홈으로ArticlesAll Issue
ArticlesQuality of Service-Aware Resource Selection in Healthcare IoT Using Deep Autoencoder Neural Networks
  • Ramachandran Manikandan1, Indu2, Victor Hugo C. de Albuquerque3, Prayag Tiwari4, Salman Ali AlQahtani5,* , and M. Shamim Hossain6,*

Human-centric Computing and Information Sciences volume 12, Article number: 36 (2022)
Cite this article 3 Accesses
https://doi.org/10.22967/HCIS.2022.12.036

Abstract

Heterogeneous network and device-to-device communication are two possible solutions for improving wireless network spectral efficiency. Techniques based on the Internet of Things (IoT) can interact between a large number of smart devices as well as heterogeneous networks. The goal of this study is to investigate proposed quality of service-aware resource selection in an IoT network for healthcare data using a deep auto encoder neural network with spectrum reuse utilizing mixed integer nonlinear programming (MINLP). The suggested MINLP spectrum reuse was used to address the optimization problem, and the spectrum allocation was done using fast Fourier transform based reinforcement Q-learning. Increased transmission repetitions have been identified as a promising strategy for improving IoT coverage by up to 164 dB in terms of maximum coupling loss for uplink transmissions, which is a significant improvement over traditional LTE technology, particularly for serving customers in deep coverage. Based on a comparison of existing methodologies, the experimental study is performed using parameters such as bit error rate of 40%, signal-to-interference plus noise ratio of 72%, sum rate of 88%, and spectral efficiency of 98%.


Keywords

HetNet, D2D, IoT, QoS, Spectrum Reuse, Healthcare


Introduction

Information and communication technology is increasingly being used in the healthcare sector as part of smart city infrastructure, with network technology serving as the foundation for data transmission and reception. Anyone who owns a mobile device or a smart sensor understands how important wireless communication has become in their daily lives. The fact that these devices and sensors are linked to a data network alters our overall perception of the Internet. After all, they can automatically communicate data with one another without the need for human intervention, resulting in machine-to-machine (M2M) traffic, also known as MTC [1] . As a result of recent advancements in communication and networking applications, there is a tremendous demand for a next-generation communication paradigm. According to global telecommunication market trends [2], wireless communication networks (WCNs) are becoming one of the most important aspects in the deployment of the Internet of Things (IoT) paradigm. Deviceto-device (D2D) communication is a promising method for increasing the spectral efficiency of WCNs. A heterogeneous network (HetNet), as depicted in Fig. 1, is composed of various tiers of networks with varying cell sizes/footprints and/or radio access methods [3].

Fig. 1. Schematic representation of heterogeneous network (HetNet).


HetNets are a good way to increase system capacity; improve network coverage, service quality, and fairness; and make it easier to integrate new types of networks, connectivity, and applications [4]. Furthermore, rather than providing best-effort services, the quality of service (QoS)-aware routing method is used to provide differentiated data transmission services. The Internet Engineering Task Force has developed three methods to ensure the QoS needs of applications are met during data transmission: IntServ, DiffServ, and MPLS [4].
With the advancement of technology over the years, it is now possible to diagnose and monitor a variety of diseases using small technologies such as smart watches. Furthermore, technological advancements have transformed the healthcare system from a hospital-centric to a patient-centric one. Because a large amount of data is collected/recorded from various sources, data storage and accessibility are especially important in the IoT system. The data collected by the aforementioned sensing devices is accessible to doctors, care providers, and other authorized individuals. The ability to share this data with healthcare providers via cloud/server allows for faster patient diagnosis and, if necessary, medical action. Users, patients, and the communication module must all work together to ensure effective and secure transmission [5]. QoS is an important component in such systems, as are usage perspectives with regard to ProSumers. IoT is one of the fastest-growing technique paradigms used in every industry, which is a critical component in such systems and usage perspectives with regard to ProSumers. Most recent research works on QoS in IoT have used machine learning (ML) approaches as one of the evaluating methodologies for increased performance and solutions. ML and related methodologies have become a common trend and requirement across a wide range of technologies and domains, including open-source frameworks, task-specific algorithms, and the application of artificial intelligence (AI) and ML methods.
The following is the research's contribution:

To improve cellular network coverage and capacity, allowing for a better customer experience, and to collect healthcare data using the IoT.

To enhance spectrum reuse using mixed integer nonlinear programming (MINLP)-based fast Fourier transform-based reinforcement Q-learning (FFT RQL).

To design a QoS-aware resource selection system based on a deep autoencoder neural network (DAENN) that maximizes a weighted total rate while meeting minimum rate and transmission power requirements.


The remainder of the paper is organized as follows: Section 2 provides a brief description of works found in literature. Section 3 describes the proposed IoT-based network for QoS-aware resource selection and spectrum reuse, as well as the network optimization model. Section 4 presents the results of the simulation. Section 5 concludes the work and suggests a few improvements for the future.


Related Work

The IoT generates large amounts of unstructured data from a variety of applications, such as smart-connected health [612], disease detection [1315], Internet of Health Things (IoHT) [16, 17] and so forth. The IoHT has brought about a widespread acceptance of IoT devices in our daily healthcare system. Healthcare data connected with IoT devices over networks generate a massive amount of unstructured data, which is quite impossible to process with traditional ML techniques. Thus, the importance of AI-powered IoT technologies for healthcare services is being regarded as one of the next-generation network's killer applications. However, owing to security flaws, an unauthorized person might get access to health-related data or operate IoT nodes linked to the patient's body, leading to adversarial and other security effects [1820].
A large number of studies on two-/multi-tier HetNets have been conducted with the goal of optimizing user association, as well as research on energy-significant channel assignment and spectrum allocation for homogeneous cellular networks [21]. A user association and interference management strategy that maximizes the sum utility of average achievable rates was presented [22]. The authors of [23] proposed a game-theoretic approach (GTA) for defining a throughput-maximizing user association problem. In [24], the author proposed a general paradigm for combined resource partitioning and offloading on a two-tier HetNet. They figured out the best way to increase the number of cell edge users. Their research, however, did not take into account QoS requirements. Authors in [25] described a downlink HetNet cell association as well as a resource allocation method for balancing network traffic. They also developed a distributed method but did not consider the QoS requirements [26]. A comprehensive evaluation of LTE uplink schedulers for M2M devices is provided in research [27].
According to the authors of this study, there are three major types of existing schedulers. Power-saving schedulers are designed to reduce power consumption in M2M devices [28]. Multi-hop schedulers and QoS-based schedulers [29, 30] are intended to provide QoS processing for various M2M applications. A QoS hybrid uplink scheduler [31] is presented in order to reduce the number of base station (BS) while improving system performance by increasing coverage area. Authors in [32] proposed a hybrid scheduling approach for a HetNet that uses human to human (H2H) and M2M communications. In their study, the authors proposed a cluster-based group paging (CBGP) technique for reducing congestion in mMTC settings [33]. Authors in [34] investigated sum-rate optimization for D2D and cellular links, as well as resource allocation (RA) and interference management issues in D2D communication underpinning cellular networks. A centralized spectrum allocation approach was developed in [35] to maximize system throughput while ensuring QoS of both cellular users (CUs) and D2D pairs. In [36], a coalitional game strategy is used to solve the joint mode selection and RA problems. A cyber-physical approach to health monitoring management had previously been proposed in [37]. A distributed deep learning-based healthcare framework is proposed in [15] that uses network functions virtualization. Using the dynamic voltage and frequency scaling method, [20] proposed a method known as multiagent deep Q-network with coral reefs optimization, which combines the coral reefs optimization method with the multiagent deep Q-network to minimize the energy consumption of data centers and cloud resources. To perform automatic optimal RA, the author of [38] proposed a new framework that employs a combinatory method based on reactive as well as proactive scaling that is sensitive to previous and present fluctuations. The authors in [39] construct encrypted data classifiers based on dubbed DataNets techniques: multilayer perceptron, convolutional neural network, and stacked autoencoder using an open data set containing over 200,000 encrypted data samples from 15 applications. A data preprocessing system is provided to process raw data packets and tested data sets in order to develop DataNets. The study in [40] focuses on the use of DL to improve human activity recognition (HAR) in IoHT situations. A semi-supervised DL framework is created and built for more accurate HAR, which efficiently uses and analyzes weakly labelled sensor data to train classifier learning models.
According to our review of state-of-the-art methods, the two main problems that the research community must address to propose a suitable solution for the services selection issue are identifying techniques used to design an appropriate IoT environment and determining methodology used to implement proposed solution. A third critical criterion is an evaluation stage for determining the efficiency and efficacy of the proposed solution. When the number of people and devices increases, or when multicellular situations are taken into account, these methodologies and approaches face challenges. Furthermore, IoT solutions receive and generate massive amounts of data that standard QoS assurance methods are incapable of handling, particularly when it comes to extracting relevant features from this data. To improve and ensure QoS in the IoT, DL methods are proposed as a possible candidate for addressing and handling the aforementioned issues.


System Model


This section discusses the proposed model of an IoT-based network for QoS-aware resource selection and spectrum reuse with network optimization using a DAENN and MINLP integrated with FFT RQL. Fig. 2 depicts the overall proposed model.

Fig. 2.Overall proposed model.



Results Discussion

Proposed Model in QoS-Aware Resource Selection using DAENN

First train an autoencoder for a DAENN by recreating raw data in relation matrix B and getting a hidden layer H1∈R (N × $t_1$). Then H1 is used as the next autoencoder's input. The rest of layers are completed in the same way. The DAENN's architecture is depicted in Fig. 3.

Fig. 3Deep autoencoder neural network (DAENN) operations: (a) forward feature abstraction and (b) backward error feedback.


Consider a training set with {$x^1$,$x^2$,…,$x^n$ } training data and $x^i$∈$R^n$ for each training data. For i∈{1,2,…,n}, the autoencoder objective is $y^i$=$x^i$, which means that the network's output will be equal to its input. With this objective function in mind, the autoencoder attempts to learn a compressed representation of the dataset, i.e., it learns the identity function where W and b are the entire network’s weights and bias vectors [41].
Forward feature abstraction: Assume that the DL network has N layers. The output of the jth node in layer I, abbreviated asn_ij, is obtained in two phases. To begin, node n_ij computes a weighted sum of all of its inputs, which is indicated by $z_{ij}$. The output $y_{ij}$ of node $n_{ij}$ is then obtained by sending $z_{ij}$ to a nonlinear function f() given by Equation (1).

(1)


where is the weight from node to node and $L_{-1}$ is the laye$r_{i-1}$ node count.
Backward error feedback: Starting weights are either random or empirical. These weight values are altered via a backward error feedback approach to increase the accuracy of the learning system’s final output. The error derivative for a node in the deepest layer, say node $n_{Nj,}$, is $y_{Nj,}$ $t_{Nj}$ where $y_{Nj,}$ and $t_{Nj}$ are generated as well as correct outputs, respectively. The error derivative of the lower layer link is then calculated using Equation (2).

(2)




3.1.1 Encoding
Encoding is the process of utilizing the mapping function $f_θ$ to transform a pre-completed user vector x0 to hidden feature vector (HFV) h. An affine transformation follows the nonlinear transformation using Equation (3).

(3)


where pre-completed user vector x0 has a dimension of d, and HFV h has a dimension of d0. The parameter set is θ = W1, b1, where W1 is a $d_0$×d matrix and b1 is an offset vector with $d_0$ dimensions.


3.1.2 Decoding
Using the mapping function decoding is the process of turning an HFV into a d-dimensional reassembled user vector y. After an affine transformation, a nonlinear transformation is usually utilized using Equation (4).

(4)


The autoencoder's is to reduce the reconstructed error between the pre-completed user vector x0 and the recreated vector y; hence the goal of autoencoder training is to discover the best parameters θ and θ, 0 to reduce the recreated error in the training set. To calculate the reconstructed error, utilize the squared error function E(x,y)=∥x-y∥^2 using Equation (5).

(5)


where vector $x_u$ is the user u's pre-completed user vector, vector $y_u$ is the user u's rebuilt user vector, and λ is a regularization parameter by Equation (6).

(6)


W_H∈$R^{N×d}$ and d_H∈$R^{N×1}$ are the weights of the coding layer and bias vector, respectively and s() is a nonlinear mapping function using Equation (7).

(7)


where $W_O$∈$R^{N×d}$,and $d_O$∈$R^{N×1}$are the weights of the decoding layer and the bias vector, respectively. The main purpose of an autoencoder is to investigate condition of a parameter = which will rebuild the original data in order to minimize raw data M and recreate deviation within data O. Fig. 4 depicts the DAENN architecture with the encoder and decoder.
Fig. 4.Deep autoencoder neural network (DAENN) architecture with encoder and decoder.


The following linear approximation of $Ψ^T$ is obtained by using Taylor's theorem at , halted to 1st order using Equation (8):

(8)


Since $x^{(t)}$ is a d-dimensional column-vector, it gives Equations (9) and Equation (10):

(9)


(10)


The above can be expressed as Equations (11) and (12):

(11)


(12)


Thus, providing Equations (13) and (14):

(13)


with

(14)


be a column-vector of input at time-stamp t, and considering Taylor expansion of yields Equation (15):

(15)


3.1.3 QoS prediction
Choose related K users as well as services for QoS forecast after computing similarity between users and services as shown in Algorithm 1. Prediction value is as follows on the user side by Equation (16).

(16)


where NS_u is the set of users as well as neighbor’s, and  $S_{u,v}$ denotes similarity between users u and v. In initial user-service invocation matrix R0, $q_{ui]$ denote the QoS value, and $N_i$ indicates the users who have called service i. The prediction value is computed as follows on the service side using Equation (17).

(17)




Spectrum Reuses and Network Optimization by FFT_RQL
Discrete Fourier transform on a finite group, which is a particular instance of a FT on a locally compact group, is evaluated by utilizing FFTs, which are low-complexity methods. For group is the bounded, continuous function using Equation (18).

(18)


The Fourier inversion formula can be used if certain assumptions are met, such that it belongs to totally integrals given using Equation (19).

(19)


The map for the fixed y is a character on Rr, i.e., a continuous group homomorphism from Rr into circle group The least common multiple given by Equations (20) and (21).

(20)


(21)


Let hence But then by Equation (22).
and hence

(22)


Consider that represents analogue of standard basis such that
Using periodicity given by Equation (23),

(23)


Consider that D2D pair k as well as CU m share the $m_{th}$ subcarrier. Task can be phrased as follows to maximize the spectrum efficiency of k and m by Equations (24), and (25):

(24)


(25)


As the D2D-$T_x$ k on the mth subcarrier’s minimum and extreme achievable powers, respectively,
should be satisfied in most cases. If , the D2D pair k and CU m cannot exchange resources. Using Equations (26) and (27), we can get the following:

(26)


(27)


By taking the first order derivative of the versus and setting the derivative to given using Equations (28), and (29).

(28)


(29)


The central unit learns internal state vector w(t) based on subsequent updating criteria to get the best access probability vector p* by Equations (30) and (31):

(30)


(31)


where λ is the proportion of times value of utility function is applied in average utility function at t time-slot. The Algorithm 2 is given for spectrum reuse based on proposed network design. The baseline is defined as the average utility function that allows for steady secondary network performance improvement in the reinforcement learning technique.
a) Spectrum sensing: The central unit collects interference values during this time in order to calculate the first access probability of STX signal transmitter. Central unit receives a measured aggregate interference value from every sensor closest to each STX. Sensor-measured aggregate interference is referred to as $S_k$.
b) Initializing: The access probability and the internal state vector are both initialized by the central unit. With the interference level Sk recorded by the kth sensors, the central unit defines the primary access probability p(0) of STXs, where k = 1, 2,..., N.
c)Access probability learning: The central unit is in charge of controlling the transmission of STXs throughout time steps t = 1, 2, T_l. The central unit gathers transmission results of STXs after they have been transmitted in order to determine the advantage of transmission of every STX.
d) Determining final access probability: Following the completion of the REINFORCE learning procedure, the central unit calculates the final access probability of STXs $p_l$ as follows using Equation (32):



Performance Analysis

Consider a scenario consisting of M = 50 APs and K = 30 MSs uniformly distributed randomly within a square of area of 11 km2. The experimental result is discussed and the parameters used for analysis are bit error rate (BER), signal-to-interference plus noise ratio (SINR), sum rate and spectral efficiency based on the comparative analysis with existing techniques.
Table 1 and Fig. 5 show a comparison of proposed and existing techniques for RA and spectrum allocation based on the number of users. The parameters measured in this case are BER, SINR, sum rate, and spectral efficiency. Based on the comparison above, the proposed technique obtained optimal results with BER of 40%, SINR of 72%, sum rate of 88%, and spectral efficiency of 98%, while the existing techniques GTA obtained BER of %, SINR of 73%, sum rate of 82%, spectral efficiency of 92.5%, and CBGP obtained BER of 42% SINR of 72.5%, sum rate of 84%, and spectral efficiency of 93%. The proposed technique yielded optimal results based on this parametric evaluation.

Table 1. Comparative analysis of proposed and existing techniques in resource selection and spectrum allocation based on the number of users (unit: %)

Technique BER SINR Sum rate Spectral efficiency
GTA 44 73 82 92.5
CBGP 42 72.5 84 93
DAENN_FFT-DQL 40 72 88 98

Fig. 5. Comparative analysis of proposed and existing technique in resource selection and spectrum allocation based on number of users: (a) BER and SINR, (b) sum rate and spectral efficiency.


Table 2 and Fig. 6 show a comparison of proposed and existing techniques for RA and spectrum allocation based on cell number. Based on the comparison above, the proposed technique achieved optimal results with BER of 45%, SINR of 76%, sum rate of 87%, and spectral efficiency of 97%, while the existing techniques GTA obtained BER of 53%, SINR of 74%, sum rate of 85%, spectral efficiency of 94% and CBGP obtained BER of 50%, SINR of 75%, sum rate of 86%, and spectral efficiency of 96%. The proposed technique yielded optimal results based on this parametric evaluation. Fig. 7 depicts a summary of the overall comparison.
The proposed approach for healthcare IoTs reduced computing delay, communication latency, and network latency, as well as network utilization and RAM consumption. The experimental results demonstrated that the proposed strategy for latency minimization using FC was more efficient.

Table 2. Comparative analysis of proposed and existing techniques in resource selection and spectrum allocation based on number of cells (unit: %)
Technique BER SINR Sum rate Spectral efficiency
GTA 53 74 85 94
CBGP 50 75 86 96
DAENN_FFT-DQL 45 76 87 97

Fig. 6. Comparative analysis of the proposed and existing techniques in resource selection and spectrum allocation based on number of cells: (a) BER and SINR, (b) sum rate and spectral efficiency.


Fig. 7. Overall comparative analysis of proposed resource selection and spectrum allocation techniques.



Conclusion

This study proposed a method for QoS-aware resource selection and spectrum reuse in heterogeneous networks. The goal here is to improve the network coverage and capacity of cellular methods, enabling an enhanced customer experience and collecting healthcare data via IoT. For IoT networks, MINLPbased FFT_RQL improves spectrum reuse. Then, we design QoS-aware resource selection using a DAENN to improve the weighted sum rate while meeting lower rate requirements and transmission power constraints for the IoT-based healthcare systems. The model is tested with and without prediction, and the results show increased bandwidth and low energy consumption, which results in the desired QoS in terms of bandwidth and energy utilization. According to the findings, the prediction model aids the system in allocating required resources based on traffic, applications, and events, thereby improving resource utilization and achieving the required QoS. The proposed technique yielded optimal results with BER of 40%, SINR of 72%, sum rate of 88%, and spectral efficiency of 98%. The proposed technique yielded optimal results based on this parametric evaluation. Analyze the dependability and security of future healthcare IoT data using various cryptographic procedures and methodologies can be explored.


Author’s Contributions

Conceptualization, RM, Indu. Funding acquisition, SAA, MSH. Investigation and methodology, RM. Resources, RM. Supervision, MSH, PT. Writing of the original draft, RM. Writing of the review and editing, MSH, SAA. Software, RM. Validation, VHCA.


Funding

The authors are grateful to the Deanship of Scientific Research at King Saud University for funding this work through the Vice Deanship of Scientific Research Chairs: Research Chair of New Emerging Technologies and 5G Networks and Beyond.


Competing Interests

The authors declare that they have no competing interests.


Author Biography

Author
Manikandan Ramachandran obtained his Ph.D. in VLSI Physical design from SASTRA Deemed University, India. He received his Bachelor of Engineering in Computer Science from Bharathidasan University and Master of Technology in VLSI Design from SASTRA Deemed University. He possesses three decades of academic and 15 years of research experience in the field of Computer Science and Engineering. He has more than 200 research contributions to his credit, which are published in referred and indexed journals, book chapters and conferences. He is presently working as Senior Assistant Professor at SASTRA Deemed University for the last 15 years.

Author
Indu received Ph. D. in Electronics and Communication engineering from Galgotias University, Greater Noida, in wireless communication especially in securing vehicular communication with trust model. She has 6 years of experience in teaching and currently she is serving as an Assistant Professor in the Department of ECE at Galgotias University, Greater Noida, U.P., India. She has more than 20 publications in various indexed journals/conferences. She has published 3 patents. With specialization in wireless Communication her field of Interest includes vehicular Ad-hoc Networks, Internet of Things, Internet of vehicles, security techniques, Smart Cities, Healthcare, Transportation and Wireless Technologies.

Author
Victor Hugo C. de Albuquerque is a professor and senior researcher in the Department of Teleinformatics Engineering (DETI) at the Federal University of Ceará (UFC), Brazil. He earned a Ph.D. in mechanical engineering from the Federal University of Paraíba (UFPB, 2010), and a M.Sc. in teleinfor- matics engineering from the PPGETI/UFC (UFC, 2007). He completed a BSE in mechatronics engineering at the Federal Center of Technological Education of Ceará (CEFETCE, 2006). He specializes in image data science, IoT, machine/deep learning, pattern recognition, automation and control, and robotics.

Author
Prayag Tiwari received his Ph.D. Degree in Information Engineering from the University of Padova, Italy. He is currently working as a Postdoctoral Researcher at the Aalto University. Previously, he was working as a Marie Curie Researcher at the University of Padova, Italy. He also worked as a Research Assistant at NUST MISIS, and he has had Teaching and Industrial work experience. He has several publications in top journals and conferences including Neural Networks, Information Fusion, IPM, IJCV, IEEE TNNLS, IEEE TFS, IEEE TII, IEEE JBHI, IEEE IOTJ, IEEE BIBM, ACM TOIT, CIKM, SIGIR, AAAI, etc. His research interests include Machine Learning, Deep Learning, Quantum Machine Learning, Information Retrieval, Health Informatics, and IoT.

Author
Salman Ali AlQahtani is currently a Full professor at the department of computer engineering, college of computer and information sciences, King Saud University, Riyadh, Saudi Arabia. He serves also as a senior consultant in computer communications, integrated solutions and digital forensics for few development companies and government sectors in Saudi Arabia. Dr AlQahtani’s main research activities are in Radio Resource Management (RRM) for wireless and cellular networks (4G, 5G, IoT, Industry 4.0, LTE, LTE-Advanced, Femtocell, Cognitive radio, Cyber Sovereignty...) with focus on Call Admission Control (CAC), Packet Scheduling, radio resource sharing and Quality-of-Service (QoS) guarantees for data services. In addition, his interests also include performance evaluation of packet switched network, system model and simulations and integration of heterogeneous wireless networks. Finally, my interests also extend to the area of digital forensics

M. Shamim Hossain is currently a Professor with the Department of Software Engineering, College of Computer and Information Sciences, King Saud University, Riyadh, Saudi Arabia. He is also an adjunct professor with the School of Electrical Engineering and Computer Science, University of Ottawa, ON, Canada. He received his Ph.D. in Electrical and Computer Engineering from the University of Ottawa, ON, Canada in 2009. His research interests are on cloud networking, smart environment (smart city, smart health), AI, deep learning, edge computing, Internet of Things (IoT), multimedia for health care, and multimedia big data. He has authored and coauthored more than 325 publications including refereed journals, conference papers, books, and book chapters. Recently, he co-edited a book on “Connected Health in Smart Cities”, published by Springer. He has served as the cochair, general chair, workshop chair, publication chair, and TPC in several IEEE and ACM conferences. He is the chair of IEEE Special Interest Group on Artificial Intelligence (AI) for Health with IEEE ComSoc eHealth Technical Committee. Currently, he is the Organizing Co-Chair of the Special Sessions with IEEE I2MTC 2022. He was also the Co-Chair of the 1st IEEE GLOBECOM 2021 Workshop on Edge-AI and IoT for Connected Health. He was the Co-Chair of the special session “AI- Enabled technologies for smart health monitoring", held with IEEE I2MTC 2021. He is the Technical Program Co-Chair of ACM Multimedia 2023. Currently, he is the Chair of Saudi Arabia Section of the Instrumentation and Measurement Society Chapter. He is a recipient of a number of awards, including the Best Conference Paper Award and the 2016 ACM Transactions on Multimedia Computing, Communications and Applications (TOMM) Nicolas D. Georganas Best Paper Award, the 2019 King Saud University Scientific Excellence Award (Research Quality), and the Research in Excellence Award from the College of Computer and Information Sciences (CCIS), King Saud University (3 times in a row). He is on the editorial board of the IEEE Transactions on Instrumentation and Measurement (TIM), IEEE Transactions on Multimedia (TMM), ACM Transactions on Multimedia Computing, Communications, and Applications (TOMM), IEEE Multimedia, IEEE Network, IEEE Wireless Communications, IEEE Access, Journal of Network and Computer Applications (Elsevier), International Journal of Multimedia Tools and Applications (Springer), Games for Health Journal, and International Journal of Information Technology, Communications and Convergence (Inderscience). Previously, he served as a guest editor of ACM Transactions on Internet Technology, IEEE Communications Magazine, IEEE Network, IEEE Transactions on Information Technology in Biomedicine (currently JBHI), IEEE Transactions on Cloud Computing, International Journal of Multimedia Tools and Applications (Springer), Cluster Computing (Springer), Future Generation Computer Systems (Elsevier), Elsevier), Sensors (MDPI), and International Journal of Distributed Sensor Networks. He is a senior member of the IEEE, the Distinguished Member of the ACM. He is an IEEE Distinguished Lecturer (DL).



References

[1] A. Kishor, C. Chakraborty, and W. Jeberson, “A novel fog computing approach for minimization of latency in healthcare using machine learning,” International Journal of Interactive Multimedia and Artificial Intelligence, vol. 6, no. 7, pp. 7-17, 2021.
[2] A. Ramirez-Arroyo, P. H. Zapata-Cano, A. Palomares-Caballero, J. Carmona-Murillo, F. Luna-Valero, and J. F. Valenzuela-Valdes, “Multilayer network optimization for 5G & 6G,” IEEE Access, vol. 8, pp. 204295-204308, 2020.
[3] S. T. Muntaha, S. A. Hassan, H. Jung, and M. S. Hossain, “Energy efficiency and hover time optimization in UAV-based HetNets,” IEEE Transactions on Intelligent Transportation Systems, vol. 22, no. 8, pp. 5103-5111, 2021.
[4] K. Wang, C. M. Chen, M. S. Hossain, G. Muhammad, S. Kumar, and S. Kumari, “Transfer reinforcement learning-based road object detection in next generation IoT domain,” Computer Networks, vol. 193, article no. 108078, 2021.https://doi.org/10.1016/j.comnet.2021.108078
[5] M. S. Hossain, and G. Muhammad, “Emotion-aware connected healthcare big data towards 5G,” IEEE Internet of Things Journal, vol. 5, no. 4, pp. 2399-2406, 2018.
[6] G. Muhammad, M. S. Hossain, and N. Kumar, “EEG-based pathology detection for home health monitoring,” IEEE Journal on Selected Areas in Communications, vol. 39, no. 2, pp. 603-610, 2021.
[7] A. Kishor, C. Chakraborty, and W. Jeberson, “Reinforcement learning for medical information processing over heterogeneous networks,” Multimedia Tools and Applications, vol. 80, no. 16, 23983-24004, 2021.
[8] N. S. Sworna, A. K. M. M. Islam, S. Shatabda, and S. Islam, “Towards development of IoT-ML driven healthcare systems: a survey,” Journal of Network and Computer Applications, vol. 196, article no. 103244, 2021. https://doi.org/10.1016/j.jnca.2021.103244
[9] A. Yassine, S. Singh, M. S. Hossain, and G. Muhammad, “IoT big data analytics for smart homes with fog and cloud computing,” Future Generation Computer Systems, vol. 91, pp. 563-573, 2019.
[10] A. Kishor and C. Chakraborty, “Artificial intelligence and Internet of Things based healthcare 4.0 monitoring system,” Wireless Personal Communications, 2021. https://doi.org/10.1007/s11277-021-08708-5
[11] X. Zhou, W. Liang, K. I. K. Wang, H. Wang, L. T. Yang, and Q. Jin, “Deep-learning-enhanced human activity recognition for Internet of healthcare Things,” IEEE Internet of Things Journal, vol. 7, no. 7, pp. 6429-6438, 2020.
[12] M. S. Hossain and G. Muhammad, “Deep learning based pathology detection for smart connected healthcare,” IEEE Network, vol. 34, no. 6, pp. 120-125, 2020.
[13] M. Shorfuzzaman and M. S. Hossain, “MetaCOVID: a Siamese neural network framework with contrastive loss for n-shot diagnosis of COVID-19 patients,” Pattern Recognition, vol. 113, article no. 107700, 2021. https://doi.org/10.1016/j.patcog.2020.107700
[14] M. S. Hossain, G. Muhammad, and N. Guizani, “Explainable AI and mass surveillance system-based healthcare framework to combat COVID-I9 like pandemics,” IEEE Network, vol. 34, no. 4, pp. 126-132, 2020.
[15] M. A. Rahman, M. S. Hossain, N. A. Alrajeh, and N. Guizani, “B5G and explainable deep learning assisted healthcare vertical at the edge: COVID-I9 perspective,” IEEE Network, vol. 34, no. 4, pp. 98-105, 2020.
[16] M. A. Rahman, M. S. Hossain, A. J. Showail, N. A. Alrajeh, and M. F. Alhamid, “A secure, private, and explainable IoHT framework to support sustainable health monitoring in a smart city,” Sustainable Cities and Society, vol. 72, article no. 103083, 2021. https://doi.org/10.1016/j.scs.2021.103083
[17] M. Shorfuzzaman, M. S. Hussain, and M. F. Alhamid, “Towards the sustainable development of smart cities through mass video surveillance: a response to the COVID-19 pandemic,” Sustainable Cities and Society, vol. 64, article no. 102582, 2021. https://doi.org/10.1016/j.scs.2020.102582
[18] A. Rahman, M. S. Hossain, N. A. Alrajeh, and F. Alsolami, “Adversarial examples—security threats to COVID-19 deep learning systems in medical IoT devices,” IEEE Internet of Things Journal, vol. 8, no. 12, pp. 9603-9610, 2021.
[19] M. A. Rahman, M. M. Rashid, M. S. Hossain, E. Hassanain, M. F. Alhamid, and M. Guizani, “Blockchain and IoT-based cognitive edge framework for sharing economy services in a smart city,” IEEE Access, vol. 7, pp. 18611-18621, 2019.
[20] A. Makkar, S. Garg, N. Kumar, M. S. Hossain, A. Ghoneim, and M. Alrashoud, An eff”icient spam detection technique for IoT devices using machine learning,” IEEE Transactions on Industrial Informatics, vol. 17, no. 2, pp. 903-912, 2021.
[21] Y. Lin, X. Wang, and R. Xu, “Semi-supervised human resource scheduling based on deep presentation in the cloud,” EURASIP Journal on Wireless Communications and Networking, vol. 2020, article no. 73, 2020. https://doi.org/10.1186/s13638-020-01677-6
[22] Y. F. Huang, C. B. Lin, C. M. Chung, and C. M. Chen, “Research on QoS classification of network encrypted traffic behavior based on machine learning,” Electronics, vol. 10, no. 12, article no. 1376, 2021. https://doi.org/10.3390/electronics10121376
[23] Y. Yin, W. Zhang, Y. Xu, H. Zhang, Z. Mai, and L. Yu, “QoS prediction for mobile edge service recommendation with auto-encoder,” IEEE Access, vol. 7, pp. 62312-62324, 2019.
[24] Y. Hao, M. Chen, L. Hu, M. S. Hossain, and A. Ghoneim, “Energy efficient task caching and offloading for mobile edge computing,” IEEE Access, vol. 6, pp. 11365-11373, 2018
[25] G. D’Angelo and F. Palmieri, “Network traffic classification using deep convolutional recurrent autoencoder neural networks for spatial-temporal features extraction,” Journal of Network and Computer Applications, vol. 173, article no. 102890, 2021. https://doi.org/10.1016/j.jnca.2020.102890
[26] H. Cha and S. L. Kim, “A reinforcement learning approach to dynamic spectrum access in Internet-of-Things networks,” in Proceedings of 2019 IEEE International Conference on Communications (ICC), Shanghai, China, 2019, pp. 1-6.
[27] K. F. Muteba, K. Djouani, and T. O. Olwal, “Deep reinforcement learning based resource allocation for narrowband cognitive radio-IoT systems,” Procedia Computer Science, vol. 175, pp. 315-324, 2020.
[28] M. Beshley, N. Kryvinska, M. Seliuchenko, H. Beshley, E. M. Shakshuki, and A. U. H. Yasar, “End-to-end QoS ‘smart queue’ management algorithms and traffic prioritization mechanisms for narrow-band Internet of Things services in 4G/5G networks,” Sensors, vol. 20, no. 8, pp. article no. 2324, 2020.
[29] W. U. Rehman, T. Salam, A. Almogren, K. Haseeb, I. Ud Din, and S. H. Bouk, “Improved resource allocation in 5G MTC networks,” IEEE Access, vol. 8, pp. 49187-49197, 2020.
[30] C. Abid, M. Kessentini, and H. Wang, “Early prediction of quality of service using interface-level metrics, code-level metrics, and antipatterns,” Information and Software Technology, vol. 126, article no. 106313, 2020. https://doi.org/10.1016/j.infsof.2020.106313
[31] J. O. Mebawondu, F. M. Dahunsi, and O. S. Adewale, “Hybrid intelligent model for real time assessment of voice quality of service,” Scientific African, vol. 9, article no. e00491, 2020. https://doi.org/10.1016/j.sciaf.2020.e00491
[32] L. Eslami, G. Mirjalily, and T. N. Davidson, “Spectrum-efficient QoS-aware resource assignment for FFR-based D2D-enabled heterogeneous networks,” IEEE Access, vol. 8, pp. 218186-218198, 2020.
[33] M. H. Bahonar and M. J. Omidi, “Centralized QoS-aware resource allocation for D2D communications with multiple D2D pairs in one resource block,” 2021 [Online]. Available: https://arxiv.org/abs/2112.10494.
[34] Q. Du and X. Zhang, “QoS-aware base-station selections for distributed MIMO links in broadband wireless networks,” IEEE Journal on Selected Areas in Communications, vol. 29, no. 6, pp. 1123-1138, 2011.
[35] T. G. Prasad and M. S. S. Rukmini, “QoS sensible coalition-based radio resource management scheme for 4G mobile networks” in Mobile Radio Communications and 5G networks. Singapore: Springer, 2021, pp. 745-755.
[36] T. Kim and J. M. Chang, “QoS-aware energy-efficient association and resource scheduling for HetNets,” IEEE Transactions on Vehicular Technology, vol. 67, no. 1, pp. 650-664, 2018.
[37] M. S. Hossain, "Cloud-Supported Cyber–Physical Localization Framework for Patients Monitoring," IEEE Systems Journal, vol. 11, no. 1, pp. 118-127, March 2017.
[38] X. Ma, T. Yao, M. Hu, Y. Dong, W. Liu, F. Wang, and J. Liu, “A survey on deep learning empowered IoT applications,” IEEE Access, vol. 7, pp. 181721-181732, 2019.
[39] R. W. Liu, J. Nie, S. Garg, Z. Xiong, Y. Zhang, and M. S. Hossain, “Data-driven trajectory quality improvement for promoting intelligent vessel traffic services in 6G-enabled maritime IoT systems,” IEEE Internet of Things Journal, vol. 8, no. 7, pp. 5374-5385, 2021.
[40] M. M. Salim, V. Shanmuganathan, V. Loia, and J. H. Park, “Deep learning enabled secure IoT handover authentication for blockchain networks,” Human-centric Computing and Information Sciences, vol. 11, article no. 21, 2021. https://doi.org/10.22967/HCIS.2021.11.021
[41] S. Rathore, J. H. Park, and H. Chang, “Deep learning and blockchain-empowered security framework for intelligent 5G-enabled IoT,” IEEE Access, vol. 9, pp. 90075-90083, 2021.

About this article
Cite this article

Ramachandran Manikandan1, Indu2, Victor Hugo C. de Albuquerque3, Prayag Tiwari4, Salman Ali AlQahtani5,* , and M. Shamim Hossain6,*, Quality of Service-Aware Resource Selection in Healthcare IoT Using Deep Autoencoder Neural Networks, Article number: 12:36 (2022) Cite this article 3 Accesses

Download citation
  • Received13 February 2022
  • Accepted13 March 2022
  • Published15 August 2022
Share this article

Anyone you share the following link with will be able to read this content:

Provided by the Springer Nature SharedIt content-sharing initiative

Keywords