Search

The Internet of Things (IoT) is a network of various interconnected objects capable of collecting and exchanging data without human interaction. These objects have limited processing power, storage space, memory, bandwidth and energy. Therefore, due to these limitations, data transmission and routing are challenging issues where data collection and analysis methods are essential. The Routing Protocol for Low-power and Lossy Networks (RPL) is one of the best alternatives to ensure routing in LoWPAN6 networks. However, RPL lacks scalability and basically designed for non-dynamic devices. Another drawback of the RPL protocol is the lack of load balancing support, leading to unfair distribution of traffic in the network that may decrease network efficiency. This study proposes a novel RPL-based routing protocol, QFS-RPL, using Q-learning algorithm policy and ideation from the Fisheye State Routing (FSR) protocol. The proposed QFS-RPL is as lightweight and agile as the standardized RPL and partially outperforms the mRPL protocol on mobile networks. This method supports multi-path routing, and at any given time in the network lifetime, all possible paths for sending data from any node to the sink are available. Therefore, QFS-RPL provides high resilience against errors, failures, and sudden network changes. To evaluate the performance of the proposed method, the Contiki operating system and Cooja simulator have been used in scenarios with mobile and stationary nodes and random network topologies. The results have been compared with RPL and mRPL. We have developed an algorithm for ease of data transfer in the IoT, which provides better performance than existing protocols, especially when dealing with a mobile network. The performance evaluation criteria considered for simulation are load balancing, energy consumption, number of table entries, Packet Delivery Ratio (PDR), End-to-End (E2E) latency, network throughput, convergence speed, and control packet overhead.

The production of blood cells is affected by leukemia, a type of bone marrow cancer or blood cancer. Deoxyribonucleic acid (DNA) is related to immature cells, particularly white cells, and is damaged in various ways in this disease. When a radiologist is involved in diagnosing acute leukemia cells, the diagnosis is time consuming and needs to provide better accuracy. For this purpose, many types of research have been conducted for the automatic diagnosis of acute leukemia. However, these studies have low detection speed and accuracy. Machine learning and artificial intelligence techniques are now playing an essential role in medical sciences, particularly in detecting and classifying leukemic cells. These methods assist doctors in detecting diseases earlier, reducing their workload and the possibility of errors. This research aims to design a deep learning model with a customized architecture for detecting acute leukemia using images of lymphocytes and monocytes. This study presents a novel dataset containing images of Acute Lymphoblastic Leukemia (ALL) and Acute Myeloid Leukemia (AML). The new dataset has been created with the assistance of various experts to help the scientific community in its efforts to incorporate machine learning techniques into medical research. Increasing the scale of the dataset is achieved with a Generative Adversarial Network (GAN). The proposed CNN model based on the Tversky loss function includes six convolution layers, four dense layers, and a Softmax activation function for the classification of acute leukemia images. The proposed model achieved a 99% accuracy rate in diagnosing acute leukemia types, including ALL and AML. Compared to previous research, the proposed network provides a promising performance in terms of speed and accuracy; and based on the results, the proposed model can be used to assist doctors and specialists in practical applications.

The Internet of Things (IoT) has rapidly grown recently, and mobile devices (MDs) have encountered widespread usage. All of these cause an increase in the demand for more powerful computing resources. Meanwhile, a new concept called mobile edge computing (MEC) was introduced as a promising technology to access powerful computing resources closer to the user side for a quick and effective response, especially for time-intensive applications. Task offloading has emerged as a solution to allocate resources among computing resources of smart devices or computational resources available in MEC. This study presents a new binary quantum approach based on an arithmetic optimization algorithm (BQAOA) for computational tasks offloading decisions on MDs with low complexity and guaranteed convergence. However, since task offloading is an NP-hard problem, there is a need to use methods that provide the optimal possible solution for various quality criteria, including response time and energy consumption. Indeed, this is where the advantages of arithmetic optimization algorithms (AOA) and quantum computing have been used to improve the performance of MDs. This paper introduces a 2-tier architecture from the user to the cloud computing server-side. Also, a Markov model is proposed to compute the average network bandwidth in the offloading problem. The proposed BQAOA is compared with the best state-of-the-art algorithms in heuristic and meta-heuristic fields in different scenarios. The simulation results showed 12.5%, 12%, and 26% improvement in energy consumption, makespan, and Energy SLA Violations (ESV) optimization parameters, respectively.

In modern societies, handwritten identification is a great practical need and challenge for forensic sciences. Moreover, few research studies have focused on automatic offline handwritten document analysis. Also, there are very few standard datasets on handwritten document identification. In addition, handwritten documents lose their nature over time due to the spread and drying of ink. From this standpoint, the present study presents an offline writer identification system (in presence of the uncertainties such as different experimental conditions and environmental noises for more realistic assumptions) that is a critical need in forensic studies. For this purpose, a comprehensive right-to-left dataset is developed and devised by gathering data from 62 participants at different time intervals under different experimental conditions. This dataset is designed based on American Society for Testing and Materials (ASTM) standards. A Deep Convolutional Neural Network (DCNN) model based on modified pre-trained networks is designed and developed to extract features from raw data hierarchically. The proposed DCNN model is tested and investigated not only on the designed dataset but also on different datasets. One notable advantage of the present study is that it has made use of heterogeneous data. Another remarkable aspect of the proposed study is that the proposed DCNN model is independent of any specific languages which can be applied on various languages. The results of the study indicate that the proposed DCNN model can learn features hierarchically from the handwriting raw data and achieve higher accuracy than other comparative methods.

Software defined network (SDN) has shown significant advantages in numerous real-life aspects with separating the control plane from the data plane that provides programmable management for networks. However, with the increase in the network size, a single controller of SDN imposes considerable limitations on various features. Therefore, in networks with immense scalability, multiple controllers are essential. Specifying the optimal number of controllers and their deployment place is known as the controller placement problem (CPP), which affects the network's performance. In the present paper, a novel controller placement algorithm has been introduced using the advantages of nature-inspired optimization algorithms and network portioning. Firstly, the Manta Ray Foraging Optimization (MRFO) and Salp Swarm Algorithm (SSA) have been discretized to solve CPP. Three new operators comprising a two-point swap, random insert, and half points crossover operators were introduced to discretized the algorithms. Afterward, the resulting discrete MRFO and SSA algorithms were hybridized in a promoting manner. Next, the proposed discrete algorithm has been evaluated on six well-known software-defined networks with a different number of controllers. In addition, the networks have been chosen from various sizes to evaluate the scalability of the proposed algorithm. The proposed algorithm has been compared with several other state-of-the-art algorithms regarding network propagation delay and convergence rate in experiments. The findings indicated the effectiveness of the contributions and the superiority of the proposed algorithm over the competitor algorithms.

There is a wide range of association rules mining algorithms concerning the importance of weighted data to extract frequent rules. In some algorithms, each item has only one weight in the data set, while, in others, each item in each segment of the data set has only one specific weight. In this work, a new efficient algorithm named weighted frequent itemset mining using weighted subtrees (WST-WFIM) has been proposed to discover the average weight of frequent rules. This algorithm uses specific trees and some new data structures on the frequent pattern growth (FP-Growth) algorithm to calculate the average weight of discovered rules. It works on the data set that each item in each transaction has a specific weight and keeps them in the dedicated tree. Moreover, we have proposed the shared transactions’ concept and WST-WFIM by using it and calculate the average weight for discovered frequent rules. The algorithm capability was evaluated by standard sparse and dense weighted data sets. The results show that, while the algorithm works on weighted transactions, in sparse data sets, relative runtime increases by a decrease in minimum support (MinSup) parameter in comparison to the FP-Growth, and memory usage is approximately the same. The variations of runtime and memory usage with the MinSup parameter are opposite for dense data sets.

Today, the relational database is not suitable for data management due to the large variety and volume of data which are mostly untrusted. Therefore, NoSQL has attracted the attention of companies. Despite it being a proper choice for managing a variety of large volume data, there is a big challenge and difficulty in performing online analytical processing (OLAP) on NoSQL since it is schema-less. This article aims to introduce a model to overcome null value in converting document-oriented NoSQL databases into relational databases using parallel similarity techniques. The proposed model includes four phases, shingling, chunck, minhashing, and locality-sensitive hashing MapReduce (LSHMR). Each phase performs a proper process on input NoSQL databases. The main idea of LSHMR is based on the nature of both locality-sensitive hashing (LSH) and MapReduce (MR). In this article, the LSH similarity search technique is used on the MR framework to extract OLAP cubes. LSH is used to decrease the number of comparisons. Furthermore, MR enables efficient distributed and parallel computing. The proposed model is an efficient and suitable approach for extracting OLAP cubes from an NoSQL database.

This paper employs machine learning algorithms to detect tax evasion and analyzes tax data. With the development of commercial businesses, traditional algorithms are not appropriate for solving the tax evasion detection problem. Hence, other algorithms with acceptable speed, precision, analysis, and data decisions must be used. In the case of assets and tax assessment, the integration of machine learning models with meta-heuristic algorithms increases accuracy due to optimal parameters. In this paper, intelligent machine learning algorithms are used to solve tax evasion detection. This research uses an improved particle swarm optimization (IPSO) algorithm to improve the multilayer perceptron neural network by finding the optimal weight and improving support vector machine (SVM) classifiers with optimal parameters. The IPSO-MLP and IPSO-SVM models using the IPSO algorithm are used as new models for tax evasion detection. Our proposed system applies the dataset collected from the general administration of tax affairs of West Azerbaijan province of Iran with 1500 samples for the tax evasion detection problem. The evaluations show that the IPSO-MLP model has a higher accuracy rate than the IPSO-SVM model and logistic regression. Moreover, the IPSO-MLP model has higher accuracy than SVM, Naive Bayes, k-nearest neighbor, C5.0 decision tree, and AdaBoost. The accuracy of IPSO-MLP and IPSO-SVM models is 93.68% and 92.24%, respectively.

Due to unstructured and large amounts of data, relational databases are no longer suitable for data management. As a result, new databases known as NOSQL have been introduced. The issue is that such a database is difficult to analyze. Online analytical processing (OLAP) is the foundational technology for data analysis in business intelligence. Because these technologies were designed primarily for relational database systems, performing OLAP in NOSQL is difficult. We present a model for extracting OLAP cubes from a document-oriented NOSQL database in this article. A scalable Naïve Bayes classifier method was used for this purpose. The proposed solution is divided into three stages of preparation, Naïve Bayes, and NBMR. Our proposed algorithm, NBMR, is based on the Naïve Bayes classifier (NBC) and the MapReduce (MR) programming model. Each NOSQL database document with nearly the same attribute will belong to the same class, and as a result, OLAP cubes can be used to perform data analysis. Because the proposed model allows for distributed and parallel Naïve Bayes Classifier computing, it is appropriate and suitable for large-scale data sets. Our proposed model is a proper and efficient approach when considering the speed and reduced the number of required comparisons.