Prof. Kamran Arshad has 18+ years of research and teaching experience in higher education and he is currently Dean of Research and Graduate Studies, and a Professor in Electrical Engineering at Ajman University, UAE. Prior to join Ajman University in January 2016, he has been associated with the University of Greenwich, UK as a Senior Lecturer (Associate Professor) and a Program Director of MSc Wireless Mobile Communications Systems Engineering. Prof. Arshad is a Senior Member of IEEE (SM-IEEE), a Senior Fellow of the UK Higher Education Academy (SF-HEA) and an Associate Editor of EURASIP Journal on Wireless Communications and Networking. Prof. Arshad research interests are in the areas of cognitive radio, LTE/LTE-Advanced, 5G, and cognitive Machine-to-Machine (M2M) communications. He was a project manager at the University of Surrey, UK for the European project QoSMOS and lead University of Surrey team involved in the project. He has a global collaborative research network spanning both academia and key industrial players in the field of wireless communications. Prof. Arshad has contributed to several European and international large-scale research projects. Prof. Arshad worked closely with leading European companies in communications, such as British Telecom (UK), Toshiba (UK), Thales (France), NEC (Japan/France), Agilent (Belgium), Alcatel Lucent (Germany), Nokia (Finland), Ericsson (Sweden) etc. and produced several joint publications. He has supervised 5+ successful PhD graduates and supervised 40+ MSc thesis. Prof. Arshad has 140+ technical peer-reviewed papers in top quality journals and international conferences. He received 3 best paper awards, 1 best R&D track award and chaired technical sessions in several leading international conferences. Prof. Arshad also authored/co-authored 25+ technical reports submitted to the European Commission (EC). He served 100+ international conferences as a Technical Program Committee (TPC) member and an editor of several international journals. Prof. Arshad returned research outputs in UK Research Excellence Framework (REF) 2014 .
This paper proposes an ultrahigh-frequency (UHF) radio-frequency identification (RFID) tag sensor for detecting fat content and spoilage in milk. This dual-slot-based tag, designed and optimized for whole milk (3.6% fat content) using characteristic mode analysis (CMA), features two open triangular slots that improve impedance matching with milk bottles. The tag antenna achieves a good impedance match and bandwidth (880 MHz - 950 MHz). Prototypes of the tag were pasted onto milk bottles with fat contents of 1.5% (low-fat cow's milk), 3.6% (full-cream cow's milk), and 4.5% (full-cream buffalo's milk). The experiments utilized an RFID reader setup with an EIRP of 30 dBm (1 W). The results demonstrate the proposed sensor tag's capability for detecting milk's fat content. Furthermore, this tag paves the way for non-invasive detection of milk adulteration (primarily dilution with water) and spoilage.
This paper presents a wideband, long-range RFID tag antenna design with platform tolerant features using characteristic mode analysis (CMA). The proposed design consists of a two-layer structure, that are separated by an air gap. The upper layer consists of T-match and dual loop-based meandered antenna optimized for achieving long read range characteristics in free space/low permittivity dielectrics. The second layer consists of multi-resonant strips printed on a grounded substrate carefully designed using CMA for producing multi-resonant modes in the required UHF RFID band. The multi-resonant strips help in achieving wide bandwidth and isolation between different environment surfaces. This tag antenna provides a wide bandwidth ranging from 900 – 940 MHz that covers both the US UHF RFID band (902 – 928 MHz) and the upper European UHF RFID band (915 – 921 MHz). Moreover, this design achieved a read range of 14.7 m in free space, 14 m on a metal plate, and 13 m on wood, and glass. Furthermore, the proposed antenna was also tested in harsh metal environments such as car number plates and cargo metallic containers. Therefore, the proposed tag can be used for tagging metallic containers, wood containers, and other harsh platforms for cargo management, Asset tracking, supply chain visibility, and Internet of Things (IoT) Applications.
This paper presented a knitted antenna design for ultra-high frequency (UHF) radiofrequency identification (RFID) and wearable internet of things (IoT) applications. The proposed antenna consists of a dual slot-match structure that provides a good impedance match with RFID microchip on high permittivity surfaces such as the human body. The slot-match structure is designed and optimized using characteristic mode analysis (CMA). The slot-match structure poses a very high inductive impedance and very low real impedance in free space. However, this impedance behavior helps in countering high capacitive effects caused by the human body and other high permittivity materials. The proposed antenna is fabricated by knitting using conductive thread. This antenna design features a read range of 2.4 m measured using an RFID reader setup after placing the tag on the human body abdomen. Moreover, the proposed antennas can be used as a sensor for vital signs or breath monitoring as its stretched state provide less read range as compared to the unstretched state. Therefore, the proposed antenna design can be used for UHF RFID, sensing, and wearable IoT applications.
Artificial Intelligence (AI) in tandem wireless technologies is providing state-of-the-art techniques human motion detection for various applications including intrusion detection, healthcare and so on. Radio Frequency (RF) signal when propagating through the wireless medium encounters reflection and this information is stored when signals reach the receiver side as Channel State information (CSI). This paper develops an intelligent wireless sensing prototype for healthcare that can provide quasi-real time classification of CSI carrying various human activities obtained using USRP wireless devices. The dataset is collected from the CSI of USRP devices when a volunteer sits down or stands up as a test case. A model is created from this dataset for making predictions on unknown data. Random forest was able to provide the best results with an accuracy result to 96.70% and used for the model. A wearable device dataset was used as a benchmark to provide a comparison in performance of the USRP dataset.
As the world is experiencing population growth, the portion of the older people, aged 65 and above, is also growing at a faster rate. As a result, the dementia with Alzheimer’s disease is expected to increase rapidly in the next few years. Currently, healthcare systems require an accurate detection of the disease for its treatment and prevention. Therefore, it has become essential to develop a framework for early detection of Alzheimer’s disease to avoid complications. To this end, a novel framework, based on machine-learning (ML) and deep-learning (DL) methods, is proposed to detect Alzheimer’s disease. In particular, the performance of different ML and DL algorithms has been evaluated against their detection accuracy. The experimental results state that bidirectional long short-term memory (BiLSTM) outperforms the ML methods with a detection accuracy of 91.28%. Furthermore, the comparison with the state-of-the-art indicates the superiority of the our framework over the other proposed approaches in the literature.
Human Posture Classification (HPC) is used in many fields such as human computer interfacing, security surveillance, rehabilitation, remote monitoring, and so on. This paper compares the performance of different classifiers in the detection of 3 postures, sitting, standing, and lying down, which was recorded using Microsoft Kinect cameras. The Machine Learning classifiers used included the Support Vector Classifier, Naive Bayes, Logistic Regression, K-Nearest Neighbours, and Random Forests. The Deep Learning ones included the standard Multi-Layer Perceptron, Convolutional Neural Networks (CNN), and Long Short Term Memory Networks (LSTM). It was observed that Deep Learning methods outperformed the former and that the one-dimensional CNN performed the best with an accuracy of 93.45%.
The next generation of health activity monitoring is greatly dependent on wireless sensing. By analysing variations in channel state information, several studies were capable of detecting activities in an indoor setting. This paper presents promising results of an experiment conducted to identify the activity performed by a subject and where it took place within the activity region. The system utilises two Universal Software Radio Peripheral (USRP) devices, operating as software-defined radios, to collect a total of 360 data samples that represent five different activities and an empty room. The five activities were performed in three different zones, resulting in 15 classes and a 16t h class representing the room whilst it is empty. Using the Random Forest classifier, the system was capable of differentiating between the majority of activities, across the 16 classes, with an accuracy of almost 94 %. Moreover, it was capable of detecting whether the room is occupied, with an accuracy of 100 %, and identify the walking directions of a human subject in three different positions within the room, with an accuracy of 90 %