International Journal of Intelligent Engineering and Systems

11

Authors: Anil Kumar Chatamoni, Rajendra Naik Bhukya, Praneet Raj Jeripotula

In this paper, a novel scheme for secure image transmission based on compressive sensing (CS) and Fractional Wavelet Transform (FrWT) is proposed. The scheme uses CS and a multi chaotic pseudo random Sine-Tent- Hénon (STH) map based measurement matrix, in the first stage of encryption. Therefore, it provides a simultaneous compression and encryption. Then the security is further enhanced through the novel approach in the second stage of encryption with an iterative procedure, in which a combination of FrWT and random pixel exchange method is applied. The key parameters used in the generation of random matrix, measurement matrix and fractional orders of FrWT are served as keys. This approach has the larger key parameters provided by the STH map and high degree of scrambling and diffusion with FrWT. With the simulation results, the novel proposed scheme has better compression performance as the PSNR value is 35.6631 dB with 0.75 compression ratio and above 25dB with only 4950 coefficients in the reconstruction of image. When the different types of attacks applied, the value of PSNR in the range of 20-30dB shows the good reconstruction robustness of the proposed scheme. Numerical value comparison with other recent CS based encryption schemes, represents the superiority of the proposed scheme in terms of security.

35

Authors: Badra Khellat Kihel, Souad Khellat Kihel, Samira Chouraqui

The sizes of medical datasets are increased with an exponential rate. Extract useful knowledge from those data, analyzing and searching on this very vast amount of data is become a hard task for the scientists. Several metaheuristic algorithms for solving such problem were proposed for dimensionality reduction. In this work, a new stochastic search strategy inspired by the Grey Wolf optimization theory is proposed for feature subset selection. Grey Wolf Optimization algorithm (GWO) is a new metaheuristic optimization technique. Its principle is to reproduce the behavior of grey wolves in nature to hunt in a cooperative way. In order to improve the disadvantage of Grey Wolf Optimizer when solving the feature selection problem, three immune operators (selection, cloning and mutation) are embedded into the standard GWO and an Hybrid Immune Grey Wolf Algorithm (IGWO) is proposed. To ensure the diversity of the population of wolves and avoid premature convergence, we cloned and mutated Alpha, Beta and Delta wolves before selection of the new first, second, and third best solutions. The experiments were performed using ten medical datasets: Parkinson, Leukemia and heart diseases, Breast, Colon, and Prostate Cancers, and the results show that the proposed approach is able to explore different regions of a search space and avoid local optima reducing the dimensionality of datasets than can significantly improve the performances of the system for diseases recognition and decrease the required classification time.

44

Authors: Malikhah Malikhah, Riyanarto Sarno, Shoffi Izza Sabilla

Pork and beef are the main resources of red meat in the world. However, not everyone can eat pork because of their religious background or other reasons. Therefore, it is very important to ensure the purity of the meat prior to being consumed. This research applied ensemble learning to optimize the classification on Electronic Nose Dataset for Pork Adulteration in Beef. Ensemble learning is one of a method that is widely used and the most successful method to improve performance. This research used several traditional machine learning algorithms and chose a machine learning algorithm, which produced the best result as the base classifier for ensemble learning to optimize the classification on Electronic Nose Dataset for Pork Adulteration in Beef. There were three ensemble learnings used in this research, namely hard voting, stacking, and bagging. The steps conducted in this research comprise (1) pre-processing by de-noising of the raw signals, (2) statistical feature extraction, (3) feature selection, (4) classification, (5) using ensemble learning to improve the performance, and (6) performance evaluation. The experiment result shows that hard voting ensemble learning using K-nearest neighbors (KNN) as base classifier is able to distinguish well between beef, pork, and pork adulteration in beef to seven classes which obtained 98.33% accuracy.

66

Authors: Ediga Sathyanarayana Phalguna Krishna, Thangavelu Arunkumar

Internet of Things (IoT) is a network that provides security for physical objects such as smart home appliance, smart machines and many more. The physical objects are assigned to a unique Internet address known as Internet Protocol (IP) that is used for data communication with the external entities of the network through the internet. The IoT devices are facing security issues due to the rapid increase in attacks that are launched by the intruders during data sharing through the internet. The detection of attacks is essential to provide a strong security mechanism for such threatening attacks. The proposed hybrid optimization algorithm utilizes the combination of Particle Swarm Optimization (PSO) and Gray Wolf Optimization (GWO) in this research. The PSO is known for its fast computation speed and has found extensive utility in data training as well as data estimation. The GWO is developed as an intrusion detection approach to classify data and to efficiently detect several of intrusions. The proposed hybrid GWO-PSO uses NSL-KDD data set with binary and multi class problem respectively for showing the effectiveness of the present work. The results obtained better accuracy value of 99.97 % when compared to the existing LSTM-RNN that achieved 97.72% of accuracy, whereas the multi class SVM obtained 98 % and modified rank-based information gain feature selection method showed 99.8 % of accuracy.

91

Authors: Moufida Achour, Abderrahim Belmadani

Research on the privacy of human genomic data has significantly increased these last years to find out effective solutions to protect the most critical information resulting from an alignment operation such as the genomic position, cigar and Read content. In this paper, the authors propose an encoding technique based on homomorphic encryption to protect genomic positions of DNA sequences for a secure storage of BAM file. This technique uses two Discrete Gaussian Sampling algorithms for comparison purposes which is Gauss and Knuth-Yao algorithm. The framework was implemented on Java from scratch using container data structure with standard java HashMap for the representation of duplicates positions elements exiting in BAM file. The proposed method was assessed on the NA12878 dataset with initially 2 million positions to encrypt. Obtained results show that the FV-Knuth-Yao technique with HashMap architecture storing 165667 positions and a polynomial degree equal to 64 ensures a fast encryption time estimated to 6,5 minutes. The decryption operation needs 1 hour with zero loss of information. The acceptable security level reached 264 operations to break the cryptosystem. To demonstrate the efficiency, security and reliability of our approach, the FV-Knuth-Yao was compared with the OPE Method as implemented in SECRAM tool which is based on a deterministic encryption scheme. Our work mainly focuses on the security of genomic data without compression of BAM file.

115

Authors: Tanairat Mata

Time-domain OFDM (Orthogonal Frequency Division Multiplexing) signal has various different levels from many subcarriers in one frame which are fatal effect in the non-linear OFDM channel. Because of the signal distortion occurring at the output of non-linear power amplifier, the performance of system has failure. This issue is called that Peak to Average Power Ratio or PAPR which is the main issue of concern for OFDM system. One of the effective PAPR reduction technique known as partial transmit sequence (PTS) with non-uniform phase factors investigating between 0 and 2π was proposed to solve this issue which can suppress the PAPR value competently by clustering their data subcarriers in frequency-domain and multiplying data subcarriers in each cluster by non-uniform phase factors in time-domain after IFFT (inverse fast Fourier transform). However the computational complexity is increased in proportion to the increasing of cluster numbers. By carefully reducing the PAPR of time-domain OFDM signal with small computational complexity, this paper proposes a modified ABC (Artificial Bee Colony) with PTS (ABC-PTS) scheme which can reduce the PAPR of time-domain OFDM signal by applying non-uniform phase factor in the weighting factor sequences with small computational complexity of PAPR reduction process by performing the modified ABC algorithm. The simulation results in this paper can demonstrate that the proposed scheme outperforms approximate 0.4dB and 0.8dB over the original PTS for adjacent and interleaved respectively and approximate 3.2dB over the conventional OFDM at CCDF=10-2 (Complementary Cumulative Distribution Function) with smaller computational complexity which is 19.14% approximately as compared with the original PTS. Moreover, the radix-2, radix-4 and split-radix DIF-IFFT (Decimation-in-Frequency IFFT) algorithms are employed to reduce the computational complexity which are approximate 50%, 62.5% and 66.67% for PAPR reduction process without affecting the optimum PAPR performance.

136

Authors: Shaymah Akram Yasear, Ku Ruhana Ku-Mahamud

The ant colony system (ACS) algorithm is one of the metaheuristics in solving combinatorial optimization problems. The control parameters namely, pheromone coefficient, heuristic coefficient, decision rule, and evaporation rate of the pheromone in ACS are important in determining the performance of the algorithm. However, the initialized values of these parameters stay constant during the search process which leads to performance degradation of the algorithm. This paper aims to tune the parameters of the ACS algorithm by using Harris’s hawk optimization (HHO) algorithm in solving the travelling salesman problem (TSP). The proposed hybrid algorithm is called Harris’s hawk optimizer ant colony system (HHO-ACS). The final process of HHO-ACS will output the path distance. The performance of the HHO-ACS has been evaluated by using different symmetric TSP instances of various scale size. The results showed the HHO-ACS provided superior performance compared to other well-known metaheuristics namely black hole, particle swarm optimization, dragonfly, genetic and ant colony optimization algorithms. Thus, it can attain a better solution with higher accuracy. The proposed algorithm was able to achieve best known optimal solution in solving bayg29, att48 and berlin52 instances and near optimal solution in solving bays29, eil51, st70, eil76 and eil101 instances. Compared to other algorithms, the HHO-ACS algorithm showed superior performance in solving the TSP instances which indicates its effectiveness. This is possible because, in the HHO-ACS algorithm, the main control parameters of ACS algorithm namely, the pheromone coefficient, heuristic coefficient, evaporation rate of the pheromone, and decision rule were tuned according to the problem instance at hand. Thus, the HHO-ACS algorithm can be used to solve problems of travelling salesman nature with minimum customization.

157

Authors: Roshan Shetty, Prasad Narasimha Sarappadi

Thoracic diseases are the various collections of diseases that are associate with the cavity of the thorax such as lungs, heart, oesophagus, chest wall, and diaphragm. The Chest X-Ray is the widely used radiological examination in the diagnosis of such diseases. Many existing methods used entire Chest X-Ray images for the process of training however, it suffered from few limitations. The misalignment of potential or exist of unrelated objects in the Chest X-Ray images resulted in irrelevant noises which limited the network performance. The pre-processing steps of existing methods during training the neural network resulted in lower resolution images. Due to loss of information, it is very difficult to find the tiny lesion regions from the images. To solve such issues, the Self-Sequential Attention Layer based DenseNet (SAL-DN) model is proposed to enhance thoracic disease prediction on Chest X-Ray. The SAL uses the co-relation among the class labels and abnormal pathology by analysing the known feature maps from DenseNet-121.The SAL-DN has the advantages of providing the best representation of images and the capacity to handle large imbalanced datasets like Chest X-Ray14. The performance of the proposed SAL-DN model is compared with existing methods in-terms of the achieved AUC score. The experimental result shows that the proposed SAL-DN outperforms the Thorax-Net method by obtaining an Average AUC score of 0.8715, whereas Thorax-Net obtained an Average AUC score of 0.7876 in patient-wise official split.

177

Authors: Issa Mohammed Saeed Ali, Mukunthan Balakrishnan

The analytics techniques in Big Data are extensively employed as an alternative to generalized for data mining due to the huge volumes of large-scale high dimensional data. Feature selection techniques eradicate the redundant and inappropriate features to decrease the data dimensionality and increase the classifiers’ efficiency. However, the traditional feature selection strategies are dearth of scalability to handle the unique characteristics of large-scale data and extract the valuable features within the restricted time. This article proposed a feature selection algorithm centered on the Population and Global Search Improved Squirrel Search Algorithm (PGS-ISSA) that tackles the problem of local optimum and reduce convergence rate in standard Squirrel Search Algorithm (SSA). The novelty of this proposed PGS-ISSA is the introduction of chaos theory to improve population initialization so that the search space is increased. Then the acceleration coefficients are used in the position update equations to improve the convergence rate in local search process while inertia weight is also applied to optimally balance the exploration and exploitation in SSA. PGS-ISSA employs the fitness function based on the minimum error rate for ensuring the selection of best features that improve the classification accuracy. The proposed PGS-ISSA based feature section algorithm is evaluated by using Support Vector Machine (SVM) classifier implemented in MATLAB tool to address the big data classification problem. The experiments performed on both small and large-scale datasets illustrated that the suggested PGS-ISSA enhances the classification accuracy by 1.7% to 5.4% better than other compared models through effective handling of the big data problems. The results obtained for the bigger Higgs dataset shows that the proposed PGS-ISSA achieved high performance than the standard SSA, existing ISSA models and other prominent optimization-based feature selection algorithms with 64.72% accuracy, 67.3194% precision, 62.1528% recall, 62.3026% f-measure, 82.7226% specificity and consumed less time of 140.1366 seconds. PGS-ISSA also achieved comparatively better results for the other benchmark datasets with 0.3% to 6% improvement on statistical metrics and 10% to 25% reduction in execution time.

204

Authors: Eman H. Zaky, Mona M. Soliman, A. K. Elkholy, Neveen I. Ghali

In the field of health care, one of the most important problems is predicting the possibility of hospital readmission due to its important role in caring for patients with chronic diseases such as diabetes. Such predictions affect the health care costs and the hospital’s efficiency and reputation. In this paper, an intelligent-based model is developed to predict the reintroduction of the patient into the hospital. This model is based on using some Machine Learning (ML) algorithms such as Logistic Regression (LR), K-Nearest Neighbors (KNN), and Support Vector Machine (SVM). Also, it proposes the use of a Deep Learning (DL) based network such as a convolutional neural network (CNN). Both ML and DL are used as classifiers to predict hospital readmission. The main problem is the input noisy data to these classifiers. These noisy data reduce the accuracy of the readmission prediction model. Sequential pre-processing steps are proposed to get over such a problem. These pre-processing steps provide solutions to missing values, feature engineering, and normalization problems. The main contribution of this work is improving readmission prediction rate by solving the data normalization problem. Two types of data normalization (e.g. z-score and min-max normalization) are applied, results show there is a difference in accuracy, z-score normalization is better than min-max normalization when comparing ML methods and DL models, CNN is the best with an accuracy of 0.894% in case of z-score normalization. Moreover, the model performance is improved with an accuracy of 0.924% when non-normalized data is used as input to the model. The proposed Non-normalization technique successes in providing superior results compared to some previous techniques which are displayed data by using Ensemble, Normalization, and Ensemble by age group techniques.

231

Authors: M. J. Niegil Francis, Mihika Preethi Keran, Rachana Chetan, B. Niranjana Krupa

This study aimed to find an optimal feature-extraction method for real-time electroencephalography-based naviga-tion. A comparative study of promising methods was done on two-class motor-imagery data. A novel combination of Hjorth parameters and Welch’s-power-spectral-density produced the highest two-class accuracy of 0.84 across five subjects. The method is computationally inexpensive and provides short training and execution times that best suit real-time applications. The method yielded an average accuracy of 0.66 on three-class data. The real-time fea-sibility of the method was tested by implementing an object-retrieving robot and running a navigation experiment. The robot successfully navigated a three-circuit maze using three commands – right, left, and forward – and auto-matically retrieved the desired object. Along with being ideal for real-time applications, this method utilized only eight electrodes, making it compatible with cheap, portable BCI devices. The method and the applica-tion-framework can find commercial utility in the field of healthcare for the elderly and physically-handicapped.

251

Authors: Awan Uji Krismanto, Irrine Budi Sulistiawati, F Yudi Limpraptono, Ardyono Priyadi, Herlambang Setiadi, Muhammad Abdillah

The increasing penetration of wind power plant introduces a various effect on power system stability, operation and control. Unpredictable, uncertain and fluctuating circumstances can significantly affect the performance of power system. Moreover, employment of novel technologies in wind power plant significantly alter the control characteristic and operation procedure of power system to deal with load variations and fast changing of generated power from wind-based power plant. One of the main focus in integrating large scale wind power plant is how to maintain voltage stability under transient and steady state scenarios. In this paper, effects of large-scale wind power plants on voltage stability of power system is investigated. Practical test system of South-West Sulawesi, Indonesia with integration of two large scale wind power plants are considered. Hence the novelty of this paper is to introduce new and practical test system for power system stability study considering renewable energy integration. The simulation results suggest that the additional power injection from wind power plant introduces beneficial effects on power system voltage stability performance. It was monitored that the increasing power injection from wind power plant enhanced the voltage profile (0.96 pu to 0.975 pu), voltage margin and load-ability (200 MW to 215 MW) of the system. Furthermore, it was also observed that additional power injection from wind power plant significantly improved the dynamic voltage stability performance of the power system, ensuring stable operation under transition stages when the power system was subjected to disturbances.

276

Authors: Joe Siang Keek, Ser Lee Loh, Yan Chiew Wong, Xiu Juan Woo, Wei Wen Lee

Interference minimization in cellular network has and will always be top priority, whether in current or future generation of cellular technology. Therefore, cellular channel assignment problem (CAP) requires continuous study and research. This paper presents the study and comparison of Genetic Algorithm (GA) and Particle Swam Optimization (PSO) for CAP in minimizing interference. GA with three variants in term of population selection – roulette wheel selection (RWS), tournament selection (TS) and stochastic universal sampling (SUS) were studied, and then compared with classic PSO. Two CAPs were derived and used to comprehensively evaluate the performances of the PSO and GAs. It was found that GA-TS is ~11% and ~7% faster than GA-RWS and GA-SUS, respectively. Although the difference is small, but it allowed GA-TS to run for few more iterations and eventually achieved better interference minimization. Moreover, it was also found that GA-SUS has less noise and produce a more consistent result. On the other hand, PSO is slower than GA-TS, but has higher potential to converge on smaller minimum value.

299

Authors: Fatima Chanaa, Radouane N'ait Zarri, Rachid Bendaoud, Houssam Amiry, Elhadi Baghaz, Said Bounouar, Bouchaib Zohal, Charaf Hajjaj, Said Yadir, Abderrahim El-abidi, Mohammadi Benhmida

The islanding of grid-connected photovoltaic (PV) installations is a difficult problem that leads to many protection and security problems. A variety of islanding detection methods (IDMs) are proposed in several studies and these techniques are categorized into passive, active and communication based methods. This study describes the IDM adopted to detect islanding situation in the new PV installation connected to LVs grid of OCP Group (Cherifian Office for Phosphates in Morocco), global leader in the phosphate and phosphate derivatives markets. After having defined the way in which the PV generators will be integrated within the OCP Group grid, the causes that could lead to islanding situation are established. The comparison on the most used methods for islanding detection in the grid connected photovoltaic system (PV) is based on three parameters: the Non-Detection Zone (NDZ), the Detection Duration Time (DDT) and Total Harmonic Distortion (THD) of each method. These performance criteria are compared with the limit thresholds authorized by the IEEE 1574 standard, which are 5% for THD and 2 seconds for DDT. The Slip Mode Frequency Shift (SMS) method is demonstrated to be the most effective, for our installation, among the simulated methods, with THD and DDT of 1.3% and 0.28s, respectively.

326

Authors: Samir M. Hameed, Sinan M. Abdulsatar, Atheer A. Sabri

Visible Light Communication (VLC) becomes a promising technology in the Optical Wireless Communication OWC field. VLC has been developed based on Optical Orthogonal Frequency Division Multiplexing (O-OFDM) in recent years. The Bit Error Rate (BER) performance for various O-OFDM is studied in this paper. The non-linearity characteristic of Light-Emitting Diode (LED), modulation constellation size, O-OFDM scheme, and channel types impacts the BER performance of VLC systems significantly. The proposed model utilizes the conVolutional encoder of 2/3 rate and seven memory cells in the transmitter with O-OFDM for BER enhancement. At the receiver, the Viterbi algorithm system is employed with hard or soft-decision decoding. The proposed system is evaluated in Line-of-Sight and Non-Line-of-Sight (NLOS) channels. In the NLOS model, Zero Forcing Equalizer (ZFE) with channel estimation is added at the receiver path for equalizing the multipath diffused components. This study proves the capability of the proposed system in mitigating the LED non-linearity and reducing BER. The simulation results show the Signal-to-Noise Ratio (SNR) enhancement of the exanimated O-OFDM-based soft-decision between 7 to 10 dB for the BER 10-5 in the LOS model, and for the NLOS model, the BER can be improved from 0.02 up to 10-5 at SNR 45dB.

350

Authors: Dedy Trisanto, Nofita Rismawati, Muhamad Femy Mulya, Felix Indra Kurniadi

The development of credit card use in Indonesia has not been matched by the security provided by credit card service providers. This resulted in significant losses both in terms of banking and customers. The difficulty in finding the characteristics of credit card fraud is one of the biggest challenges. Currently, many are developing machine learning models that can identify credit card fraud to help banks. Unfortunately, the model created is mostly biased towards the class which has more dominant data. This problem is caused by the imbalance of the data on the available dataset. In the previous research, Wang et al found the implementation of Focal loss in XGBoost improve the precision, recall of imbalanced data. However according to Qin et al, the parameter loss from Focal loss have poor judgement in several case of imbalanced data and to handle this weakness, they proposed Weighted – Cross Entropy Loss (W-CEL) loss in Focal loss. According to previous research, we propose a Modified Focal Loss method for Imbalanced XGBoost by entering another parameter from W-CEL loss to Focal Loss to improve the ability of Focal Loss. Focal loss itself is a method that is often used to give weight to classes that are often misinterpreted, so that with the use of imbalance parameters (φ) from W-CEL loss is expected to improve the weighted value of the Focal Loss. We tested our proposed method in credit card fraud dataset from Université Libre de Bruxelles (ULB) machine learning group. Our proposed method produced 100 % accuracy, 0.97 precision, 0.56 recall and 0.72 MCC score in scenario 1 with extreme imbalanced data, in scenario 2 with the mild imbalanced data, the result delivered 99% accuracy, 0.88 precision, 0.87 recall and 0.89 MCC and in scenario 3 with the medium imbalance data, the result delivered 100% accuracy, 0.97 precision score, 0.72 recall score and 0.83 MCC score. The results obtained in this study proved that the proposed method makes machine learning models valid and unbiased, especially in mild-imbalanced data. However, many improvements still need to be made to medium and extreme imbalanced data.

384

Authors: Sajjad Amiri Doumari, Fatemeh Ahmadi Zeidabadi, Mohammad Dehghani, Om Parkash Malik

Numerous designed optimization problems in different disciplines of science should be solved using appropriate techniques. population based optimization algorithms are one of the powerful tools in solving optimization problems. The innovation of this paper is to present a new optimization algorithm called Mixed Best Members Based Optimizer (MBMBO) that can be used to solve various optimization problems. The main idea in designing the proposed MBMBO algorithm is to create a mixed member of several top members of the population in order to guide and update the algorithm population. The main feature and advantage of the MBMBO is the lack of control parameters. Therefore, the proposed MBMBO does not need to adjust the parameter. The various steps of the MBMBO are described and then mathematically modeled for implementation in solving optimization problems. The performance of the MBMBO in solving optimization problems is evaluated on a set of twenty-three standard objective functions. These objective functions are of three different types, including seven unimodal objective functions, six high dimensional multi-model objective functions, and ten fixed dimensional multi-model objective functions. The results of evaluation of single-model objective functions indicate the high exploitation power and also the results of evaluation of multi-model objective functions indicate the high exploration power of the proposed MBMBO algorithm. Also, the results obtained from the simulation of the MBMBO are compared with the results of 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 optimizing the objective functions of unimodal and multi-modal types using MBMBO show the acceptable ability of the proposed algorithm to provide suitable solutions. Comparison of the simulation results shows that the proposed MBMBO is much more competitive than the other eight optimization algorithms.

410

Authors: Ahmed Hesham Mostafa, Hala Abdel-Galil El-Sayed, Mohamed Belal

Facial Expression Recognition (FER) is one of the most important research problems in computer vision and Artificial Intelligence (AI) due to its potential applications, many studies were proposed for the FER, whether based on using handcrafted (Craft) features with traditional machine learning techniques or using end to end convolution neural network (CNN). In this paper, we proposed a new model called CNNCraft-net based on combining the advantages of CNN and traditional models by concatenating features outputs from CNN, autoencoder, and handcrafted features such as scale-invariant feature transform (SIFT), speed up robust feature (SURF) and Oriented Fast Rotated Brief (ORB), computed by the bag of visual words (BOVW) to recognize eight facial expressions for static RGB images. For the comparative analysis, multiple metrics were used such as Accuracy, Loss, F-measure, precision, and recall. The high imbalanced AffectNet and FER2013 datasets were used to evaluate the proposed model where the proposed model achieves accuracy 61.9% for eight expressions and 65% for seven expressions for AffectNet and 69% for FER2013.

437

Authors: Hapsari Peni Agustin Tjahyaningtijas, Dewinda Julianensi Rumala, Cucun Very Angkoso, Nurul Zainal Fanani, Joan Santoso, Anggraini Dwi Sensusiati, Peter M.A van Ooijen, I Ketut Eddy Purnama, Mauridhi Hery Purnomo

Brain tumors are among the most common diseases of the central nervous system and are harmful. Early diagnosis is essential for patient proper treatment. Radiologists need an automated system to identify brain tumor images successfully. The identification process is often a tedious and error-prone task. Furthermore, brain tumor binary classification is often characterized by malignant and benign because it involves multi-sequence MRI (T1, T2, T1CE, and FLAIR), making radiologist's work quite challenging. Recently, several classification methods based on deep learning are being used to classify brain tumors. Each model's performance is highly dependent on the CNN architecture used. Due to the complexity of the existing CNN architecture, hyperparameter tuning becomes a problem in its application. We propose a CNN method called en-CNN to overcome this problem. This method is based on VGG-16 that consists of seven convolutional networks, four ReLU, and four max-pooling. The proposed method is used to facilitate the hyperparameter tuning. We also proposed a new approach in which the classification of brain tumors is done directly without priorly doing the segmentation process. The new approach consists of the following stages: preprocessing, image augmentation, and applying the en-CNN method. Our new approach is also doing the classification using four MRI sequences of T1, T1CE, T2, and FLAIR. The proposed method delivers accuracy on the MRI multi-sequence BraTS 2018 dataset with an accuracy of 95.5% for T1, 95.5% for T1CE, 94% for T2, and 97% for FLAIR with mini-batch size 128 and epoch 200 using ADAM optimizer. The accuracy was 4% higher than previous research in the same dataset.

463

Authors: Mohammed K. Hussein, Nasser N. Khamiss

The 5G approximately provides 1000x higher mobile data traffic and 10 to 100x numbers of connected devices than the 4G. The 5G is becoming a reality after deploying it by some of the main cities in the world. Millimeter-wave (mm-wave) communications play a crucial role in the 5G networks owing to its enormous available bandwidth. The hybrid beamforming (HBF) presents better multiplexing gains than analog beamforming. It is a trade-off between cost and performance. As a result, the HBF is a promising approach that exploits the small digital beamforming and the high analog beamforming to meet these challenges. This study proposes and simulates a limited feedback hybrid beamforming for outdoor environments, including the analog precoding/combining based on a quantization codebook and the digital precoding based on Lagrangian optimization with the mathematical model in detail compared with other algorithms in the literature. It achieves an increase in the spectral efficiency satisfied, such as at 20dB SNR, it is approximately 11.75 bps/Hz while the other solutions are 11.18, 11.4, and 11.8 bps/Hz. Also, it presents work well in all possible and satisfies the quality of the results for all cases, such as increasing the number of users, number of multi-path, the SNR, and the BS antennas. Also, it offers the BER approximately 10-4 at 10.3 SNR, better than other solutions.

480

Authors: Ali A. Hasan, Hussain F. Jaafar, Mahmoud Shaker

In modern digital control system such as model reference control and model predictive control, where the real time calculation is the main challenge, the model order reduction has become very important issue to minimize the execution time. In this work, our aim is to construct a novel technique for reduction of high order discrete time systems. This could be achieved by computation algorithm model from a given high order pulse transfer function. The proposed model is based on matching the weighting sequence of the original parameters with those adopted in the low-order model. The generalized least squares method is then used to determine the reduced model parameters. The efficiency of the proposed algorithm is validated by using the integral squared error minimization between the original system and the reduced model. An example is presented and discussed to validate the efficiency of the proposed low-order model. Performance comparisons with many recent related works showed that the proposed model is promising in terms of low error indices and time responses.

514

Authors: Kaoutar Ouarid, Abdellatif El Assoudi, Mohamed Essabre, El Hassane El Yaagoubi

This paper aims to present a new approach in the design of observers for a Class of Discrete-time Takagi-Sugeno Singular Models (DTTSM) in the case of unmeasurable premise variables, allowing to estimate simultaneously the states of the system, and the faults in the actuators and sensors. The approach used is based on augmenting the state vector. The exponential convergence is studied using the Lyapunov theory. The stability conditions are presented as Linear Matrix Inequalities (LMIs). Lastly, a numerical application is used to evaluate the efficiency of the proposed dynamic system. It shows that the estimated variables catch up rapidly (around 0.3s), and accurately the real variables during the application time of the actuator and sensor faults.

539

Authors: A K M Zakir Hossain, Win Adiyansyah Indra, J. A. Jamil Alsayaydeh, Safarudin Gazali Herawan

In this article, a microstrip-feed planar monopole ultra-wide band (UWB) antenna is proposed for the chipless RFID (CRFID) tag-reader systems with the intension to reduce the size of the tag. This antenna is keyhole shaped and small in size of 32 × 20 mm2 compared with other existing antennas in the retransmission-based chipless CRFID system. The partial ground plane technique is applied here to obtain the bandwidth of 6.81 GHz within the UWB frequency spectrum. The results show it is suitable for both identification and sensing application in the retransmission-based CRFID. The antenna has a good omni-directional radiation pattern with realized gain up to 4 dBi and the realized gain to antenna size ratio goes as high 0.39/cm2 which is comparatively higher than the other existing antennas in retransmission-based CRFID. The other feature of this antenna is it has the radiation efficiency up to 99% whereby it never goes below 96% throughout the working bandwidth and the measured antenna efficiency goes as high as 95%. Furthermore, to check the suitability to apply in CRFID sensor application, the proposed antenna has been realized on three flexible substrates of polyimide, PET and paper, and very little deviations are found compared to the rigid Rogers RT 3003 substrate in terms of S11, realized gain, efficiency and radiation pattern which is a proof of robustness of the design. This proposed antenna will motivate the researcher to develop a smaller sized UWB CRFID tags.

561

Authors: Eman A. Radhi, Mohammed Y. Kamil

The task of segmenting breast tumours in mammograms is very difficult, as its difficulty lies in the lack of contrast between the tumour and the surrounding breast tissue, especially when dealing with small tumours that are not clear boundaries and hidden under the tissues. Segmentation algorithms often lose the path of tumor boundaries in an attempt to determine the position of them. Active contours are used widely for segmentation as a high-level technology for boundary recognition. The main aim to create a clear contrast between the tumour and the normal breast region. In this study, two approaches to active contour are applied: snakes and level sets. The proposed methods were applied to all abnormal mammogram images taken from mini-MIAS database. The first approach showed a weakness in the segmentation of this type of image, while the other approach was able to segment all the mammogram's tumours. The Chan-Vese method was the most superior of all the active contour segmentation methods. The proposed models were tested in two ways, the first is statistical the best result was for the Chan-Vese method and it came as follows (90%,95% 98%, 97%, 97%) for Jaccard, Dice, PF-Score, Precision, and Sensitivity respectively. And the other is based on the segmented region's characteristics, Chan-Vese was able to accurately determine the location and shape of the tumor. The proposed Chan-Vese approach is appropriate in adopting computer assisted detection systems to predict tumor boundaries and locations in mammography for its reliability and superior performance over other algorithms.