenses that typically use signature-based techniques and are unable to detect malicious executables previously unknown. Malware family variants share typical patterns of behavior which indicate their origin and purpose. The behavioural trends observed either statically or dynamically can be manipulated by using machine learning techniques to identify and classify unknown malware into their established families. This survey paper gives an overview of the malware detection and analysis techniques and tools.

Performance evaluation review, Apr 26, 2023

Motivation. Since its inception [6], Federated Learning (FL) has been enjoying a strong interest ... more Motivation. Since its inception [6], Federated Learning (FL) has been enjoying a strong interest from the privacypreserving AI research community. FL also has been commercially implemented in popular applications such as Google's Gboard and Apple Siri. In FL, Machine Learning (ML) models are trained cooperatively by many clients (e.g., mobile phones) without explicitly sharing their identity and private data. The training is done through iterative communication rounds between a central parameter/model server and clients. In each communication round, the global model is supplied to a subset of clients who use their private data to do local training on their respective devices. The updated models from each participating client are anonymously sent back to the central server and aggregated to get the global model for the next communication round.

ACM SIGMETRICS Performance Evaluation Review

Motivation. Since its inception [6], Federated Learning (FL) has been enjoying a strong interest ... more Motivation. Since its inception [6], Federated Learning (FL) has been enjoying a strong interest from the privacypreserving AI research community. FL also has been commercially implemented in popular applications such as Google's Gboard and Apple Siri. In FL, Machine Learning (ML) models are trained cooperatively by many clients (e.g., mobile phones) without explicitly sharing their identity and private data. The training is done through iterative communication rounds between a central parameter/model server and clients. In each communication round, the global model is supplied to a subset of clients who use their private data to do local training on their respective devices. The updated models from each participating client are anonymously sent back to the central server and aggregated to get the global model for the next communication round.

Proceedings of the Twentieth ACM Conference on Embedded Networked Sensor Systems

Server-level power monitoring in data centers can significantly contribute to its efficient manag... more Server-level power monitoring in data centers can significantly contribute to its efficient management. Nevertheless, due to the cost of a dedicated power meter for each server, most data center power management only focuses on UPS or cluster-level power monitoring. In this paper, we propose a low-cost novel power monitoring approach that uses only one sensor to extract power consumption information of all servers. We utilize the conducted electromagnetic interference of server power supplies to measure its power consumption from non-intrusive single-point voltage measurement. Using a pair of commercial grade Dell PowerEdge servers, we demonstrate that our approach can estimate each server's power consumption with ∼3% mean absolute percentage error.

2022 IEEE International Conference on Edge Computing and Communications (EDGE)

Federated Learning (FL) offers a privacy-preserving massively distributed Machine Learning (ML) p... more Federated Learning (FL) offers a privacy-preserving massively distributed Machine Learning (ML) paradigm where many clients cooperatively work together towards training a shared machine learning model. FL, however, is susceptible to data heterogeneity problems as the FL clients have diverse data sources. Prior works employ auto-weighted model aggregation to mitigate the heterogeneity issue to minimize the impact of unfavorable model updates. However, existing approaches require extensive computation for statistical analysis of clients' model updates. To circumvent this, we propose, FedASL (Federated Learning with Auto-weighted Aggregation based on Standard Deviation of Training Loss) which uses only the local training loss of FL clients for auto-weighting the model aggregation. Our evaluation under three different datasets and various data corruption scenarios reveals that FedASL can effectively thwart data corruption from bad clients while causing as little as onetenth of the computation cost of existing approaches.

2017 International Conference on Current Trends in Computer, Electrical, Electronics and Communication (CTCEEC), 2017

This paper presents the design and prototype implementation of USenSewer, an automated sewerage m... more This paper presents the design and prototype implementation of USenSewer, an automated sewerage management system that uses Arduino microcontroller coupled with an ultrasonic sensor, a NRF module and a GSM module to automate the routine checkup and removal of drain blockage vital for the continuous waste water flow in the big cities. The proposed system consists of three main components; the blockage detection system which is placed inside the drain, the blockage removal system which is placed above the ground level and a control center which controls and coordinates the collection of waste materials placed by the first two components. Several pilot installations of the system in the city of Dhaka have shown that the proposed system significantly outperforms the state-of-theart manual systems prevailing in the city from the scalability, flexibility and economic point of view.

2017 International Conference on Current Trends in Computer, Electrical, Electronics and Communication (CTCEEC), 2017

The urging need for seamless connectivity in mobile environment has contributed to the rapid expa... more The urging need for seamless connectivity in mobile environment has contributed to the rapid expansion of Mobile IP. Mobile IP uses wireless transmission medium, thereby making it subject to many security threats during various phases of route optimization. Modeling Mobile IP attacks reasonably and efficiently is the basis for defending against those attacks, which requires quantitative analysis and modeling approaches for expressing threat propagation in Mobile IP. In this Paper, we present four well-known Mobile IP attacks, such as Denialof-Service (DoS) attack, bombing attack, redirection attack and replay attack and model them with Stochastic Game Petri Net (SGPN). Furthermore, we propose mixed strategy based defense strategies for the aforementioned attacks and model them with SGPN. Finally, we calculate the Nash Equilibrium of the attacker-defender game and thereby obtain the steady state probability of the vulnerable attack states. We show that, under the optimal strategy, an IDS needs to remain active 72.4%, 70%, 68.4% and 66.6% of the time to restrict the attacker's success rate to 8.5%, 6.4%, 7.2% and 8.3% respectively for the aforementioned attacks, thus performing better than the stateof-the-art approach.

2019 1st International Conference on Advances in Science, Engineering and Robotics Technology (ICASERT), 2019

Phasors are significant to power system analysis in both steady-state and dynamic conditions. IEE... more Phasors are significant to power system analysis in both steady-state and dynamic conditions. IEEE Standard C37.118.1 2011 along with its recent amendment C37.118.1a2014 define synchrophasor measurements with the focus on the behavior of phasor measurement units (PMUs) under steady-state and dynamic operating conditions. The Synchrophasors can find many applications in modern power systems including real-time status monitoring, precise state estimation, validating simulation models, protection, taking centralized control actions using wide-area measurements and integrating renewable and intermittent resources. In this paper, various PMU algorithms are compared to find out the best algorithm which gives best result in specific conditions. As different research papers suggested and focused on different criterions, it was very difficult and cumbersome task to accumulate all data in a common platform and compare them. After vigorous research a unique way is developed to present overall comparison result in a well-defined and more understandable manner.

ArXiv, 2020

While the COVID-19 pandemic continues to be as complex as ever, the collection and exchange of da... more While the COVID-19 pandemic continues to be as complex as ever, the collection and exchange of data in the light of fighting coronavirus poses a major challenge for privacy systems around the globe. The disease's size and magnitude is not uncommon but it appears to be at the point of hysteria surrounding it. Consequently, in a very short time, extreme measures for dealing with the situation appear to have become the norm. Any such actions affect the privacy of individuals in particular. For some cases, there is intensive monitoring of the whole population while the medical data of those diagnosed with the virus is commonly circulated through institutions and nations. This may well be in the interest of saving the world from a deadly disease, but is it really appropriate and right? Although creative solutions have been implemented in many countries to address the issue, proponents of privacy are concerned that technologies will eventually erode privacy, while regulators and priva...

International Journal of Network Security & Its Applications, 2020

The huge amounts of data and information that need to be analyzed for possible malicious intent a... more The huge amounts of data and information that need to be analyzed for possible malicious intent are one of the big and significant challenges that the Web faces today. Malicious software, also referred to as malware developed by attackers, is polymorphic and metamorphic in nature which can modify the code as it spreads. In addition, the diversity and volume of their variants severely undermine the effectiveness of traditional defenses that typically use signature-based techniques and are unable to detect malicious executables previously unknown. Malware family variants share typical patterns of behavior that indicate their origin and purpose. The behavioral trends observed either statically or dynamically can be manipulated by using machine learning techniques to identify and classify unknown malware into their established families. This survey paper gives an overview of the malware detection and analysis techniques and tools.

International Journal of Network Security & Its Applications (IJNSA), 2020

The huge amounts of data and information that need to be analyzed for possible malicious intent a... more The huge amounts of data and information that need to be analyzed for possible malicious intent are one of the big and significant challenges that the Web faces today. Malicious software, also referred to as malware developed by attackers, is polymorphic and metamorphic in nature which can modify the code as it spreads. In addition, the diversity and volume of their variants severely undermines the effectiveness of traditional defenses that typically use signature-based techniques and are unable to detect malicious executables previously unknown. Malware family variants share typical patterns of behavior which indicate their origin and purpose. The behavioural trends observed either statically or dynamically can be manipulated by using machine learning techniques to identify and classify unknown malware into their established families. This survey paper gives an overview of the malware detection and analysis techniques and tools.