International Journal of Intelligent Engineering and Systems

11

Authors: Haneen Qteat, Mohammed Awad

Diabetes mellitus is one of the deadliest and chronic diseases that affect persons who have an increase in their blood glucose levels. Type 1 Diabetes Mellitus “T1DM” is considered one of the most dangerous types of diabetes as it is the reason why diabetes is called the silent killer. Due to the common symptoms of type 1 and 2 diabetes, diabetes specialists face doubts about their diagnosis of the type of diabetes in the patient where the uncertainty about the diagnosis of the disease may lead to delays in controlling the potential complications, especially if they have T1DM. The prime motto of this work is to classify the diabetes types accurately. In this work, we have collected a local Palestinian dataset “DataPal” With the assistance of the Palestinian Diabetes Institute. The "DataPal" dataset was applied using machine learning algorithms to predict diabetes types. The “DataPal” consists of 9 predictors used to diagnose diabetes types. The dataset consists of 314 instances of diabetic females. Thus, our samples were for females with both types 1 and 2 diabetes, aged between 5 and 89 years. The local dataset “DataPal” was preprocessed using the K Nearest Neighbor (KNN) algorithm to fill their missing values, wherein medical diagnosis there is no room for error. The Support Vector Machine (SVM) algorithm was applied to the dataset to select the most optimal features to predict diabetes types. Both the two-fold and four-fold cross-validation methods were applied to the datasets to evaluate the applied models fairly. A hybrid Model Particle Swarm Optimization with Multi-Layer Perceptron Neural Networks (PSO-MLPNNs) uses the PSO evolutionary algorithm to train an MLPNNs and find the optimal weight values of the trained network. The PSO-MLPNNs model was applied to the preprocessed dataset. Then the performance of the model was evaluated using different metrics such as the overall accuracy, recall, specificity, and others. The obtained results show that the proposed PSO-MLPNNs model outperformed all models applied in this work in the classification of diabetes types with an overall accuracy (ACC= 98.73%).

31

Authors: Shimaa Ouf

This paper introduces an architecture to improve the pharmaceutical supply chain's security by using the Internet of Things, semantic web, and blockchain. This architecture increases transparency and visibility of drug flows and improves the representation of the increasing amounts of data that have been generated from pharmaceutical supply chain transactions. The pharmaceutical companies have problems dealing with the complexity of their supply chains, which represent the full life cycle of drugs from extracting raw materials, production, distribution, tracking of drugs, and quality assurance to use by patients. They need to create an efficient, effective, transparent, immutable, and secured supply chains to achieve a competitive advantage in a fast-changing market. A proposed architecture which applies the semantic web technology is implemented to enhance the representation capability of the IoT-blockchain based pharmaceutical supply chain data by annotating them with semantically rich languages to conduct formal reasoning, and aggregating data from heterogeneous sources in easy way and an interoperable manner. The proposed architecture integrates IoT, blockchain, and semantic web to help pharmaceutical companies improving their supply chains in transit and storage, improving patient satisfaction, trust through transparency, preventing drug counterfeit and sharing and reusing knowledge related to the pharmaceutical supply chain with other systems.

54

Authors: Kun Nursyaiful Priyo Pamungkas, Waskitho Wibisono, Supeno Djanali

Wireless sensor networks (WSN) have been widely applied to monitor objects in harsh and hazardous environments. The limited energy resource in WSN is a major problem. Energy efficiency is required for WSN to operate for a long time. The cluster in WSN is a strategy that can help to extend the life of the network. However, the optimal cluster formation and cluster head (CH) selection is a non-deterministic polynomial-time hard problem. Besides, many clustering-based protocols build clusters after defining the cluster heads among all sensor nodes on the network. This process triggers inefficient energy consumption. Furthermore, the unbalanced load between CH and cluster members (CMs) also affects the network lifetime. This paper proposes an advanced clustering protocol based on a modified differential search algorithm to save energy consumption and extend network life. To form clusters, this protocol determines the optimal number of clusters. In this proposed protocol, CH selection takes place at the cluster level. The process of cluster formation and CH selection is proposed by utilizing a modified differential search algorithm. Furthermore, the rotation system and the determination of relay nodes are proposed to control the load distribution. The proposed protocol's performance is compared with various data gathering protocols for WSN, namely LEACH, PEGASIS, LEACH-GA, and HFAPSO. The experimental results show that the proposed protocol can increase network lifetime in terms of First Node Dies (FND), Half of Nodes Die (HND), and Last Node Dies (LND). On average, the MDSAC protocol increases FND by 1596 rounds, 2456 rounds, 1211 rounds, and 1060 rounds compared to LEACH, PEGASIS, LEACH-GA, and HFAPSO. The MDSAC protocol also increases the average HND by 5764 rounds, 4918 rounds, 5221 rounds, and 4308 rounds compared to other comparison protocols. Even the MDSAC protocol managed to improve the average LND of 7784 rounds, 8739 rounds, 9001 rounds, and 7161 rounds, compared to the four comparison protocols. These results highlight the contribution of the proposed approach in maintaining network lifetime in WSN as one of the important challenges in WSN research and development.

82

Authors: Yohanes Artha Setiawan, Amirullah Amirullah

This paper aims to design and implement a single-phase dynamic voltage restorer (DVR) supplied by battery energy storage (BES) using load voltage controlled by the unit vector template generation (UVTG) method. The UVTG method on the DVR system is implemented using the Arduino-Uno hardware. LabVIEW-based interface simulation is used for monitoring the parameter in real-time during 80% voltage sag period i.e. source voltage, injection voltage, load voltage, source current, load current, and DC-link voltage. The true power factor measurements are also carried out to obtain total harmonic distortion (THD) of the source current and load current. The single-phase DVR system is connected to load i.e. a 5-watt bulb lamp (linear-load), 5-watt fluorescent (FL) lamp (non-linear load), and 5-watt light-emitting diode (LED) lamp (non-linear load). During voltage sag, the single-phase DVR-BES system is able to maintain load voltage i.e. bulb lamp, FL lamp, and LED lamp of 216 volts, 256 volts, and 185 volts, respectively. The percentage of load sag voltage deviation for each load is 6.4%, 26.13%, and 8.87%, respectively. The measurement of source power-factor with voltage sag results in source true power factor of three loads of 0.985 pu leading, 0.985 pu leading, and 0.990 pu leading respectively. On the same system with voltage sag, single phase DVR-BES is able to result in source current THD for three loads of 1.523%, 1.523%, and 1.010% respectively. The proposed study is able to give the best performance because it is able to produce source current THD lower than the previous studies. By using the Arduino-Uno interface, this model can be run and monitored in real-time with LabVIEW better from the simulations conducted by previous researchers which were still carried out off-line using Matlab/Simulink.

109

Authors: Ayomi Sasmito, Asri Bekti Pratiwi

This paper proposes an efficient optimization algorithm called Chaotic Student Psychology Based Optimization (CSPBO) to solve bi-objective permutation flowshop scheduling problem (BPFSP). The SPBO algorithm does not require any tunning parameters which makes it simpler in computational experiments. The original SPBO classifies students related to their efforts in improving their performances. In this paper, we use chaotic maps to enhance student efforts in each category and present different strategic approaches. Logistic, iterative, sine, tent, and singer maps are integrated using proposed strategies to find the best map and also its strategy, named CSPBO. To prove its performance, the CSPBO is compared with several well-known metaheuristic algorithms. The comparison results are evaluated in regard to mean, standard deviation, best terms and ARPD. Since BPFSP is an NP-hard problem, computational experiments on BPFSP instances are carried out according to type size of the dataset. Three types of dataset, small, medium and large are tested. Based on the experiment result, the SPBO integrated with logistic map has significantly given better performance compared to other chaotic maps. Furthermore, CSPBO shows compatible performance compared with other metaheuristic algorithms in solving BPFSP.

128

Authors: Dheyaa T. Al-Zuhairi, Abbas Salman Hameed, Isam Salah Hameed

In this paper, a system for the purpose of signals encryption using the technique of independent component analysis has been proposed. The proposed system mixes the original signal with arbitrary number of random signals in order to obtain a highly encrypted signal like noise. The number of random signals is indicated by an attached key signal preceding the encrypted signal. In order to increase the encryption strength, the number of random signals is changed at each signal transmission. An independent component analysis technique is utilized to separate signals and select the original signal by a designed selector. A set of parameters has been adopted to measure the encryption quality and decryption capability. Promising encryption results represented by up to - 48.35 dB for segmental spectral signal to noise ratio and 0.00037 correlation coefficient between the original and the encrypted signals are obtained. As well, an efficient signal decryption is achieved with 45.85 dB signal to noise ratio and nearly 1 correlation coefficient value between the original and reconstructed signals. In conclusion, the proposed system is a good candidate to be adopted in developing highly reliable security transceiver systems.

150

Authors: Ameer A. Badr, Alia K. Abdul-Hassan

Lately, with the rapid growth of various technologies, identifying the information of gender and ‎age give short speech utterances has become a necessity for many applications in daily life like ‎human-robot interaction, targeted ‎marketing, identifying suspects in criminal cases, etc.‎‏ ‏Despite ‎the comprehensive studies carried out to extract descriptive features, the recognition accuracy is ‎still not satisfactory.‎‏ ‏In this study, an automatic system is proposed to classify age and gender in ‎short speech utterances without depending on the text. Firstly, two groups of features are ‎extracted from each utterance frame, followed by measuring ‎10 statistical functionals for each ‎extracted feature dimension. After that, the extracted features dimensions are normalized using ‎the Quantile technique. Then, the CatBoost machine is utilized as an important features detection ‎to select the most discriminatory features for speaker age and gender recognition tasks. For ‎classification purposes, the selected feature dimensions are fed into the Support Vector Machine ‎‎(SVM). Experiments are conducted on the aGender data-set for measuring the suggested ‎system's performance. The unweighted accuracies (UA) of the proposed system for gender, age, ‎in addition to gender & age is ‎89.62%, 72.29%, and 71.96%, respectively. The achieved results ‎outperform recent results on the same data-set.

177

Authors: Mohammed El Mahfoud, Badre Bossoufi, Najib El Ouanjli, Mahfoud Said, Mohammed Taoussi

This research paper deals with the improved direct torque control (DTC) of a doubly fed induction machine (DFIM) operating in motor mode. The conventional DTC is a powerful tool to ensure robustness and good torque dynamics. However, the variable switching frequency as well as the presence of torque and flux ripples due to the use of hysteresis controllers are the major drawbacks of this strategy. For this purpose, the control by space vector Modulation (SVM) is developed to improve DTC performance, by replacing hysterisis controllers by PI controllers to reduce electromagnetic flux ripple and torque ripple in order to minimize mechanical vibration and acoustic noise while maintaining the benefits of DTC control. The proposed control algorithm is simulated and tested in MATLAB/Simulink. A comparative analysis between this technique and conventional DTC is carried out, highlighting the efficiency of the proposed improvement approach. The performance indexes and comparative results show the effectiveness of the proposed control in reducing torque ripple and stator and rotor flux ripple by up to 43%, 42% and 31%, respectively, resulting in Total Harmonic Distortion (THD%) 61% lower, in addition the switching frequency is controlled, and the Rejection Time is improved by more than 76%.

201

Authors: Mahmoud O. Elish, Karim Elish

Software defect prediction is one of the most important quality assurance activities during software development. This paper contributes empirical insights into the effectiveness of three resampling ensemble methods (bagging, boosting, and dagging) of Deep Learning Neural Networks (DLNN) for cross-project software defect prediction, compared to individual DLNN. An empirical study was conducted using five datasets. The results indicate that the bagging ensembles of DLNN offer only 0.24% increase in accuracy, on average, compared to the individual DLNN models, whereas the boosting and dagging ensembles degrade the accuracy. Furthermore, the results show that the three resampling ensembles of DLNN outperform the individual DLNN models in precision; with a maximum improved precision by 25.15% on average using the boosting ensembles. The results however indicate that none of the resampling ensembles improve the recall. Lastly, for a balanced performance in terms of both precision and recall, the results indicate improvements ranging from 0.98% on average by applying the bagging ensembles to 11.67% on average by applying the boosting ensembles.

222

Authors: Muhammad Fuad, Trihastuti Agustinah, Djoko Purwanto

The need for mobile robots that can operate in the human environment is increasing during the Covid-19 pandemic. To accomplish this task, robot navigation must be supported by collision avoidance to maintain human safety and comfort. Collision avoidance methods generally only maintain a safe distance from surrounding objects. Some methods that provide comfort still have difficulty in overcoming pedestrians that move with unexpected direction, changing speed, and unknown trajectories. This paper proposes Hybrid Velocity Obstacle-based modified Headed Social Force Model (HVO-based modified HSFM) to avoid disturbed groups of pedestrians while navigating in the complex and dense workspace. HVO is used to calculate linear and angular velocities to avoid obstacles with nonlinear trajectories. The linear velocity that computed by HVO acts as desired velocity for generating target force of modified HSFM that drive robot to target location. While interaction force of modified HSFM that guide robot to circumvent obstacles is determined by static and moving objects in the surrounding of robot. The angular velocity from HVO is used to produce steering command to avoid collision. For evaluating the proposed method, several simulation scenarios had been run by implementing HVO-based modified HSFM into a two-wheeled differential-steering mobile robot that navigate in the indoor human environment. The results show that our approach is capable to avoid collision by maintaining safety and comfort with disturbed groups of pedestrians with average value 0.14 of Threat Level Index (TLI) from two scenarios. It is envisioned that proposed method can be implemented into real transport robot that operate in human environment.

255

Authors: Salah Mortada, Yuhanis Yusof

Artificial bee colony (ABC) is one of the widely used swarm intelligence algorithms in solving combinatorial optimization problems. In this study, the existing Modified ABC (MABC) algorithm is revised to solve the vehicle routing problem with time windows (VRPTW). The reason is that even though MABC is reported to be successful, its exploitation process lacks a selection neighborhood structure during the intensification process. The proposed algorithm, termed as enhanced Modified ABC (E-MABC), offers a neighborhood search that exchanges neighborhood structure between two different routes in the same solution, rather than one route. To evaluate the effectiveness of E-MABC, experiments are performed on 56 instances of VRPTW. The results of E-MABC are compared against the ones produced using MABC and other metaheuristic algorithms. Based on the total travel distance and number of vehicles, the proposed E-MABC is shown to be a solution for VRPTW. The interchange neighborhood search, which is implemented by bees during the exploitation process, improves the solution quality, hence producing an optimal outcome. The proposed E-MABC results are better as compared to MABC in terms of the total travel distance by 71.42 % and the number of vehicles by 35.71%.

282

Authors: Rafika El idrissi, Ahmed Abbou, Mohcine Mokhlis, Mahdi Salimi

The environmental condition changes lead to obvious fluctuation in photovoltaic panels’ output power. Therefore, to make efficient use of photovoltaic (PV) systems the maximum power point tracking (MPPT) controllers are required. Many classical methods are proposed to track the MPP, but they will lead to a high power drop when rapid changes in the atmospheric conditions occur, which necessities a robust controller with high performance. In such a manner, the proposed controller is designed for this purpose. There are two stages of the proposed controller: The artificial neural network (ANN) based the first stage that generates the PV panel optimal voltage and the second phase consists of a non-linear adaptive backstepping control, which is able to follow this optimum voltage by acting on the DC/DC boost converter’s duty cycle. The input-output linearization technique is based the suggested controller. The last is robust and safe from the parameters fluctuation, load variation, and the atmospheric condition changes. In the proposed control design, unknown and estimated converter parameters are assumed, especially the inductor and input capacitor, considering an adequate Lyapunov function. A single-phase inverter connects the boost converter to the grid. However, for this connection to be made the power factor should be unified and the inverter current should be synchronized with the grid voltage. The sliding mode controller (SMC) is designed to operate on the inverter duty cycle for the resolution of these tasks. The controller is used to enhance the robustness and the system's rapidity. Besides, the DC bus is regulated using the proportional and integral (PI) controller. Matlab/Simulink software is used to simulate the overall system. Furthermore, for accurate results, the proposed controller is compared with perturb & observe (P&O), ANN-backstepping sliding mode (ANN-BSMC), and ANN-backstepping integral sliding mode (ANN-BISMC) techniques . The findings show that the proposed method outweighs other methods in terms of tracking rapidity, steady-state error and the fluctuations around the MPP under severe assumptions.

303

Authors: Hanaa Khater, Ahmed ElSawy, Assem Tharwat, Ihab El-Khodary

The max-min separable optimization problem is depended on max-min algebra. This optimization problem focuses on the worst case for minimization of objective function. Additionally, this problem has separable max-min constraints. It was solved in another article by using difficult and complicated algebraic method for finding one solution only. In this paper, the Max-Min Separable problem under Max-Min Separable Linear Constraints (MMSMMLC) is solved by hybrid PSO. The main idea of the presented algorithms is combining GA processes with PSO algorithm to be suitable for solving MMSMMLC optimization problem under all types of constraints (equality and/or inequality). Finally, the new algorithms applied through several benchmark functions show that the new algorithms present all optimal solutions in best result with 100% and improve the optimal solution in some cases rather than algebraic method.

332

Authors: Hassan Almazini, Ku Ruhana Ku-Mahamud

Modified graph clustering ant colony optimization (MGCACO) algorithm is an unsupervised feature selection (UFS) algorithm used in determining a subset of effective genes from microarray data. The feature subset construction is based on the ant colony optimization (ACO) algorithm, which guides the search process from clusters. However, the MGCACO algorithm is unable to choose all significant features from the clusters to form an optimal feature subset. This paper proposes an enhanced graph clustering ACO (EGCACO) to overcome the problem of feature selection in the MGCACO algorithm. A principal point of this algorithm is utilizing an adaptive selection technique that guides ACO for subset construction from the clusters of features. The adaptive technique for ant selection is based on the state of the search space. Experimental results indicated that the proposed EGCACO achieves the highest classification accuracy than five other common UFS algorithms on four classifiers, where it obtained 87.13%, 86 .19 %, 87.38 % and 90.80 % for support vector machine, k-nearest neighbor, decision tree and random forest classifiers, respectively. In particular, the proposed algorithm can select the genes of the deoxyribonucleic acid microarray with consideration of relevance and redundancy among the genes. Therefore, the proposed EGCACO can be implemented to handle the high dimension feature space, such as image processing, text classification, and microarray data processing, which is critical for good and reliable results.

359

Authors: Dinial Utami Nurul Qomariah, Handayani Tjandrasa, Chastine Fatichah

Diabetic retinopathy is a disease caused by complications of chronic diabetes and can lead to blindness. Therefore, automatic detection of microaneurysms on retinal fundus images as an early warning system is very necessary and challenging since the size of the microaneurysm is very small compared to the anatomical structures of the retina such as blood vessels and other elements, and hemorrhages as other red lesions. In this study, we propose a novel deep learning network that modifies UNet using residual units with modified identity mapping (MResUNet) to perform microaneurysm segmentation. The purpose of identity mapping modification using convolutional layers and batch normalization is to enrich features and add Residuals in UNet to overcome feature degradation as the network becomes deeper. The mean weighted loss function is used in training to solve the problem of pixel imbalance between the microaneurysms and the background. The proposed architecture is evaluated using the IDRID and DiaretDB1 datasets. The experimental results show that the proposed architecture (MResUNet) achieves higher sensitivity values of 61.96% and 85.87% on the IDRID and DiaretDB1 datasets, respectively, compared to AutoEncoder, FCN16, FCN8, and UNet.

388

Authors: Lamia Moudoubah, Kamal El Guemmat, Khalifa Mansouri, Mohammed Qbadou

In terms of mental health, COVID-19 has a great impact on people's psychology. It can cause a certain degree of fear, worry, boredom, depression, eating disorders, loneliness and sometimes self-aggression or suicidal behaviour. The purpose of this article is to explore the psychological consequences of health problems related to COVID-19. During this period, some literature appears to analyze the psychological behaviour of individuals. However, in the sense of domain ontology, there is a lack of knowledge about modelling specific knowledge in the field of psychology and to have formal semantics. Through this article, we present a psychology-based method to analyze human behaviour and frailty during COVID-19. We chose IR (Information Research) based on TOR (Termino-Ontology Resources) as the method to construct the resulting ontology modelling the state of mental health in a COVID-19 context, an ontology that avoids ambiguities in the concepts that model the domain in question. The result of our research is the ontology described in Fig. 5, 6, 7, 8 and 9. The concepts that make up our ontology are derived from the results of the calculation of scores through Eq. 6, which consists of calculating the Score (Cij), then the CI(c) through the calculation of P(c) which is 0.18 in the term ti= anxiety, and therefore (-log (0.25)) for the calculation of CI(c). These results concern the term "anxiety" in our ontology, following the same logic for the other terms. The lack of a psychological ontology modelling the concepts, those correspond to the fragility of people's mental health during the COVID-19 pandemic, so we thought about developing an ontology that would fill this gap.

412

Authors: Elly Matul Imah, Erina Seviyanti Dewi, I. G. P. Asto Buditjahjanto

Electroencephalography (EEG) is a method for recording the electrical activity of the brain. In the EEG various frequency signals can analyse the brain and the brain's behaviour. EEG can detect the abnormalities in the brain, one of the abnormality is Autism Spectrum Disorder (ASD). ASD is a condition where a person has a combination of neurological disorders in social communication and behavior, limited interest in something and sensory behavior. Joint attention (JA) is the ability in sharing attention between interactive social partner with third external elements as objects or events. Joint attention can also be said as social interaction behaviors to follow the attention of others, and direct attention of another. Joint attention is the key point affecting social communication in individuals with autism spectrum disorder (ASD). In this study, the detection of the ability of ASD sufferers to respond to instructions to see a targeted object based on the EEG signal recording was conducted. The dataset used is BCIAUT P-300 that is non-linear separable and imbalanced class with a ratio of 1: 8. In handling the imbalanced data, under-sampling was applied. The feature extraction method that compared are wavelet and principal component analysis. Based on the experiment result, Pz has the best channel to classify the Joint Attention of ASD using EEG P-300 data. The best accuracy is GRLVQ and the best G-mean is SVM. Over all the results, based on accuracy, g-mean, training and testing time show that GRLVQ is better performance than others, and SVM is the runner up. The differences of accuracy 9%, training time GRLVQ faster around 45 seconds than SVM, and testing time faster around 3 seconds.

435

Authors: Vandana Charaliparambu Pathayapuram, Ajeet Annarao Chikkamannur

Internet of Things (IoT) paradigm merges the physical world into digital world with proliferation of smart solutions in various domains. These smart solutions leverage the sensing and actuating power of heterogeneous IoT devices. The user applications must discover the right IoT resource at the right time. To serve this objective, the most important attributes of these diverse IoT devices must be identified and managed by network management systems (NMS) for effective discovery. In this paper, an ameliorated feature selection approach is proposed which selects minimum number of discriminating and service relevant features of multiple diverse IoT devices. Proposed feature selection technique has two phases where the features are ranked based on discriminating capability. Minimal subset of discriminating features is then chosen and ensemble with service relevance features. The proposed model is validated with dataset constructed from annotated IoT resources based on our previous work. Analyzes is carried out by comparing with existing state of art hybrid feature selection techniques and the proposed method has outperformed in terms of accuracy, quality and size of feature set selected. Further, these selected feature set derived from the proposed feature selection method is used to enrich IoT protocol, Constrained Application Protocol (COAP) CoRE Link format. This semantic enriched COAP protocol is used for IoT resource discovery on COOJA simulator. The results show that enriched COAP based discovery diminishes the discovery time by 33% compared to traditional COAP. Experimental work is carried out to showcase the effectiveness of the proposed feature selection model in terms of resource discovery success rate, query response time, service type detection accuracy.

458

Authors: Qasim Hadi Kareem, Malik Jasim Farhan

This paper presents a compact linear and circular polarization multi-input multi-output (MIMO) antenna for wireless local area network (WLAN) / worldwide interoperability for microwave access (WiMAX) and C-band wireless communication systems. This work has been performed with two configurations of six-port MIMO antennas with a total size of 60 x 60 〖mm〗^2 on low-cost FR-4 substrate. There are two rectangular slots - one is situated on the upper part of the circular patch to radiate left hand circular polarization (LHCP) waves, and the other is located on the lower portion of the circular patch to emit right hand circular polarization (RHCP) waves. It is a wave absorbing slot that allows electromagnetic waves on one side to be separated from the other. Another design of the four elements is the structure that consists of four orthogonal radiators. The four feed-through pair antenna elements are located orthogonal to each other for best isolation and provide vertical and horizontal polarization. The two configurations have a broad bandwidth (640MHz and 400MHz), high efficiency (over 75 %), good isolation (>-20 dB), low correlation coefficient (<0.01), high channel capacity (22.5 bps/Hz), and stable gain characteristics. The proposed MIMO antenna is simple, generalized, and can be adapted for a broad range of frequencies. It was found that there is a close agreement between the measured/calculated antenna parameters and experimental performance.

481

Authors: Ali Majeed Mahmood, Azad Raheem Kareem

Increasing the number of mobile network connections represented by humans and machines leads to a higher demand for resources, which have to be allocated for all User Equipment (UEs) with an acceptable level of fairness. However, the cost of allocating the required resources for a great number of connections may result in growing the computational complexity of scheduling algorithm and can delay the response of the Base Station (BS). In this paper, a new processing approach is proposed for Fifth Generation - Internet of Things (5G-IoT) networks. The computational complexity of the scheduling process is minimized in two folds. First, the Spark framework is used as a distributed and parallel processing paradigm, which can reduce the computational complexity from O(N^2) to O(〖N/M)〗^2. Then, to minimize the scheduling latency of the scheduler itself. The second level of complexity minimization is presented by using a low computational scheduling algorithm via modifying the traditional proportional fair scheduling algorithm to further decrease the complexity from O(〖N/M)〗^2 to O(N/M). Simulation results demonstrate the efficacy of the proposed scheduling approach by applying the concept of divide and conquer in achieving the scheduling process. Hence, the effectiveness of the new scheduling approach represented by scaling down the time of scheduling large number of UEs to the time of small sub-clusters of UEs. The desired percentage of gain in Scheduling Time (ST) depends on the size of sub-clusters such as deploying 250 UEs in form of clusters of (50, 30, or 10) UEs will reduce the ST compared to the classical Proportional Fair (PF) with a percentage of 85%, 89%, and 97% respectively. This property can be exploited to support network scalability at low scheduling latency in next-generation mobile networks.

503

Authors: Zakaria Boucetta, Abdelaziz El Fazziki, Mohamed El Adnani

Road deteriorations threaten people ‘safety as they are considered one of the major causes of road accidents leading to significant human and material losses. Automatic detection of road deteriorations can help massively in avoiding the tragic consequences. However, this task represents a real challenge. This work describes an attempt to tackle this issue by establishing a crack notification and road condition monitoring system. The proposed system will help drivers to avoid dangerous cracks and enable authorities to supervise and manage the road surface. In order to automatically detect and classify the road surface images, a convolutional neural network (CNN) is used, as this model has proven to be more suitable for this task than traditional methods due to its ability to extract features implicitly. The novelty in this paper consists on finding the optimal CNN model architecture for crack detection and classification into 10 distinguished classes. The proposed CNN model is trained and tested using historical 3D pavement images and can perform the classification with a high accuracy (above 95%). To evaluate and supervise each road segment condition, a novel method for severity indexes computation that uses the resulting classes is suggested as well as the generation of a weighted road graph in which the edge weights evolve according to the severity indexes. All data processing tasks are performed over a Hadoop-based framework using HBase and MapReduce.

528

Authors: Doan Perdana, Sarah Hafidzah, Bayu Erfianto

The IEEE 802.11ah standard was developed from IEEE 802.11, which works on a Wireless Local Area Networks (WLAN). This standard works in the sub-band 1GHz, to increase the range of an Access Point (AP) up to 1 km2, with the ability to handle approximately 8000 stations (STA). The IEEE 802.11ah standard occurs in the MAC and PHY layers. However, the high number of STAs produces high collision, therefore this standard introduces the Restricted Access Window (RAW) at the MAC layer. This research accurately examines a surrogate model to predict RAW performance under hidden nodes scenario. The solution to the surrogate model was analyzed using the Markov chain and numerically simulated with Matlab. Furthermore, the Markov chain solution is used to determine the performance measure parameters, namely throughput, energy consumption, and average delay. This research also examines the effect of hidden nodes on the IEEE 802.11ah standard network’s performance, especially in RAW, with variables obtained using Bianchi’s approach. The result showed that the simulated RAW slot duration of 52 µs on the IEEE 802.11ah standard made performance results better than 104 µs and 156 µs. The effect of the hidden nodes makes the successful transmission time probability longer due to its increment.

545

Authors: Sajjad Amiri Doumari, Hadi Givi, Mohammad Dehghani, Om Parkash Malik

There are many optimization problems in different scientific disciplines that should be solved and optimized using appropriate techniques. Population-based optimization algorithms are one of the most widely used techniques to solve optimization problems. This paper is focused on presenting a new population-based optimization approach called Ring Toss Game-Based Optimization (RTGBO) algorithm. The main idea of RTGBO is to simulate the behaviour of players and rules of the ring toss game in the design of the proposed algorithm. The main feature of the proposed RTGBO algorithm is the lack of control parameters. Steps of implementing RTGBO are described in detail and the proposed algorithm is mathematically modeled. The ability of RTGBO to solve optimization problems is evaluated on a set of twenty-three standard objective functions. These functions are selected from three different groups including unimodal, high-dimensional multimodal, and fixed-dimensional multimodal. The performance of RTGBO is also compared with eight other well-known optimization algorithms including Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Gravitational Search Algorithm (GSA), Teaching Learning-Based Optimization (TLBO), Gray Wolf Optimizer (GWO), Emperor Penguin Optimizer (EPO), Hide Objects Game Optimization (HOGO), and Shell Game Optimization (SGO). The results of optimization of objective functions of unimodal type indicate the high exploitation ability of RTGBO in solving optimization problems. On the other hand, the results of optimizing the multi-model type objective functions indicate the acceptable exploration ability of RTGBO. The results also confirm the superiority of the proposed RTGBO algorithm over mentioned optimization techniques.

565

Authors: Omar Abdul Razzaq Abdul Wahhab, Ahmed Sabah Al-Araji

This paper presents an improvement of the actual output trajectory tracking performance of a mobile robot based on convolutional neural network controller with off-line and on-line tuning Back-Propagation algorithms. The goals of this strategy are to find the optimal path to direct its movement and to design Convolutional Neural Network Trajectory Tracking (CNNTT) controller in order to control the nonlinear kinematics mobile robot system. Therefore, a hybrid swarm optimization algorithm uses for solving the two most important problems of path planning; the first is that the path must avoid collision with obstacles, and the second it must reduce the length of the path to a minimum. This paper will discuss the finding of the shortest path with the optimum cost function by using three optimizations’ algorithms; Chaotic Particle Swarm Optimization (CPSO) algorithm, A-star algorithm, and a hybrid swarm optimization algorithm (ACPSO). The task of the proposed feedback (CNNTT) controller is to obtain precisely and quickly the robust left and right wheels velocity which are used to control the position and orientation of the mobile robot system. These algorithms are simulated by MATLAB in a fixed obstacles environment to show the effectiveness of the hybrid swarm optimization algorithm in terms of the minimum number of an evaluation function and the shortest path length as well as the results of the proposed method showing that the (CNNTT) controller is accurate in terms of the mobile robot follows the desired paths quickly through fast obtaining the (CNNTT) controller’s parameters and a smooth linear wheels velocity actions are generated for mobile robot system with minimum number of cost-function evolutions that minimized the tracking error x-position around 4 cm and y-position around 2.5 cm and zero approximately orientation error as well as no oscillation in the responses. Finally, we confirm the effectiveness of the numerical simulation results of the proposed control strategy through comparison other types of controller simulation results.