Internet of Things and Connected Technologies

The emergence of Industry 4.0 has made a breakthrough by providing state of the art services when it comes to the manufacturing and material handling sectors. Continuous research and development are being made to ascertain efficiency. In spite of technological advancements, several bottlenecks still exist that require its mitigation to a great extent. Hence, in this paper, a strategy has been suggested which encompasses the fully automated Autonomous Guided Vehicles (AGV), Green energy, Automatic Storage and Retrieval Systems (ASRS) and the Internet of Things (IoT) to coordinate the status of the operation by acquiring data through a supervisory control in order to optimize vehicular paths with the help of a dynamic routing algorithm. Based on the operating area, the ZigBee protocol is seen to be best suited for this purpose. Thus, by formulating a prescribed working environment, manufacturers can conserve energy, reduce costs, eliminate machine downtime, and increase operational efficiency.

Malware are capable of evolving into different variants and conceal existing detection techniques, which relinquishes the ineffectiveness of traditional signature-based detectors. There are many advanced malware detection techniques based on machine learning and deep learning, but they cannot fulfill the real issues in industries. Malware variants are evolving at a rapid pace and labelling each of them is not practical and feasible. So, industries are considering a lot of the unlabeled samples as benign, while only a few are labelled. Consequently, the authentic malware samples are mislabelled. Bias created by mislabelling the samples severely restraints the accuracy. Also, the user is unsatisfied with malware detection system, since there is poor negotiation between the speed and accuracy.In this research article, we propose a hybrid positive-unlabeled learning technique for malware detection that can address some important challenges. Here, we use an ensemble model comprising of Logistic regression (cost-sensitive boosted), Random Forest and Support vector machine, to detect the malware variants. Along with that, we demonstrate that features in the form of a triplet vector are optimal while training a model. Experimental outcomes show that our proposed model attains 91% malware detection accuracy having a false alarm rate less than 0.005, while the earlier state-of-art approaches can only achieve 76.4% to 89% accuracy. The detection speed of our approach is 0.003 s.

Air pollution is one of the new civilized world’s major concerns, which has a serious impact on human health and the environment. The main areas that are affected by toxic pollutants are the industrial areas and areas around it. Air quality prediction focuses mainly on these industrial areas. Industrial level usage of this project requires expensive sensors and an enormous amount of power supply. The World Health Organization (WHO) states that major air pollutants include particulate pollution, Carbon Monoxide (CO), Sulfur dioxide (SO2) and Nitrogen Dioxide (NO2). In addition to these mentioned gases, PM or Particulate Matter and VOC or Volatile Organic Compounds components also pose as grave threats. Long and short-term exposure to air suspended or air-borne toxicants has a different toxicological impact on humans. Some of the diseases include asthma, bronchitis, some cardiovascular diseases and long-term chronic diseases such as cancer, lung damage and in extreme cases diseases like pulmonary fibrosis.In this proposed system, an IoT prototype of a large-scale system using high-end and expensive sensors that measures the different air pollutants in the atmosphere is designed. Gas sensors are used in this prototype to record the concentration of the various pollutants that are encountered in the air on a regular basis. The data collected is stored in a cloud environment and analysis is performed that predicts the possible diseases, illness or health issues related to a particular pollutant, along with other factors like time of exposure to a particular pollutant etc. The framework uses stored data to train the model using multi-label classification with Random Forest algorithm, XG Boost algorithm in the local system. The real time data obtained using the different sensors is tested and the results obtained would be used to predict the possibilities of diseases such as asthma, lung cancer, ventricular hypertrophy etc. and the Air Quality Index (AQI) is calculated. In addition to this, preventive suggestions are also provided which is merely a cautionary message displayed on our LCD display to vacuum clean the room or mop the room thoroughly.

Effective market information systems help to reduce information asymmetries, increase competitiveness, and improve efficiency in the marketing network. Thus, lack of dissemination of market information and the bargaining capability of the traders across the agricultural supply chain is a major concern among small holder farmers in Nigeria. The advent of mobile phones serves as a great tool for awareness and information dissemination to people. A Mobile Based Market Information System is presented in this study. It serves as a means of enhancing farmers marketing strategy by providing market visibility for transacting business both within and outside the region. The proposed system employs Google map API which shows the location of the registered markets. The system was implemented using JavaScript, PHP/MySQL and Phonegap/Cordova. XAMMP database server was used for data management. The application was tested and validated by 50 respondents, the results returned high acceptance rate, high perception on usage and ease of use.

This paper aims to reduce the sidelobe levels of a Multicarrier Complementary Phase Coded (MCPC) radar signal. MCPC radar signal takes advantage of the properties of orthogonal frequency division multiplexing (OFDM). A MCPC signal is obtained by phase modulating the N subcarriers and every subcarrier are spaced apart by 1/tb duration, which forms an OFDM signal. The detection range for given radar is independent of the waveform which led to the implementation of MCPC signal using orthogonal transforms. The sidelobe level and Peak to Mean Envelope Power Ratio (PMEPR) are the problems which should be addressed to improve the performance of radar. Simulation result shows, the proposed method called Two Samples Sliding Window Adder (TSSWA) algorithm results in lower sidelobe levels.

Internet of Things (IoT) is a revolution in our daily life ranging from tiny devices to large industrial setup. Since the last decade, the proliferation of IoT has raised major security concerns which are not often considered by the manufacturers and the end-users as well. IoT devices are more vulnerable to attacks due to weak passwords, lack of standard architecture, availability, and plug and play services, etc. Most of the existing IoT security literature primarily focus on network layer security. However, this paper mainly focuses on the application layer security of IoT for secure cloud-based sensor data handling. We provide a detailed taxonomy of various attacking and security tools at the application layer of IoT. We also provide a comparative analysis of these tools. Finally, a Denial of Service (DoS) case study is performed in the simulated IoT testbed environment integrated with Amazon Web Services (AWS)-IoT using open source tools. We hope that our work will be helpful to the researchers working in the area of IoT security.

The advancement in technology has provided a lot of ease and solved many safety concerns in various fields. It had also catered various applications in automotive sector and has the capability to solve safety issues in order to prevent the occurrence of accidents to a large extent. Accident Prevention of Automobile using Real-time Tracking (APART) system is one such system which uses the real time tracking of all the vehicles and provides a map view containing real-time plots of all registered vehicle to each of the drivers, So that the driver will have prior knowledge of vehicles in its vicinity to avoid the accidents which would predominantly occur at the junctions, fog areas and blind curves, moreover it also increases the fuel efficiency by reducing the braking and accelerating instances. This paper presents a complete overview of APART system comprising of a raspberry pi and a GPS module to send the location to the real-time database and a map development process by using Google API key in order to display the resultant map.

Time series Forecasting has attracted attention over the last decade with the boost in processing power, the amount of data available and the development of more advanced algorithms. It is now widely used in a range of different fields including Medical Diagnostics, Weather Forecasting, Financial time series etc. In this paper, we propose a model of attention mechanism that allows for attended input to be fed to the model instead of the actual input. The motivation for the model is to show a new way to view the input so that the model can make more accurate predictions. The proposed LSTM model with the attention mechanism is then evaluated on common evaluation metrics and the results are compared with state of art models like CNN-LSTM and Stacked LSTM to show its benefits.

Network function virtualization (NFV) and service function chains (SFCs) effectively improve the flexibility of network service provisioning and increase the extent to which scaling can be done. However, finding an efficient deployment of virtual network functions (VNFs) for steering service function chain (SFC) requests is an NP-hard problem. The objective of our study is to obtain an optimal communication cost in VNF deployments and to allow for effective traffic steering in NFV enabled multicasting, when the number of SFC requests is large. Specifically, we have first formulated the problem and proved that it is NP-hard. We then present a 6 $$\alpha $$ α approximation algorithm for the centralized approach, where $$\alpha $$ α is the approximation factor, and a O(logn) approximation algorithm for the distributed approach, where n represents the number of nodes. Through extensive simulations on synthetic and real-world networks, we have evaluated the performance of our proposed approach on multi-cast traffic to be better by 18.06% as compared to current state-of-the-art algorithms.

LPWAN technologies such as-LoRa, Sigfox, and NB-IoT has been the key enabler in the advancement of Internet-of-things and Industry 4.0. LPWAN technologies known for long distances communication in low power devices at low operation cost for battery powered things. In this paper we have studied the performance of LPWAN in peer-peer (P2P) models for IoT-applications. We present a comparative analysis with respect to minimum data distribution delay for P2P system among LoRa, Sigfox, and Nb-IoT technologies, which shows that NB-IoT is best in LPWAN.

IoT is a buzzword nowadays and of course, it should be. The widespread of electronic and electromechanical devices with connecting ability to the Internet makes IoT be dominant from the user, manufacturer and services/goods provider perspective. Via IoT, the status of almost anything can be tracked, configured and maintained by different computing techniques using user devices or remotely from server ends. Determination of status can be easily known with data mining techniques that follow a distinct ladder until the representation of knowledge. In this survey work, we examined articles published from 2010 to date in the area of IoT. We followed a systematic literature review approach and scrutinize the different data mining steps followed by various scholars, and further classify the data mining techniques used in IoT as a conventional and non-conventional approach. Data cleaning, regression, model visualization, and summarization techniques were considered as challenging tasks due to the nature of IoT settings. This in turn demanded a new direction of research so as to come up with enhanced service provision in the area of IoT. Overlooked data mining techniques and comparison of the different approaches were criticized and reported. Moreover, the interdependency of IoT technologies with data mining approaches is discussed. Ultimately, an attempt has been made to indicate the research trend of IoT.

Community question answering (CQA) websites like Quora, Yahoo!Answers, Reddit enables users to ask questions as well as to answer questions. These sites are online communities that are popular now a days on the internet due to the increase of Question Answering (QA) websites and covers a wide variety of topics. Answer Processing task is classified as the ranking of answers, selection of answer through voting correlation, predicting the answer, selecting an appropriate answer from the candidate answers by classifying answer in good, bad, and potential category and then performing Yes/No task on selected answers or through best answer prediction or best answer selection. The shortcomings in the current approaches are the lexical gap between text pairs, dependency on external sources, and manual features which leads to a lack of generalization ability and to learn the associate patterns among answers. These shortcomings are resolved by already proposed work but they lack generalization ability and their performance is not satisfying. Feature extraction based methods mostly involve manual featurization which are not generalized form, therefore it can be avoided by deep learned feature. Whereas to focus on rich quality answers attention mechanism can be integrated with the neural network.

Human emotions can be identified in numerous ways, ranging from analyzing the tonal properties of speech to the facial expressions created before speech delivery and even the body gestures that can suggest various emotions without saying anything. Knowing the correct emotions of an individual can help is understand the situation and even react to it. This phenomena is even true for many feedback system used for day-to-day communication with humans, specifically the ones used for smart home solutions. The field of automated emotion recognition involves use-cases in different fields of research from computer vision, physiology to even artificial intelligence. This work focuses on classifying emotions into eight categories which are neutral, happy, sad, angry, calm, fearful, disgust and surprised based on the way those sentences have been spoken, using the “Ryerson Audio-Visual Database of Emotional Speech and Song” (RAVDESS). We propose a novel approach for emotion classification of audio conversations based on speech signals. Acoustic properties based emotion classification is independent of any spoken language and it can be used for cross-language emotion classification. The aim of the contribution was to develop a system capable of automatically recognising emotions for real-time speech. We performed several simulations and were able to achieve the highest accuracy of 82.99% with our shallow CNN model.

Cloud computing leverages virtualization as the most popular technique to deploy enterprise applications on virtual machines. Since the cloud system dynamically adapts to workload changes depending on the time of the day. It is required to ensure elasticity as a robust technique to efficiently model the changing workload requirements. However, it is an extremely challenging task, as several users may enter and depart from the cloud system over time. Predicting the different resource usage metrics of dynamically arriving jobs can help the cloud service providers (CSPs) in better capacity planning to fulfill the service level agreements (SLAs). In this paper, we propose a k clustering-based stacked bidirectional LSTM (BiLSTM) deep learners to model the multi-variate resource usage predictions for highly varying cloud workloads. We evaluate the proposed model on the Google cluster trace and validate its performance with the current approaches.

It is a very challenging task to recognize unconstrained Bengali handwritten text due to its cursive nature. This paper introduces an offline technique of recognizing handwritten Bengali sentences based on BiLSTM architecture and connectionist temporal classification (CTC) output layer. The traditional approach of detecting handwritten sentence recognition rises the computational complexity due to the difficulty of combining isolated characters. The proposed method only segments the sentences into words, recognizes separately and combines them to represents the output sentence. The proposed technique is evaluated with 450 Bengali sentences and achieved 92% accuracy for recognizing sentences considering 15 distinct writing styles.

The development of robots has enabled humans to allow them to carry out simple tasks and activities for easier and faster living. Robotic vehicles are found in our houses and industries and would continue to be integrated into more aspects of human living. In the present age of technology, most robotic systems found in industries and our everyday lives are controlled by a traditional input method or are pre-programmed. This makes them a little tedious for operation by a common man without technical know-how or a certain level of expertise. Due to this, there is a need for a natural mode of communication between humans and robots. This study develops a robotic vehicle system that can be controlled by a user’s simple hand gestures. The system employs the use of an accelerometer to detect hand gestures and then processed by Arduino boards to relay instructions to the robotic vehicle. The development of a hand gesture control system for a robotic vehicle provides a more natural mode of communication for human-robot interaction. This technology would increase the applications of robotic vehicles in various sectors through its implementation.

The rapid growth of location-based networks during the twenty- first century has greatly increased, so there is a need of providing suggestions to personals about their interest activities. Nowadays, location-based social networks (LBSN) become a common platform for users to share interests. In this paper, our main concern is to design a recommendation system that will provide suggestions to the user according to their interests. We have developed a framework based on Collaborative Filtering (CF) that analyses user activities to find the similar user. CF helps us to enrich each user profile by rating unvisited places which we can include in their interest hierarchy. Then we calculate the similarity of the user profile with the Point of Interest (POI) extracted from the user’s current location and make recommendations. Here Grid Structure is used to analyse the POIs extracted from Google.

Most of the classification problems in the real-world suffer from class imbalance. The performance of traditional classification algorithms is biased towards the majority class while handling a class imbalance problem. Several methods have been used to handle the class imbalance problems. These methods are based on three main approaches which are data level, algorithmic, and ensemble. Weighted Kernel-based Synthetic Minority Oversampling Technique (WK-SMOTE) is a recently proposed method that utilizes the benefits of both the data level approach and algorithmic approach. Inspired by the idea and performance of WK-SMOTE this work proposes a novel Synthetic Minority Oversampling based Weighted Kernelized Extreme Learning Machine (SMOTE-WKELM). SMOTE-WKELM is a variant of Weighted Kernelized Extreme Learning Machine (WKELM), which uses SMOTE for oversampling of the minority class instances. Experiments are performed on 15 datasets with varying imbalance ratios, downloaded from keel dataset repository for performance evaluation. The results on these datasets show that the proposed method performs better than the other state of the art methods in consideration.

Virtual wireless sensor network (WSN) is an approach towards automated monitoring system with minimum cost and maximum throughput. vWSN is an approach to convert physical WSN [1] into virtual device using VM and docker container combined with Bluetooth 5, this designed can be further transformed into on demand IOT device that is efficient, secure and real time device used for monitoring system. In this manuscript we will propose the end to end design to achieve the desired result having lot of advantages. This model can be deployed on the cloud. We will also discuss the security aspect of vWSN.

Breast Cancer is one of the most prevalent diseases among women. Its early diagnosis helps to increase the survival rate. Among many modalities, thermography is considered to be an early diagnostic procedure, which depicts the temperature values of the hot regions and further provides scopes in locating the tumor. In this work, features from Gray Level Co- occurrence Matrix (GLCM) and Gray Level Run Length Matrix (GLRLM) are extracted using the breast thermograms. Dimensionality reduction technique i.e., Autoencoder is applied to the extracted features. It gives the non-linear pixel intensities of the breast thermograms. Further, the reduced feature set is directed towards the statistical analysis of the features with three different methods viz. Filter, Wrapper and Embedded methods on the breast thermograms for the selection of best features set. Random Forest and Decision tree based classification algorithms are further applied for the features selected using three statistical tests. Among both the classifiers, Random forest with Recursive feature Elimination method gives a better performance in detecting the tumors between healthy and unhealthy breasts, giving an accuracy level of 81.63%.

Photo steganography is a rising research field for improvising secure data hiding and transmission over the network. The base concept of the proposed system which is Least Significant Bit (LSB) based Steganography along with Visual Cryptography using Genetic Algorithm is implemented over a network of distributed systems connected to each other via LAN. The original message is converted into cipher text and then hidden into the Least Significant Bit – LSB of the original image. Visual Cryptography and Genetic Algorithm are implemented for enhancing the security. Genetic Algorithm is basically used to modify the pixel location of steganography image and thus making the detection of the message complex. To encrypt the visual information, Visual Cryptography is used. To achieve it, we break the image into two shares based on a given threshold. The complete workload of the above proposed systems is divided among Distributed Systems with the help MPI, NFS and SSH and the work is done in parallel among the machines in the distributed network. The purpose of the proposed method is to improve the speed of execution of the enhanced secure algorithm to make the proposed system viable to larger data and a wider spectrum of applications.

The term Internet of Things refers to the huge network of interconnected smart devices (sensors, actuators, RFID tags and readers) that can share and communicate information without any human intervention. One of the diverse applications of IoT is in the field of dairy farming for cattle behavior classification, automated heat detection (time period when the cows are sexually receptive) and calving time prediction. In this paper, first the work proposed by various researchers for cattle behavior classification using supervised and unsupervised machine learning techniques has been summarized. One of the limitations found in previous work was that the researchers used LM35 sensor for cow’s body temperature measurement which read environmental temperature readings thus giving inaccurate results. Second limitation was quick discharge of wearable device due to high power consumption modules like Wi-Fi module. The proposed model discussed in this paper overcomes these limitations. The sensor system is divided into two parts: wearable transmitter module and receiver module. The wearable collar transmitter module consisted of only sensors and wireless transceiver module. It uses contactless infrared temperature sensor (Tmp006) instead of LM35. The high-power consumption modules were placed in the receiver module which had continuous power supply. Hence, the power requirements were minimized and the battery life was increased in the wearable from few hours to 4 days.

The identification of correct edge location is an important step in edge based image steganography. The edge pixels in an image are characterized by sudden sharp change in intensity; changes made at these locations are inconspicuous to the human eye making it optimal for modification. In this paper, use of Fuzzy inference system to identify the correct edges have been extended to perform edge image steganography, that hides secret message using a hybrid LSB technique. The aim of the proposed hybrid LSB method is to achieve high payload embedding that ensures statistical resemblance between the cover and the generated stego image. It embeds two message bits in the seventh and eighth edge bit using LSB-Matching (LSBM) and LSB-Replacement (LSBR). The performance of our method shows substantial improvement in terms of security and image quality compared to existing edge based image steganography approaches. The experimental analysis shows that the stego image achieves high PSNR of 61 dB and SSIM of 0.99 with minimum level of distortion at high embedding rate observed by gradually increasing % pixels modifications. The method is validated to be secure against RS analysis and Pixel-wise Histogram steganalysis, tested with 2000 grayscale images from BOSSBase 1.0 database.

Extreme Learning Machine (ELM) is proved to be fast, cost-effective and efficient for solving the binary classification problems. On the other hand, learning with the universum samples is also considered useful as these samples are incorporated as the prior information with the classifier. In our proposed work, we design the novel Universum based nonparallel kernelized Extreme Learning Machine (UNPKELM) which is a variant of Extreme Learning Machine. The two ELMs are learned and trained with the two nonparallel hyperplanes and hence make a robust classifier. This work also presents the mathematical formulation of UNPKELM and equations are derived to determine the output weights. Our proposed classifier UNPKELM is evaluated using 10 benchmark datasets and its performance (G-Mean) is compared to KELM, WKELM, CCR-KELM which verifies its effectiveness.

The authors propose a digital image watermarking mechanism to improve data security, robustness, and capacity by combining singular value decomposition with the discrete wavelet transformation. In the proposed approach, an image is decomposed horizontally as well as vertically into their low and high sub-frequency bands, respectively by using DWT. A horizontal low-frequency sub-band is selected and converted into an M × M block size. After that, the watermark images are embedded into each block using the SVD approach. After the completion of this process, where the watermarked images were subjected to various kinds of attacks such as cropping, rotation, sharpness, noise, contrast adjustment, filtering etc., the inserted watermark image was recovered from the cover image. The quality of recovered watermarks was compared based on correlation coefficients. The proposed algorithm’s robustness was found to be comparatively better than the previously existing methods. Tables show the results.

Medical data contains sensitive information and can have big impact if stolen. The term medical identity theft is coined for such thefts. Medical records are collected from various sources like hospitals, diagnostic labs, physicians, pharmacy and health insurance companies and includes all details of patient including his demographic information, test reports like X – rays, CT scans, MRIs, etc. With the advent of digitization, these medical records are now stored in digital form to make access and sharing easier. However, storing and sharing these data electronically opens the threat of data theft and misuse. Health insurance companies often bear the brunt of fraudulent claims based on stolen medical data. So, the current need is to enable storing and sharing these data with security and make a prohibition on making copies of such data. Hence, considering the importance of healthcare data, the use of blockchain technology can be promising to maintain the security, privacy, immutability, authentication, and reliability of the data to the intended users. The aim of this paper is to investigate the current state of blockchain technology and its diverse applications. The paper also discusses the application of the blockchain technology in the domain of secure medical health record sharing.

Memory design is the most complex part of any electronic devices. In this paper, comparative analysis of static random access memory (SRAM) cells such as conventional 6T, 9T and 10T has been performed and are simulated on cadence virtuoso tool using 18 nm Fin-FET technology. Fin-FET technology is more preferable for low power devices. The SRAM characteristics likewise delay, power and stability are determined out with cadence virtuoso tool. The read and write access time of 10T SRAM cell has been found to be 54.62% and 73% better as compared to Conventional (Conv.) 6T SRAM cell. The read stability and write ability of 9T SRAM cell is 1.21 times and 3.15 times better as of Conv.6T and 10T SRAM cells respectively. The read and write powers in 10T SRAM cell have been observed to be 73% and 31% less as compared to Conv.6T SRAM cell at 0.6 V supply voltage.

Today’s large enterprises run on a wide variety of independent applications, data sources, and infrastructure which is either disconnected or tied together using point-to-point integration or integration using middleware technologies. With the focus on Move to Cloud many new applications are hosted on Public cloud while many legacy applications, for a variety of compliance and security reasons, are still hosted on On-premises data centers or in private cloud across different networks/firewalls. In recent years, Hybrid cloud is a promising paradigm in cloud computing space. It provides seamless integration between on-prem, private and public cloud services and applications. There are many advantages to move to a hybrid cloud as compared to the existing middleware network. Migration to hybrid is as simple as ‘a lift and shift’, provided, the applications are containerized and cloud-ready. The native non-cloud applications can still use the benefit of hybrid cloud by replacing legacy middleware tools and achieving a point-to-point integration with hybrid cloud clusters. In this article, we explore the current middleware landscape with it’s associated pitfalls and compare it with the Hybrid Cloud approach. Additionally, this article will cover a brief introduction about Redhat OpenShift, a hybrid cloud product from IBM.

Authors have proposed the optimization of Multirate Finite Impulse Response (FIR) and Polyphase filters with the Multiple Constant Multiplication (MCM) shifting-adding concept. Proposed logics have reduced Multiplier, adders, and latches in the design. Due to which, it reduces area, power dissipation, and the circuit complexity in the system to a great extent. The primary objective is an optimization using Multirate DSP rather than targeting a single rate Digital Signal Processing (DSP). The proposed system approach can implement for the improvement in the important Parameters area, Power Dissipation, Speed and complexity. An approach provides the design and testing of the FIR filter, Multirate Polyphase filter using multiple constant multiplication and digit-serial architecture. It is an effective way to reduce the adders and subtractors in the Multirate finite impulse response filter. Investigational results are shown the effectiveness of the projected technique and the importance of different architectures. The simulation of essential parameters of the design is analyzed by using the software Active HDL, Synopsis 45 NM, and Xilinx. The Result is verified on the FPGA platform.

Internet of Things is found everywhere in today’s life, they are simplifying our life and are reliable. As they are connected to the Internet for communication, they are vulnerable to security attacks. Here, we have proposed a Key generation protocol for Internet of Things (IoT) devices for securing the communication between them. The proposed protocol is based on the discrete logarithm problem (DLP) and is secure. In the proposed mechanism the nodes/entities generate the session key and the communication is facilitated by server. The session key is generated by the individual nodes. The mechanism uses group theory and DLP. As the key is generated by the nodes and no exchange of key is done, the proposed mechanism is secure. It requires very fewer resources and is suitable for IoT networks.

The introduction of electric vehicles (EVs) brings out various challenges like deploying more charging stations, building its supportive infrastructure, managing the EVs and their various resource requests. In order to maintain all the resource requests of the EVs, network slicing is used which provides an efficient way to satisfy various use case demands of the EVs. In this work, we perform network slicing that partitions the physical network into three slices, infotainment and safety message slices that belong to downlink communication and charge state information slice that belongs to uplink communication. If there are large number of resource requests made by the EVs to the slices, it might lead to collision in the channel. Unsupervised machine learning is performed on the EVs using local scaling as the scaling parameter which handles multi scale data and also performs better clustering. For an efficient communication between the clustered EVs and charging station, slice leaders of every cluster is determined. We have also proposed an algorithm that efficiently performs resource allocation to the EVs to increase the throughput with low latency. Slice leaders forward the resource requests made by the EVs of the respective clusters to the charging station through RSUs and a slice block allocation is performed by giving higher preference to the critical message requests.

Investing in the financial market is not a new or exceptional case in real. Recently, many people are interested to invest in financial market to maximize the ROI. To invest in financial market is not always gives a profit. So for investing in market need more analytical research or study or expert knowledge to identify the best investment which maximize the ROI. Every investor doesn’t has expertise to perform analytical study so these type of modelling helpful them to increase the profit. This paper focus on the investment strategy base available methods and perform comparative analytical study to provide the optimize investment strategy for peak bulk deal selection. Predominant bulk deal picking is basic among dynamic value subsidize directors. The explanation bulk deal picking aptitude isn’t identified by numerous exhibition contemplates is over-enhancement, which overpowers the prevalent execution of top possessions. Utilizing self-announced model for arranging the dynamic value universe liberates supervisors to seek after a barely characterized technique and aides in distinguishing effective chiefs inside every bulk deal.

Searching of the target(s) in unknown Scenarios is a very complex task in the field of Robotics. This paper is an introduction to the works that are seminal in the field of Swarm Robotics. Swarm Robotics is the application of the principles used in swarm Intelligence. Different types of algorithms are used to decide the path in different types of situations or environments. Due to the robustness, scalable, fault-tolerant, and many other properties of the swarm of robots, the SRS is used in many search and tracking applications. This is the review that focuses on the different problems in SRS. Firstly an algorithm that deals with the local minima problem is explained, further, there are two more algorithms that are used to search the target and guide the robots to the target are explained. These algorithms perform differently for different scenarios. Further, seven different types of algorithms are compared with each other based on some parameters. Most of these algorithms perform well in some situations but may fail to perform well in some other situations.

Cricket is most lovable game in the world. It is played in many formats. One of the most popular format is IPL (Indian Premier League) which is hosted by BCCI (Board of Control for Cricket in India) every year. IPL is involved in business and money as the players across the globe are auctioned by the businessmen to form their teams. This paper presents the analytics of IPL match using the dataset of previous year matches i.e. from year 2008 to 2019. Various attributes like DLS method applied, venue, toss decision, toss winner and many more have been analyzed to check whether they contribute in predicting the winner or not. In this work, some graphs has also been plotted to visualize the performance of teams and players on the basis of different attributes.

Anomaly detection is an effective approach of dealing with problems in decision making process. Rapid development in technology has elevated the requirement of resourceful detection system using machine learning, deep learning in order to detect new and advanced outliers. Most of the real time applications employed with wireless sensor networks, which are positioned in unkind and unattended atmospheres, where these situations turns to a major causes on the production of anomalous or low quality sensor readings. The erroneous and unreliable readings may increase generation of false alarms and erroneous decisions; hence it is essential to identify outliers resourcefully and exactly to make sure the authentic decision-making. In this paper, first simple cluster algorithm performed based o residual energies of sensor nodes and then denoising auto-encoder with Gaussian kernel applied on each cluster head to detect outliers. Experimental analysis shows that designed technique achieves high detection rate as well as low false alarm rate.

Attacks through the web pages containing malicious content have become an increasingly threat to the web security in the recent years. Thus, detection of the malicious URL is an important task to reduce the security threats. To detect malicious URL or web pages, there are several ways but the most traditional technique is through the Black List detection. The Black list contains the list of malicious web pages that are maintained so that user can be aware about the web pages before accessing any webpage. But, the problem with the black list is that it is not an effective method as malicious web pages change frequently, and also growing numbers of web pages that pose scalability issues. A part from blacklist technique, various research techniques have been proposed that use machine learning technique and some use CNN (Convolution Neural Network) to classify web pages into category: malicious or benign. In the paper, a literature survey on classification of malicious web pages is presented that compares various machine techniques with parameter: precision, recall, and F1 score. This survey shows that the Machine learning techniques are better if the features used are textual but when there are images in web page, CNN performs better for the malware image classification.

The acceptance of various computing aspects of cloud-based systems gets hampered by the evolution of various security threats. Security and privacy issues are considered as the primary challenges to an adoption of cloud computing. In order to develop a secure cloud, there is a need of proper analysis of security threats and their associated detection as well as prevention techniques. In this study, an attempt has been made to offer an ontology-based analysis and design approach for the cloud security. The security provision of cloud systems is accomplished by using security patterns, which are often specified by using UML (Unified Modeling Language) diagrams. In order to specify cloud pattern notations, a semantic modeling approach i.e., Web Ontology Language (OWL) has been considered. Description Logic (DL) has been considered for analyzing security requirements, which is supported by the OWL editor i.e., Protégé.

Network slicing is the key technology in 5G wireless communication, which aims to provide services based on latency, availability, reliability, throughput and more. With the rapid development of mobile networks and new networking applications, it is turning out to be more difficult to meet the Quality of Services (QoS) under the current mobile traffic and mobile-network architecture. Mobile Traffic forecasting is one of the domains that can benefit the mobile companies in optimizing their resources. In this paper, we consider a dataset with Internet usage patterns by users over a period of six days. Based on past time-steps trends we tries to predict the current network slice that would be classified into streaming, messaging, searching, and cloud classes. We compared the four deep learning models namely MLP, Attention-based Encoder Decoder, GRU and LSTM and we evaluated these models on recall, precision and f1 score performance matrices. We found that MLP, Encoder-Decoder models performed average for mobile-traffic forecasting while the GRU, LSTM performs well and out of them LSTM obtained best result.

Traditional light weight wireless communication modules like ZigBee, Bluetooth, LoRa have fixed frequency band and fixed protocol stack and hence lack flexibility of changing frequency, modulation techniques, waveforms etc. and may also prone to interference, jamming. Due to obstacles in between the transmitter and receiver, the communication coverage range is not large. But in emergency situations like war field or any disaster, there is need for robust communication in which the soldiers or public have to receive crucial information from base station. Software defined Radios like Universal Radio Software Peripheral (USRP), WARP boards needs to interface with computer for signal processing. Hence it is not suitable for drone based communication. So we proposed a solution to overcome bulkier system and limited communication range problem. In this work, a light-weight communication platform was developed which acts as a repeater for broadcasting voice commands that were transmitted from base station. The repeater is mounted on the drone and it is hovered at a certain altitude to make line of sight communication possible with both transmitter and receiver. This system was designed by interfacing Radio Frequency (RF) board present in USRP 2900 with Raspberry Pi 4, a credit card sized computer. The RF board overcomes the difficulty of standard wireless communication modules mentioned above, could easily mounted on drone, capable of communicating in different modulation schemes at multiple frequency bands and also one could implement cognitive radio features like spectrum sensing and taking automatic decisions based on data which is obtained from various sensors that are connected to Raspberry Pi.

Software reliability depends on the number of faults present in the software system (more number of faults in the software means reliability is less and vice versa). In this paper, two software fault prediction models are developed using Support Vector Machines (SVM) and Multi-layer Perceptron (MLP). Genetic Algorithm (GA) is applied to select the best features. Then these reduced features are considered as input to the two classifier algorithms such as SVM and MLP. The developed models are applied on 6 different datasets collected from Github Repository (Nasa Defect Dataset). From the results, it is observed that MLP with GA model is performing better than other models.

Fault prediction is the major step in large industries where the complexity of the software is rising at an exponential rate. Assigning the proper damaging level of faults encountered in complex and large software, would help the developers to plan for fixing the faults. Traditional fault prediction studies mainly concentrate on designing hand-crafted features, which are input into machine learning classifiers to identify faulty modules. In this paper, we have developed a model based on deep learning techniques. Some of the extracted features from programs are used to train the Deep Neural Network and then other features are passed for testing. The proposed model is validated using open source Promise data repository. It is clear from the results that the performance of predicting faults by the proposed model is better than the existing models.

Recently, the classification of the head pose has gained incremented attention due to the rapid development of HCI/HRI interfaces. The resoluteness of head pose plays a considerable part in interpreting the person’s focus of attention in human-robot or human-human intercommunications since it provides explicit information of his/her attentional target. This paper proposes a geometrical feature-based human head pose classification using deep convolution networks. An MTCNN framework is implemented to identify the human face and a ResNet50 layered architecture built to classify nine head poses. The system is trained with 2, 85, 000 and tested by 1, 15, 500 head pose images. The proposed system achieved $$90.00\%$$ 90.00 % precision for nine head pose classes.

Recently, many multifactor remote user authentication schemes have emerged to cover-up the security weaknesses of single factor user authentication systems. Among these schemes, the two-factor multi-server user authentication scheme has drawn a considerable amount of attentions of researchers. Subsequently, many two-factor user authentication schemes for multi-server architecture have been introduced in recent past. However, they were unable to prevent the security vulnerabilities like password guessing attack, user impersonation attack, privileged insider attack, server masquerading attack, denial of service attack, replay attack, etc. To address these security deficiencies, we propose a robust two-factor remote user authentication scheme for multi-server architecture using smartphone and biometric that can prevent all the major security vulnerabilities shown by existing schemes. We use ProVerif to demonstrate that our scheme fulfils all the required security properties. Through a comprehensive heuristic security and performance analysis, we show that our proposed scheme can overcome drawbacks of existing systems.

Packet error in IEEE 802.15.6 MAC protocol is one of the main sources that degrades the performance and its inconsistency. This protocol does not have the ability to control the packet errors which is caused by the transmission failures or collisions. To alleviate the issues presented in the traditional protocol, this paper aims to develop an analytical model $$(p_{r})$$ ( p r ) to study the effects of packet errors and validate the performance of the IEEE 802.15.6 MAC protocol using Carrier Sense Multiple Access with Collision Avoidance (CSMA/CA) mechanism. The $$p_{r}$$ p r -model decreases the packet error rate by incorporating the optimum number of Retry limit ( $$R_l$$ R l ) and Contention Window (CW) size for all User Priority (UP) nodes. Therefore, this model can be deliberated as the first-ever analytical model for optimizing the UP and $$R_{l}$$ R l assignment in order to improve the overall performance of the Wireless Body Area Network system. The simulation results suggest that minimizing transmission failure probability has more effect on saturation throughput than multiplying the CW size after even number of failures.

Web text categorization is a procedure of deliberately assigning a web text document into one of the pre-defined classes or categories. It is a very challenging task to manipulate, organize, and categorize an enormous amount of web text data in manually. This paper proposes an automatic text categorization framework to classify Bengali web text data using deep learning. The proposed framework comprises of three key constituents: text to feature extraction, training, and testing. The categorization framework is trained, validated, and tested at 120K, 12K, and 36K datasets, respectively. The proposed system achieved $$99.00\%$$ 99.00 % accuracy in the training phase, $$96.00\%$$ 96.00 % in the validation phase, and $$95.83\%$$ 95.83 % in the testing phase, respectively.

Stability and power consumption in the circuit are the major column of any SRAM cell idolization. In this paper, a read decoupled 8T SRAM cell has been implemented and compared with conventional 6T SRAM and Differential 8T SRAM cell, analyzed on 45 nm technology with Cadence virtuoso tool. It has been observed that read stability and write ability of considered cell is improved by 1.4× and 1.02 × in comparison of conventional 6 T SRAM cell while 1.7× and 1.07× with respect to Differential 8T SRAM cell. The read delay and write delay of read decoupled 8 T SRAM cell is 46.40% and 29% better in comparison of conventional 6T SRAM cell. Additionally, the read power of 8T SRAM cell is reduced by a factor of 55% as comparison of conventional 6T at 1 V supply voltage.

VLSI designers are inspired by the widespread use of portable low power devices. In this paper, a 9T SRAM cell has been analyzed and implemented at 45 nm technology node with Cadence virtuoso tool. The read stability and write ability of considered cell is improved by 2.05 $$\times $$ × and 1.13 $$\times $$ × in comparison to conventional 6T SRAM cell. The write access time of 9T SRAM cell is 3.37 $$\times $$ × and 2.94 $$\times $$ × better in comparison of conventional 6T and differential (DF) 8T SRAM cell respectively. Furthermore, the write power of 9T SRAM cell is reduced by a factor of 2.07 $$\times $$ × and 1.77 $$\times $$ × as comparison of conventional 6T and Differential 8T SRAM cell respectively at 0.5 V supply voltage. The data retention voltage of 9T SRAM cell is better at all corners in comparison of conventional 6T and differential 8T SRAM cell respectively. The 9T SRAM cell may be utilized in IoT based devices such as medical equipments, space applications, etc.