Junaid Qadir
Qatar University, Department Of Computer Science & Engineering, Faculty Member
- Junaid Qadir completed Ph.D. from University of New South Wales, Australia in 2008 and his Bachelors in Electrical En... moreJunaid Qadir completed Ph.D. from University of New South Wales, Australia in 2008 and his Bachelors in Electrical Engineering from UET, Lahore, Pakistan in 2000. He is an Associate Professor at the Information Technology University (ITU)—Punjab, Lahore since December 2015.edit
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Wireless communication has achieved great proliferation in communication technologies. IEEE 802.11 WLANs have been widely deployed as wireless access mechanism technology. To keep up with growing bandwidth and data rate requirements,... more
Wireless communication has achieved great proliferation in communication technologies. IEEE 802.11 WLANs have been widely deployed as wireless access mechanism technology. To keep up with growing bandwidth and data rate requirements, legacy standards such as IEEE 802.11 have been deeply researched for their operation in multi-channel environment through usage of multiple interfaces. In the multi-channel multi-interface environment, it is common to assume that number of available channels are greater than or equal to number of interfaces. We evaluate performance of IEEE 802.11 DCF (Distributed Coordination Function) in multi-channel multi-interface environment in which number of channels are less than number of interfaces; contradicting common assumption where number of channels are greater than or equal to number of interfaces. In our study, we present scenarios under which number of available channels can become less than total number of interfaces on a node to motivate the problem. Through evaluation of accurate analytical models, we suggest that it is not possible to achieve higher average network throughput by tuning multiple interfaces on one channel. By keeping aggregate traffic constant for a channel, greater number of interfaces on a channel increases collisions, resulting in less effective usage of channel.
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
With the advances in machine learning (ML) and deep learning (DL) techniques, and the potency of cloud computing in offering services efficiently and cost-effectively, Machine Learning as a Service (MLaaS) cloud platforms have become... more
With the advances in machine learning (ML) and deep learning (DL) techniques, and the potency of cloud computing in offering services efficiently and cost-effectively, Machine Learning as a Service (MLaaS) cloud platforms have become popular. In addition, there is increasing adoption of third-party cloud services for outsourcing training of DL models, which requires substantial costly computational resources (e.g., high-performance graphics processing units (GPUs)). Such widespread usage of cloud-hosted ML/DL services opens a wide range of attack surfaces for adversaries to exploit the ML/DL system to achieve malicious goals. In this article, we conduct a systematic evaluation of literature of cloud-hosted ML/DL models along both the important dimensions—attacks and defenses—related to their security. Our systematic review identified a total of 31 related articles out of which 19 focused on attack, six focused on defense, and six focused on both attack and defense. Our evaluation reveals that there is an increasing interest from the research community on the perspective of attacking and defending different attacks on Machine Learning as a Service platforms. In addition, we identify the limitations and pitfalls of the analyzed articles and highlight open research issues that require further investigation
Research Interests:
This work addresses the problem of impersonation detection in an underwater acoustic sensor network (UWASN). We consider a UWASN consisting of $M$ underwater sensor nodes randomly deployed according to uniform distribution within a... more
This work addresses the problem of impersonation detection in an underwater acoustic sensor network (UWASN). We consider a UWASN consisting of $M$ underwater sensor nodes randomly deployed according to uniform distribution within a vertical half-disc (the so-called trusted zone). The sensor nodes report their sensed data to a sink node on water surface on an additive white gaussian noise (AWGN) reporting channel in a time-division multiple-access (TDMA) fashion. The ongoing communication on the shared reporting channel is at risk of potential impersonation attack by an active-yet-invisible adversary (so-called Eve) present in the close vicinity, who aims to inject malicious data into the system. To this end, this work proposes a novel, two-step method at the sink node to thwart the potential impersonation attack by Eve. We assume that the sink node is equipped with a uniform linear array of hydrophones; and therefore, the estimates of the distance, angle of arrival, and the location of the transmit node are available at the sink node. The sink node exploits these measurements as device fingerprints to carry out a number of binary hypothesis tests (for impersonation attack detection) as well as a number of maximum likelihood hypothesis tests (for transmitter identification when no impersonation is detected). We provide closed-form expressions for the error probabilities (i.e., the performance) of most of the hypothesis tests. Furthermore, extensive simulation results (for various scenarios of Eve's location) are provided, which attest to the efficacy of the proposed scheme.
Research Interests:
Research Interests:
Research Interests:
Research Interests:
The observation and management of cardiac features (using automated cardiac auscultation) is of significant interest to the healthcare community. In this work, we propose for the first time the use of recurrent neural networks (RNNs) for... more
The observation and management of cardiac features (using automated cardiac auscultation) is of significant interest to the healthcare community. In this work, we propose for the first time the use of recurrent neural networks (RNNs) for automated cardiac auscultation and detection of abnormal heartbeat detection. The application of RNNs for this task is compelling since RNNs represent the deep learning technique most adept at dealing with sequential or temporal data. We explore the use of various RNNs models and show through our experimental results that RNN delivers the best-recorded score with only 2.37\% error on the test set for automated cardiac auscultation task.
Research Interests:
This paper describes the design process by which we designed an Android application equipped with audio, textual menus and visuals components for use by farmers of diverse literacy levels looking for vital weather information after the... more
This paper describes the design process by which we designed an Android application equipped with audio, textual menus and visuals components for use by farmers of diverse literacy levels looking for vital weather information after the conclusion of research-work that productivity lags due to information inadequacies. The intervention provides more timely access to accurate information to low-literate farmers and thereby help in making the agricultural ecosystem more robust. We discuss the various design and implementation features of our system and presents our findings from the field on the usability of our application. We have also openly released our source code so that other users and developers can also benefit from our work.
Research Interests: Computer Science and Urdu
Research Interests:
Research Interests:
Research Interests:
In higher educational institutes, early grade prediction is an important area of interest as it allows instructors to improve students’ performance in their courses by providing special attention at the early stages. Machine learning... more
In higher educational institutes, early grade prediction is an important area of interest as it allows instructors to improve students’ performance in their courses by providing special attention at the early stages. Machine learning techniques can be utilized for students’ grades prediction in different courses. However, the performance of these techniques is highly dependent on the quality of data that made the selection of model a challenging task. Therefore, in this paper, we evaluate different state-of-the-art machine learning techniques for university students grade prediction. Ultimately we find that Restricted Boltzmann Machines (RBM) can more accurately predict students’ grades. The predicted grades by these techniques visualize uncertainty on student learning and can be used for confidence gains, student degree planning, personalized advising, and to enable instructors to identify potential students who might need assistance in relevant courses.
Research Interests:
Cognitive radio networks (CRNs) have emerged as a promising solution to the problem of spectrum under utilization and artificial radio spectrum scarcity. The paradigm of dynamic spectrum access allows a secondary network comprising of... more
Cognitive radio networks (CRNs) have emerged as a promising solution to the problem of spectrum under utilization and artificial radio spectrum scarcity. The paradigm of dynamic spectrum access allows a secondary network comprising of secondary users (SUs) to coexist with a primary network comprising of licensed primary users (PUs) subject to the condition that SUs do not cause any interference to the primary network. Since it is necessary for SUs to avoid any interference to the primary network, PU activity precludes attempts of SUs to access the licensed spectrum and forces frequent channel switching for SUs. This dynamic nature of CRNs, coupled with the possibility that an SU may not share a common channel with all its neighbors, makes the task of multicast routing especially challenging. In this work, we have proposed a novel multipath on-demand multicast routing protocol for CRNs. The approach of multipath routing, although commonly used in unicast routing, has not been explored for multicasting earlier. Motivated by the fact that CRNs have highly dynamic conditions, whose parameters are often unknown, the multicast routing problem is modeled in the reinforcement learning based framework of learning automata. Simulation results demonstrate that the approach of multipath multicasting is feasible, with our proposed protocol showing a superior performance to a baseline state-of-the-art CRN multicasting protocol.
Research Interests:
Deep machine learning techniques have shown promising results in network traffic classification, however, the robustness of these techniques under adversarial threats is still in question. Deep machine learning models are found vulnerable... more
Deep machine learning techniques have shown promising results in network traffic classification, however, the robustness of these techniques under adversarial threats is still in question. Deep machine learning models are found vulnerable to small carefully crafted adversarial perturbations posing a major question on the performance of deep machine learning techniques. In this paper, we propose a black-box adversarial attack on network traffic classification. The proposed attack successfully evades deep machine learning-based classifiers which highlights the potential security threat of using deep machine learning techniques to realize autonomous networks.
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Twitter has largely become a central online social network for arguments on various global controversial topics. Detecting and analysing such topics could prove to be beneficial in understanding the sentiments of trending topics in... more
Twitter has largely become a central online social network for arguments on various global controversial topics. Detecting and analysing such topics could prove to be beneficial in understanding the sentiments of trending topics in developing regions. In this paper, we perform a systematic sentiment study of trending controversial topics on Pakistan's Twitter user-base. From the data collected we build retweet graphs, partition graphs into communities, measure community influence, and label the communities as 'for' or 'against' per topic. To the best of our knowledge this is the first work to categorise and study sentiments attached to controversial topics in a developing region.
Research Interests:
Speech-based intelligent systems using deep learning are becoming increasingly important due to their wide range of applications in our routine life. Most of the efforts on voice signal processing are limited for the English language.... more
Speech-based intelligent systems using deep learning are becoming increasingly important due to their wide range of applications in our routine life. Most of the efforts on voice signal processing are limited for the English language. However, little effort has focused on voice signal processing for the Arabic language or for the Quran, which is the central religious book of Islam. In this study, our objective is to develop a deep learning based speaker identification using Quran recitations. We propose the use of Bidirectional Long Short-Term Memory (BLSTM)– a type of Recurrent Neural Networks (RNNs), which are well known for being particularly suitable for speech modeling and processing–for the task of Quranic speaker identification. Our results show that our BLSTM-based Quranic speaker identification delivers significantly improved results compared to previous approaches and is also computationally less expensive.