Artificial Intelligence and Evolutionary Computations in Engineering Systems

In order to have a clear understanding of the market structure as well as the customer trends toward various products, there is a need for every company to collect, monitor, and analyze the user data generated online. In this paper, the online reviews of products from two leading camera manufacturers have been utilized to analyze the user trends. After preprocessing the data, sentiment analysis techniques have been employed to mine the textual content of customers’ opinion and classify them into different polarities according to the theoretical conceptualization of service and performance. The sentiment analysis results using Support Vector Machine provide a high level of accuracy in encapsulating and measuring the sentiments of customers toward the products and services as compared to the other text mining strategies. Further, a competitive analysis technique based on K-means clustering has been implemented to examine the most frequent word which is discussed by the customer. The combination of these two methods provides benefit not only to obtain the best classification but also to help the user focus on the most relevant categories that meet his/her interest.

The networking is the backbone that supports the vast area of Information Technology. SDN is the new road that takes the conventional networking to greater heights. SDN is going to aid all future innovations and developments in the field of networking. SDN stands for Software Defined Networking, this separates the network into two planes namely data plane and control plane. A data plane is the abstraction of all the hardware side of the network and the control plane is the central unit that acts like a brain controlling the entire network. This dual architecture thus helps to maintain a network that is centralized, highly scalable, flexible etc. The programmability of the network opens the window of scope for greater innovations and developments. SDN can gracefully accommodate technology shifts. At the same time SDN posses certain security issues that need to be addressed. As a widely flourishing and developing networking method, these security issues need to be tackled. In this paper we are trying to address the security issue of rewriting flow entries in switches. We propose an algorithm for the detection of incongruence between firewall rules and flow rules and thus we overcome the threat caused by modification of flow entries. The proposed system is for Open Flow based Firewalls. The system is intended to boost the security capabilities of SDN, thereby minimizing some of the security challenges in SDN.

With the advancement of information technology, there is a visible change in economical, commercial, and technological ways. Privacy of information still remains the main concern. Audio steganography is one of the most popular techniques for shielding of information. To raise the surveillance, we focus on Audio steganography along with noise-free environment. For ensuring this, hiding in silence interval are considered because most of the current audio steganography embedding methods are improper for noise-free environment. This paper proposing an attainment upgrade in audio steganography with using RSA algorithm of three distinct prime numbers as well as hiding text in silence speech signals with high hiding capacity employing summate silence interval in speech signal. Our simulation results show that this approach maintains noise-free environment, security, and achieves a higher hiding capacity.

Path planning is an important aspect of any mobile robot navigation to find a hazard-free path and an optimal path. Currently, the A* algorithm is considered to be one of the prominent algorithms for path planning in a known environment. However, with the rise of neural networks and machine learning, newer promising algorithms are emerging in this domain. Our work compares one such algorithm namely the Hopfield neural network-based path planning algorithm with A* in a static environment. Both the Hopfield network and the A* algorithm were implemented while minimizing the total run times of the programs. For this, both the algorithms were run in MATLAB environment and a set of mazes were then executed and their run times were compared. Based on the study, the A* algorithm fared better and the Hopfield network showed promising results with scope for further reduction in its run time.

We propose a novel vision-based lane departure warning algorithm. The initial step of lane departure warning system is lane detection. In this paper, we present a lane detection method based on template matching. Kalman filtering is used to track the detected lanes. To make the algorithm robust to shadows, inpainting technique is done on regions with shadows. The curved lanes are handled with the help of Bezier spline fitting. Our algorithm determines the lane departure based on vanishing point margins. Experimental results show that the proposed method performs better than Caltech lane detection algorithm.

Tourism plays an important role on Indian economy. In this paper, we are proposing a system for finding valuable tourists and forecasting the arrival of the tourists. For analysis, we collected data through online survey and can make prediction on valuable domestic tourist arrival using classification based on association rules generated in a bidirectional approach (CARGBA). By predicting the tourist arrival in India, we can recommend valuable tourist destinations and can promote and improve the facilities in Indian tourism.

Data Mining is all about analyzing the data gathered from different sources, based on different context and summarizing it into meaningful information. The belief of utilizing design pattern to improve the data mining application is relatively high. Design Pattern is a broad and general repeatable solution to a problem that occurs frequently in a software design. There are only a few number of researches illustrating their benefit. In this paper, we try to expose the relation between data mining architecture and design patterns. We have here taken a layered architecture for data mining environment and analyzed the impact of design pattern in every components of given architecture. This paper presents a survey on various design patterns used in data mining which is used to fulfill functional and nonfunctional requirements (quality attributes).

In emerging lifestyle, people prefer online shopping as the most convenient way for purchasing their products. In order to provide secure transaction in e-commerce site, biometric features are incorporated in authentication. Biometric features are considered to be easier and secure because it solves the exasperation by the customer to recall the long account number and their password and it also has unique feature all over the globe. The use of biometrics authentication in online site provides more secure way of transaction than any other means [1–3]. In this work, we summarize the study of the various kinds of biometric traits that are used for online payment compared with one another.

Near field communication (NFC) is a technology that facilitates communication between NFC-enabled devices by utilizing low power and radiating signals around a close proximity. The interaction is always binary, between the initiator that emits signals within a range of 4 cm and the target which enters the field of the initiator to begin the interaction. The communication may be one way (passive) or two way (active). The data shared in the session is neither encrypted not authenticated. These two factors aid in potential communication and data transfer. However, this becomes downside to this upcoming technology, when the data provided by the initiator may be subjected to data spoofing or data corruption. When the target processes such data, it could leads to unintended behaviour thereby compromising the integrity of the device. When a forensic investigator is handed a compromised device and asked to recreate the alleged crime, he relies on the presence of nonvolatile data on the device. In the Android operating system, there is no mechanism to provide the nonvolatile artefacts ensuing an NFC interaction. Therefore, any digital crime on Android devices abetted by NFC remains unsolved and the case gets deferred. The main aim of this research is to develop a logging mechanism for Android devices that will log all the interactions taking place through the NFC hardware, and the presence of these nonvolatile logs along with other volatile artefacts would benefit the forensic investigator to comprehend the exact sequence of activities that jeopardized the conventional operation of the android device.

Aerial and ground robots have been widely used in tandem to overcome the limitations of the individual systems, such as short run time and limited field of view. Several strategies have been proposed for this collaboration and all of them involve periodic autonomous precision landing of the aerial vehicle on the ground robot for recharging. Intelligent control systems like neural networks lend themselves naturally to precision landing applications since they offer immunity to system dynamics and adaptability to various environments. Our work describes an offline neural network backpropagation controller to provide visual servoing for the landing operation. The quadrotor control system is designed to perform precise landing on a marker platform within the specified time and distance constraints. The proposed method has been simulated and validated in a Gazebo and robot operating system simulation environment.

Estimation of the channel has been one of the major concerns in any communication system. Although various channel estimation techniques are available for MIMO (Multiple Input Multiple Output) systems, the problem of wastage of bandwidth due to the transmission of training sequences prior to the transmission of actual data exists. Use of superimposed pilot sequence that accomplishes transmission of both the pilot and data simultaneously has been proposed as an effective alternative. The work carried out includes both simulation and real-time implementation of channel models and channel estimation techniques. Appropriate channel models chosen include shadowing model and Markov model to address large-scale fading effects, and Rician model for small-scale fading effects. Employing channel estimation techniques for the above listed channel models is the novel idea aimed for satellite applications. Also different algorithms for channel estimation like least square (LS), minimum mean square error (MMSE), and linear minimum mean square Error (LMMSE) have been compared. Furthermore, merits and demerits of superimposed and conventional pilots are analyzed by making performance comparisons. Hardware implementation is done using Universal Software Radio Peripheral (USRP).

This paper introduces an optimization technique that uses an artificial bee colony (ABC) algorithm to solve the fault section estimation (FSE) problem. FSE is introduced as an optimization problem, where the objective function includes the status of protective relays and circuit breakers. The ABC algorithm is a new population-based optimization technique inspired by behavior of the bee colony to search honey. In order to test the effectiveness of the proposed technique, two sample systems are tested under various test cases. Also the results obtained by the proposed ABC algorithm is compared with those obtained using two other methods. The results show the accuracy and high computation efficiency of the ABC algorithm. The ABC algorithm has a main advantage that it has only two parameters to be controlled. Therefore, the tuning of the proposed algorithm is easier and has a higher probability to reach the optimum solution than other competing methods.

Exponential growth of information in the web became infeasible for user to sieve useful information very quickly. So solution to such problem now a day is text summarization. Text summarization is the process of creating condensed version of original text by preserving important information in it. This paper presents for the first time a nature inspired cuckoo search optimization algorithm for optimal selection of sentences as summary sentence of intelligent text summarizer. The key aspects of proposed summarizer focus on content coverage and length while reducing redundant information in the summaries. To solve this optimization problem, this model uses inter-sentence relationship and sentence-to-document relationship by considering widely used similarity measure cosine similarity. The inputs for this model are taken from DUC dataset. Whereas the result is evaluated by ROUGE tool and compared with state-of-the-art approaches, in which our model in multi-document summarization have shown significant result than others.

The topographical features like craters are forming in different size and shape in different categories that will give more information about the planet. The 2D shape descriptor from top view of the crater is very difficult to interpret the crater type. Here 2D shape descriptor is developed based on the intensity values of the crater image. The reverse order summation and smoothing function is applied to these intensity values, which will depict the cross-section of the crater image. This can be used for classification, 3D model development, and to retrieve other information.

Determination of optimal thrust direction or steering angle for lunar soft landing trajectory is attempted in this article. The problem is complex due to the presence of system constraints and local minima. An exhaustive search of optimal thrust direction incurs high computational costs. The problem was solved as an optimum initial value estimation problem. Taboo evolutionary programming (TEP) is utilized to compute the optimal estimates. The study gives the integration of TEP technique in solving the governing nonlinear differential equations where a control parameter involved. The results are compared with available results in literature and it shows that the solution based on TEP algorithm is comparable to the counterpart. Further, sensitivity of design parameters such as terminal altitude, true anomaly, and terminal velocity over the final landing mass at the touch down is also examined.

In this paper presents the analysis of pre-filters based SRF PLL for distorted grid conditions. An exact detection of phase and fundamental frequency of grid current is essential for the control algorithm of grid connected power converter circuit. A control model of the SRF PLL is developed and is made on tuning the system under distorted grid conditions like harmonics and dc offsets. The SRF PLL can be completely implemented in software with and without pre-filters. The pre-filters are band pass and high pass filters these can be used to reduce harmonics present in the input and high pass filter alone can reduce the dc offset present in the input. The effects of high pass and band pass filters on dc offsets are analyzed. The superior performance of proposed pre-filter-based SRF PLL phase detection system is studied and the obtained results are compared with SRF PLL-based phase detection to confirm the feasibility of the study under different grid environment such as high-harmonic injection and dc offset. All analytical results are verified using MATLAB software.

Surmount demand requirement in electrical energy has given a provision of integrating renewable energy to the three-phase grid-connected system. In particular, the penetration of distributed generation system has contributed more in electrical power generation. But the system connected produces adverse effects in grid due to the unpredictable atmospheric conditions. To handle such condition, in this research article, PV system is considered under study and a new two-stage CPG-P&O MPPT method is used for interfacing PV with grid systems. In addition, a wide range of feed-in power to grid capacity at inverter level is briefly addressed. Further, the algorithm is designed to reduce the increased thermal stresses in switch and to get rid of the grid overshoot and power loss. The desired objective was acknowledged by the simulation results by ensuring delivery of optimal performance.

Lack of knowledge on the mechanical soil properties have resulted in large inaccuracy of the rover’s mobility prediction in the past. This paper deals with the prediction of mechanical properties of the soil based on the experimental and computational model of a six-wheeled rocker bogie rover. The work is divided into two parts. First, a physical model of the rover was fabricated and was made to travel on an unknown loose soil on earth. For this, a known reference value of revolutions per minute (RPM) was given to the direct current (DC) motors and the corresponding linear speed of the rover was measured. Next, a terramechanics based dynamics model was developed for a nominal value of the mechanical soil properties. The RPM needed to maintain the same linear speed as the experimental value was computed for the assumed mechanical soil properties. These soil properties were altered within a range such that the RPM obtained from the experimental and the computational results were similar to maintain the same linear velocity. The results were tested and validated for different RPM values for the predicted mechanical soil properties, which proved to be satisfactory.

The use of visible light communication (VLC) in the area of smart home automation is presented in this paper. The VLC link between two Arduino development board containing microcontroller to transfer the controlling ASCII character through LED and LDR is also discussed. The Proteus ISIS simulation software is used to realize the data transfer between two microcontrollers. Simulation result revealed that different electrical gadgets can be controlled by using VLC link. The Experiment is also carried out to show its validation.

There are several models of uncertainty found in the literature like fuzzy set, rough set, intuitionistic fuzzy set, soft set, and hesitant fuzzy set. Also, several hybrid models have come up as a combination of these models and have been found to be more useful than the individual models. In everyday life we make many decisions. Making efficient decisions under uncertainty needs better techniques. Many such techniques have been developed in the recent past. These techniques involve soft sets and intuitionistic fuzzy sets. It is well known that intuitionistic hesitant fuzzy sets are more general than intuitionistic fuzzy sets. In this paper, we define intuitionistic hesitant fuzzy soft sets (IHFSS) and we also propose a decision making technique, which extends some of the recently developed algorithms. We also provide an application from real-life situations, which illustrates the working of the algorithm and its efficiency over the other algorithms.

This paper presents a method for generating multi-document text summary building on single document text summaries and by combining those single document text summaries using cosine similarity. For the generation of single document text summaries features like document feature, sentence position feature, normalized sentence length feature, numerical data feature, and proper noun feature are used. Single document text summaries are combined after calculating cosine similarity between the different single document text summaries generated and from each combination, sentences with high total sentence weight are extracted to generate multi-document text summary. The average F-measure of 0.30493 on DUC 2002 dataset has been observed, which is comparable to two of five top performing multi-document text summarization systems reported on the DUC 2002 dataset.

The need to analyse social media content is on the rise with more people using it for sharing their day-to-day activities. This has prompted multiple studies in the area of social data mining. One innovative approach is to analyse tweets posted by the user in twitter, and generates a structure that uniquely captures all the information related to the user in a single location. This project collects metadata from twitter users, analyses their tweets and gets a set of associated probabilities for a set of topics generated from a Latent Dirichlet Allocation model. The collected information is mined and is stored in a unique data structure consisting of various layers that represent a user. This structure is stored using a hash function, which facilitates the easy storage and retrieval of the structure. A destination tree is built based on the data mined from the tourist websites of states in India, the leaves of the tree representing a tourist destination in India and each having its associated set of probabilities for a set of topics generated from the Latent Dirichlet Allocation model. Finally, the user node built is used to recommend tourist destinations in India by comparing the user node with the leaves of the tourist destination tree.

There has been an ever-growing demand for electrical energy causing supply–demand imbalance in the power system. To overcome this imbalance, a reliable, efficient, less centralised, and more interactive energy management system (EMS) has to be realized. One of the methods to bridge the supply–demand gap using EMS is through Load Management. The evolution of EMS allows loads to respond to the demand (Demand Response) and assist customers to make informed decisions about their energy consumption, adjusting both the timing and quantity of their electricity use. This paper deals with the development of Fuzzy logic-based load management scheme using Demand Response Programs in smart microgrids. The developed system is tested on a smart micro-grid simulator (SMGS) operated in islanded mode installed in the Renewable Energy Laboratory in Amrita Vishwa Vidyapeetham, Coimbatore.

Radar and satellite communication uses X-band antennas. Also, high gain and minimal size antennas are key requirements to latest wireless and satellite communication systems. The main parameters or constraints to the design of patch antennas include length and width of the antenna and substrate thickness and type. In this paper, the proposed 4 × 2 corporate feed microstrip patch antenna which is designed to operate at 10 GHz. It was found that the 4 × 2 patch antenna delivers a higher gain and return loss. The significant reduction in mutual coupling has been experimented using Defective Ground Plane (DGS). The paper introduces design of patch feed network using the high frequency structural simulator (HFSS) and the simulation results were compared with that of 4 × 1 patch antenna. The gain, return loss, and radiation pattern are analyzed and presented.

Movement is the main characteristic of living species and is the major source of biologically inspired computational systems. The migration of cell organism can take the form of either movement of cells or movement within cells, and it is also capable of changing the shape as a result of reversible or irreversible contraction. This paper simulates synergistic fibroblast optimization (SFO), a multi-agent heuristic technique that models the migration and methodical behavior of the fibroblast. The proposed SFO technique exhibits the role of fibroblast in dermal wound healing process, by migrating the individual cell through the connective tissue, and synthesizes the collagen in the extracellular matrix for the new tissue formation during wound healing. Compared to the related technique particle swarm optimization (PSO), SFO produces better results in terms of both accuracy and performance. An analysis of the proposed algorithm indicates that the two most important factors contributing to SFO effectiveness are fibroblast collaborative nature and goal-oriented feature. The results suggest that SFO is a promising new optimization technique, which may be particularly applicable to find optimal maxima or minima, among the candidate solutions in the nonlinear complicated optimization problem.

The aim of the paper is to design evolvable hardware (EHW) architecture for Finite Impulse Response Filter using Genetic Algorithm. Evolvable hardware refers to hardware that can change its behaviour (parameters such as coefficients) according to the changes in its environment. To update the filter coefficients adaptively, genetic algorithm was used. The proposed filter architecture was implemented with Xilinx Spartan 6 FPGA (XC6SLX45-CSG324) Trainer Kit. Hardware design was synthesized using the EDK (Embedded Development Kit) platform and the genetic algorithm was implemented in SDK (Software Development Kit) of Xilinx Platform Studio tool (XPS) 14.6.

As cloud computing technology becomes the major storage and sharing of data in Internet era, outsourcing cloud server for storing data becomes the latest trend. All small business and major social networking sites rely on cloud storage for dynamic data storage. Storing and maintaining cloud data are not easy, it require lots of effort and space to manage cloud data. Since the cloud is accessible by everyone connected in the network the security risk of cloud data is also high. To improve the cloud storage and make the storage more effective we are implementing the audit function in cloud data to reduce the duplication of data in cloud storage. This audit is done using file id and hash key generated for every file during the time of uploading. Proof of Ownership is used to track the owner of the files uploaded. Clustering of user data is included to monitor user file and improve service to the user for that particular type of file. To avoid data remanence attack in the cloud data, we are implementing a secure data deletion scheme which required key to delete the files which cannot be recovered during the attack.

There are several models of uncertainty found in the literature like fuzzy set, rough set, soft set and hesitant fuzzy set. Also, several hybrid models have come up as a combination of these models and have been found to be more useful than the individual models. In everyday life we make many decisions. Making efficient decisions under uncertainty needs better techniques. Many such techniques have been developed in the recent past. These techniques involve soft sets and fuzzy sets. In this paper we redefined the hesitant fuzzy soft sets (HFSS) with the help of membership function. We also provide a decision making algorithm.

Nowadays, Cloud computing becomes a looming computing prototype. Users can get a variety of services such as high computation power, storage, etc. Thus, applications of users can be more cost-effectively put on cloud by utilizing various commodity computers together. But, cloud computing faces some security concerns even if they provides many services. Some of such important concerns are data security and privacy of data. Some personal data like personal healthcare records and financial records contain sensitive information which can be analyzed and mined for public researches although these records offer important human assets. Data should be privacy preserved because malicious cloud users or untrusted cloud providers can get the data with less effort. To deal with these problems, privacy-aware set-valued data publishing on cloud for personal healthcare records has been proposed. An efficient privacy-aware system, named PHKEM (Personal Healthcare k-anonymity Encryption Model), for eliminating privacy breaches in publishing of personal healthcare data on cloud as well as data querying is designed. A data anonymization technique, named k-Anonymity with extended quasi-identifier partitioning (EQI-partitioning), interactive differential privacy, and AES encryption is applied to preserving personal healthcare records to prevent unauthorized access. Therefore, the security is efficiently enhanced with a natural and expressive fashion.

Research has witnessed a wide range of trust-based recommendation systems especially for P2P networks considering the open nature of networks. The recommendation systems have been devised with parameters, factors, and techniques to mitigate attacks on the recommendations. Though the ground functionalities are the same (trust management, reputation querying, recommendation filtering, and aggregation of fair recommendations) each of them have come up with different models which are unique in their own way. An important thread of research focuses on trust models for P2P systems based on the rule-based anomaly detection using genetic algorithm, evolving recommendations using genetic programming, etc. While, attempts have been made using binary coded GA, a decimal coded GA has not received attention so far. A decimal coded GA consumes less memory space and it gives accurate results. Our work focuses on evolving a self-organizing trust model for P2P systems using decimal coded genetic algorithm for mitigating service-based and recommendation-based attacks. A detailed discussion on responsiveness of P2P trust models to particularly malicious peer behavior, brings out the feasibility of the proposed model.

The systems in which missing values (NULL) occur are called incomplete information systems and computations on these may lead to biased conclusions. The structured difference of the datasets and importance of attributes compels us to depend on uncertainty-based approaches for finding the null values. This paper presents a hybrid approach for solving null value problems using the concepts of rough set theory and neural network. In this, complete tuple set is used for training the NN. The incomplete tuples are then tested using the model. Level of dependency is used to judge the importance of association rules [11]. Testing the dataset after reducing unwanted attributes, yields a reduced error percentage. The system produces result with better efficiency as observed by the values of accuracy, completeness, and coverage. Thus, the proposed algorithm can be suitably modified for different scenarios using the algorithm step-by-step to solve the null value problem.

Software defined networking (SDN) is an incipient network paradigm in which the control plane is moved out of the individual network nodes and into a separate centralized controller which is capable of exploiting the complete knowledge of the network to optimize flow management. SDN is a promising way to support the dynamic nature of networks at present and in the future. OpenFlow is the most commonly used SDN protocol. OpenFlow protocol governs the communication between SDN controllers and the underlying network infrastructure. Routing not only implies mere forwarding of data packets, but also refers to the choosing of best path for the data traffic based on certain metrics. In this paper, a routing algorithm namely Network Protocol-based QoS Routing is proposed and simulated using the network emulation tool, Mininet. The working of the algorithm is verified to correctness using the network protocol analyser Wireshark. The performance analysis is done by considering the QoS parameters.

This paper presents the application of gravitational search algorithm (GSA) and particle swarm optimization (PSO)-based approach along with multiple FACTS (flexible AC transmission system) devices for the economic operation of an interconnected power system under different loading condition. Two different types of FACTS devices such as static Var compensator (SVC) and thyristor-controlled series capacitor (TCSC) are used in this paper. The location of the FACTS devices is obtained by the reactive power flow in the transmission lines. The reactive loading of the system have been increased from the base value to 110 and 120% of base reactive loading. Finally, results have been compared between both the techniques in terms of minimization of active power loss and operating cost.

This paper works out with the comparison of dynamic responses of closed-loop proportional integral derivation (PID) and artificial neural network (ANN) in wind generator-fed Γ-ZSI-controlled PMSM drive system. Voltage-type Γ-ZSI is proposed for PMSM drive for simulation. Γ-ZSI can boost the input voltage significantly and the speed of the drive is controlled using V/f control method. The drive system is developed using blocks of MATLAB/Simulink and the results are presented in terms of its rise time, settling time, and steady-state error.

Power and performance have become significant metrics in the designing of multicore processors. Due to the ceasing of Moore’s law and Dennard scaling, reducing power budget without compromising the overall performance is considered as a predominant limiting factor in multicore architecture. Of late technological advances in power management techniques of the multicore system substantially balance the conflicting goals of low power, low cost, small area, and high performance. This paper aims at ascertaining more competent power management techniques for managing power consumption of multicore processor through investigations. We highlight the necessity of the power management techniques and survey several new approaches to focus their pros and cons. This article is intended to serve the researchers and architects of multicore processors in accumulating ideas about the power management techniques and to incorporate it in near future for more effective fabrications.

Direct Power Control (DPC) method for grid-connected Voltage Source Inverter is popular due to number of advantages such as elimination of inner current control loop, direct voltage vector selection, direct control of instantaneous active and reactive powers, and improved dynamic response. For voltage source inverter, effect of particular vector is to produce finite variations in instantaneous active and reactive power in a given sector. This paper investigates different switching patterns of DPC method for grid-connected Voltage Source Inverter. The performance of various switching tables such as Noguchi Table and Eloy-Garcia Table is studied. Based on the above tables, a modified table is proposed for 12-sector approach. The proposed method has less active as well as reactive power errors and can be applied to certain applications such as Active Power Filter and Unified Power Quality Conditioners. These methods are simulated and experimentally validated using RT-LAB + MATLAB Simulink®/Sim Power System tool.

Only limited support is given to the resource management of graphics processing unit (GPU) by operating system in commodity software. Operating system manages the graphics processing unit as a peripheral device and restricts the use of graphics processing unit in various applications. GPUs are not only intended for graphics applications, but could also be used in applications that require high performance thereby giving more attention for its development in recent years. A desirable part of GPU memory is left unutilized in systems that are not used for GPGPU(general purpose GPU) computing. GPU memory of such systems can be viewed as disaggregated memory and by developing proper interfaces, it could be used as swap device or cache memory because of its low access latency to improve the operating systems performance. The proposed system manages GPU as buffer cache in operating systems and ensures security using Twofish encryption algorithm.

In this chapter, we propose an effective method to remove uniform and heterogeneous fog from the image using dark channel prior (DCP) and guided filter. Variation in thickness of the fog present in variety of image has helped to analyze and upgrade the dark channel prior image quality. Fog acts as veil which obscures the original scene radiance. A significant amount of fog could be observed around the edges of the objects in fog free image which is carried from the foggy image during intermediate processing steps. Efficient selection of dark channel prior kernel parameters helps us to carry minimum fog from the input foggy image to fog free image which has certainly reduced the halo effect around the edges of the objects in the fog free image. Also the use of guided filter to preserve edges of the objects in image is a fast and cost effective approach toward generation of fog less image.

A reentry control system is proposed for an assured crew reentry vehicle (ACRV), where the control law is tuned using multi-objective genetic algorithm-based sliding mode controller. The controller designed guarantees the robustness properties with respect to parametric uncertainties and other disturbances. The system state remains in the neighborhood of a reference attitude and the control signal is close to a well-defined equivalent control. The amplitude of the sliding mode controller is tuned using an evolutionary optimization technique, i.e., Genetic algorithm. Multi-objective optimizer is used for the controller as it is to minimize the error in the Bank Angle (degree), Angle of Attack (degree), and Sideslip Angle (degree). The reference attitude is obtained in terms of the outputs given by the trajectory controller and the navigational system. A pulse width pulse frequency (PWPF) modulator is designed to modulate the attitude controller through the thrust torque developed. The simulation results show the effectiveness of the proposed method.

Multilabel classification is a classification technique in which each sample may be related with more than one class labels. This paper deals with a comparative study of mutual information (MI) technique and other methods in different classifiers. MI is a technique in filter approach type feature selection. It is a fine indicator, which measures the information or data that common between two variables, it audits and evaluates how one of the variables reduces the uncertainty of the other. We consider other two classifiers for the study; they are Naïve Bayesian (NB) and ID3. The experiments were done using MI and compared with the two classifiers for different benchmark data set from UCI repository flag, music, and yeast. The results were verified using evaluation measures. An accurate precision and recall value can be obtained in MI technique rather than using the classifiers NB and ID3.

This paper determines the incorporation of photovoltaic source and battery with a PMSM motor fed with a centrifugal pump based on conventional Direct Torque Control technique (DTC). For the boosting of solar output, interleaved DC-DC converter with fuzzy logic-based MPPT control is studied and in order to convert the DC output of converter for the required form of motor input, three-phase inverter has been taken. With the help of a bi-directional converter, regulation of battery is carried out which depends on load and source. Output results of motor as well as pump are obtained using MATLAB Simulink environment.

In the field of video surveillance, person re-ID (reidentification) is an assignment of identifying an individual caught by various cameras in the system at various place and time. With the developing security dangers, this undertaking has an incredible significance in observation out in the open spots like air terminal, railroad station, shopping buildings, and so forth. This assignment recognizes the individual of interest among the gathering of individuals caught by various cameras in the system put at various places and tracks the individual in various camera views. This undertaking faces numerous difficulties and has pulled in the scientists to this field to discover the answer for defeat the difficulties. In this paper, we have discussed about the latest research works that have been made to assault these difficulties.

There are several algorithms used for data clustering and as imprecision has become an inherent part of datasets now days, many such algorithms have been developed so far using fuzzy sets, rough sets, intuitionistic fuzzy sets, and their hybrid models. In order to increase the flexibility of conventional rough approximations, a probability based rough sets concept was introduced in the 90s namely decision theoretic rough sets (DTRS). Using this model Li et al. extended the conventional rough c-means. Euclidean distance has been used to measure the similarity among data. As has been observed the Euclidean distance has the property of separability. So, as a solution to that several Kernel distances are used in literature. In fact, we have selected three of the most popular kernels and developed an improved Kernelized rough c-means algorithm. We compare the results with the basic decision theoretic rough c-means. For the comparison we have used three datasets namely Iris, Wine and Glass. The three Kernel functions used are the Radial Basis, the Gaussian, and the hyperbolic tangent. The experimental analysis by using the measuring indices DB and D show improved results for the Kernelized means. We also present various graphs to showcase the clustered data.

AC drives fed by inverters find applications in a number of fields which are uncountable. Inverter performance can be enhanced by controlling the harmonics at various levels which can be at the input, output, and within the inverter. Hence, a hybrid method of controlling the harmonics on all the three levels can be put together in a inverter-fed permanent magnet system. A novel method of implementing a harmonic filter and multipulse converter on the input side and a space vector modulation technique used within the inverters will be a better solution in reducing the level of harmonics so as to improve the quality of output power fed to a permanent magnet synchronous motor drive system. This paper mainly focusses on improving the quality of power by various hybrid techniques which also involves the design of transformers using dimetrical connection for a multipulse system. The results are proved by reduced total harmonic distortion using PSIM and Matlab softwares.

The modern computing technology has a huge dependence on biometrics to ensure strong personal authentication. The mode of this work is to increase accuracy with less data storage and providing high security authentication system using multimodal biometrics. The proposed biometric system uses two modalities, palm print and palm vein. The preprocessing steps begin with image acquisition of palm print and palm vein images using visible and infrared radiations, respectively. From the acquired image, region of interest (ROI) is extracted. The extracted information is encrypted using encryption algorithms. By this method of encryption, after ROI extraction, the storage of data consumes less memory and also provides faster access to the information. The encrypted data of both modalities are fused using advanced biohashing algorithm. At the verification stage, the image acquired is subjected to ROI extraction, encryption and biohashing procedures. The biohash code is matched with the information in database using matching algorithms, providing fast and accurate output. This approach will be feasible and very effective in biometric field.

Real-world problems usually have bounded search space and, therefore, the performance of optimization algorithms on them is also related to choose a proper boundary constraint handling methods. There are several classical approaches in the literature to handle the bounds. Usually, optimization algorithm use does not pay attentions to choose a proper boundary constraint handling scheme. In this study different boundary constraint handling schemes such as evolutionary scheme and classical schemes (including reflecting scheme, absorbing scheme, toroidal scheme, and random scheme) are evaluated on a recent evolutionary algorithm called interior search algorithm (ISA). In this paper, all the boundary constraint handling approaches have been adopted to ISA to solve a wide set of global numerical benchmark problems. The conclusions are made based on statistical results which clearly show the importance of different boundary constraint handlings in the searching process. The results obtained using the evolutionary boundary constraint handling scheme are better than the ones obtained by the other well-known approaches and it seems this scheme is suitable for a wider range of evolutionary optimization problems with very good convergence rate.

In recent days, automatic emotion detection is a field of interest and is used in fields such as e-learning, robotic applications, human–computer interaction (HCI), surveillance, ATM monitoring, mood-based playlists/YouTube videos, psychological studies, medical fields like supporting blind and dumb people, for treating autism in children, entertainment, animation, etc., The proposed work describes detection of human emotions from a real-time video or image with the help of classification technique. The major part of human communication constitutes of facial expression, which is around 55% of the total communicated information. The basic facial expressions that are considered by the psychologists are: happiness, sadness, anger, fear, surprise, disgust, and neutral. The proposed work aims to classify a given video into one of the above emotions using efficient facial features extraction techniques and SVM classifier. The author’s contribution is to increase the efficiency in emotion recognition by implementing the above mentioned superior feature extraction and classification methods.

Parse tree generation from supervised sentence is crucial as well as essential task necessary for natural language processing. Syntax analyzers play a vital role in natural language processing. Prevailing system for parse tree generation is subject to compromise on ambiguous situation. Our approach is to generate a parse tree using Hidden Markov bigram model from tagged sentence results in improved parse tree generation. The user will be given the sentence to be parsed after part of speech tagging. If the sentence is syntactically correct, the parser will generate a correct parse trace.

The important aspect in research is to publish data by conserving one’s privacy. Enormous techniques have been proposed to tackle this issue. The main concept is to concentrate on how cogently one protects individual privacy in publishing data without the exact attribute missing in the research. As per our knowledge is concerned, there raised a great scope for research on vertically partitioned distributed databases. This paper mainly concentrated on privacy which is addressed on vertically partitioned distributed data. It is the prime responsibility of the publisher to see that one’s personal information is protected. In order to maintain such sustainability, organizations like hospitals, government agencies, etc., store information at multiple sites by vertically portioning attribute with a solution to integrate all these attributes without violation and distraction of any meaning. To implement this, good models are required to integrate information from multiple sites for publication or for research, and models proposed also require maintaining privacy. In this paper, modernistic models ring and hierarchical model are proposed to preserve privacy for vertically portioning distributed data.

In this paper, we propose a novel one-class classification approach for text document classification using One-Class Support Vector Machine (OCSVM) and Latent Semantic Indexing (LSI) in tandem. We first apply t-statistic-based feature selection on the text corpus. Then, we apply OCSVM on the rows corresponding to the negative class of the document-term matrix of a collection of text documents and extract the Support Vectors (SV). Then, in the test phase, we employ LSI on the query documents from the positive class to compare them with the SVs extracted from the negative class and match score is computed using the cosine similarity measure. Then, based on a prespecified threshold for the match score, we classify the positive category of the text corpus. Use of SV for comparison reduces the computational load, which is the main contribution of the paper. We demonstrated the effectiveness of our approach on the datasets pertaining to Phishing, and sentiment analysis in a bank.

System identification (SI) is referred to as the procedure of building mathematical models for the dynamic systems using the measured data. Several modeling methods and types of models were studied by classifying SI in different ways, such as (1) black box, gray box, and white box; (2) parametric and non-parametric; and (3) linear SI, nonlinear SI, and evolutionary SI. A study of the literature also reveals that extensive focus has been paid to computational intelligence methods for modeling the output variables of the systems because of their ability to formulate the models based only on data obtained from the system. It was also learned that by embedding the features of several methods from different fields of SI into a given method, it is possible to improve its generalization ability. Popular variants of genetic programming such as multi-gene genetic programming is suggested as an alternative approach with its four shortcomings discussed as future aspects in paving way for evolutionary system identification.

In this paper, the problems posed during the integration of small hydro power generating units with the existing power grid are smoothly handled. The proposed scheme considers a pulse width modulated (PWM) converter and an improved D-STATCOM. The results obtained from MATLAB-Simulink-based simulation successfully validate the smooth interfacing of small hydro power units with the grid.

In the recent past , there has been increasing usage of machine learning algorithms for classifying the data for various real-world applications such as weather prediction, medical diagnosis, network intrusion detection, software fault detection, etc. The instance-based learning algorithms play a major role in the classification process since they do not learn the data until the need of the developing the classification model. Therefore, these learning algorithms are called as lazy learning algorithms and implemented in various applications. However, there is a pressing need among the researchers to analyze the performance of various types of the instance-based classifier. Therefore, this chapter presents a pragmatic investigation on performance of the instance-based learning algorithms. In order to conduct this analysis, different instance-based classifier namely instance-based one (IB1), instance-based K (IBK), Kstar, lazy learning of Bayesian rules (LBR), locally weighted learning (LWL) are adopted. The performance of these algorithms is evaluated in terms of sensitivity, specificity, and accuracy on various real-world datasets.

In recent past, the data are generated massively from medical sector due to the advancements and growth of technology leading to high-dimensional and massive data. Handling the medical data is crucial since it contains some sensitive data of the individuals. If the sensitive data is revealed to the adversaries or others then that may be vulnerable to attack. Hiding the huge volume of data is practically difficult task among the researchers. Therefore, in real life only the sensitive data are hidden from the huge volume of data for providing security since hiding the entire data is costlier. Therefore, the sensitive attributes must be identified from the huge volume of data in order to hide them for preserving the privacy. In order to identify the sensitive data to hide them for providing security, reducing the computational and transmission cost in secured data transmission, this chapter presents a pragmatic approach to identify the sensitive attributes for preserving privacy. This proposed method is tested on the various real-world datasets with different classifiers and also the results are presented.

The objective of this paper is to assist the driver during nighttime. Many papers were published which concerned with detection of vehicle during daytime. We propose a method which detects the vehicle during nighttime since it is problematic for human to analyze the shape of vehicles in nighttime due the limits of human vision. This project endeavors to implementation of vehicle detection based on vehicle taillight and headlight using technique of blobs detection by Center Surround Extremas (CenSurE). Blobs are the bright areas of the pixels of headlight and taillight. First stage is to extract blobs from region of interest by applying multiple Laplacian of Gaussian (LoG) operator which derive the response by manipulating variance between surrounding of blob and luminance of blob which are grabbed on road scene images. Compared to the automatic thresholding technique, Laplacian of Gaussian operator provides more robustness and adaptability to work under different illuminated conditions. Then vehicle lights which are extracted from the first stage are clustered based on the connected—component analysis procedure, to confirm that it is vehicle or not. If vehicle is detected, then tracking of vehicle is done on the basis of the connected components using bounding box and different tracking parameters.

This paper gives an idea of how to utilize the processor-in-the-loop (PIL) and reduce the development time. The rapid advancement of the computer system and user interactive software empowers the lot of simulation software which are available for technical applications. In this paper, the simple mathematical formulations using simulation block sets are formed. The simulated mathematical formulations are transformed into embedded C code. The generated code is launched into the embedded processor. The communication link between host PC and target processor is formed to view real-time variation where part of the simulation runs in PC and part in processor. The simulation results are compared with launched program executed by the processor. This approach unified the designer with a program and reduced the core technical manpower effort, simultaneously increases more functionality to test the program in processor. The simulation platform is used as MATLAB and Simulink, and STM32F4 Discovery board is used as a target to launch the program in processor.

From the view point of revealing different preferences among the individuals, the assessment of individual typology, i.e., chronotype, is increasing largely in the recent past. Chronotype, recognized as a human attribute, is usually studied by self-reported instruments designed to find individual time of preference for daily activities in an easy manner. One of the criticisms of using these self-reported instruments includes the fact that total scores may not always reflect the actual chronotype of an individual. On the other hand, linguistic terms are used to address some of the items of these self-reported questionnaires. In this paper, an interval valued type-2 Mamdani fuzzy inference system has been proposed to assess chronotype of an individual. An illustrative case example is discussed for validation of the proposed model in the field chronobiology.

Biometrics plays a major role in image identification field in the security environment, mainly in image recognition and detection. The pose variation in image leads to mismatch with its original image and it is one of the major frequently occurring problems. This proposed work explains new efficient techniques to recognize people with the help of their biometric features like Face and Ear. The main objective of this work is to recognize a person’s face, ear image correctly, with respect to his/her pose varied images. Novel techniques named as, Max–Min Comparison, Red matrix Eigen vector recognition, are proposed for recognition processes. These techniques are tested on standard database images and its results are analyzed using mismatching and runtime parameters, and its accuracy yields high recognition rate of 99%.

Cloud-based storage is becoming a very good method for sharing any type of data in upcoming days. This method can reduce the burden of storing and retrieving data in local machines. But there exist many issues. The vendor itself can attack the user’s data and it can even delete or hide the data; if the user did not access the data for a long period of time. It also faces problems for synchronizing and uploading files and the vast range of internal and external threats for data integrity. Another problem is based on data security in cloud storage, where it does not have a proper encryption technique. Storing data in the cloud is financially good for long-term large scale data storage, although it does not fully offer guarantee on data integrity and availability. In this paper we are focusing the integrity of the data using a technique that divides the data into number of pieces and encrypting it by newly introduced method. Through this technique cloud ensures more security and integrity.

In last decades, electronics and telecommunications have known unexpected development; the number of nonlinear loads has increased, along with this the disturbances in power quality also raised. Poor quality in power causes more losses and will cause malfunction, leading to replacement of particular connected load. This paper presents the preliminary results of Power quality survey done in academic and hostel buildings of VIT University. The power quality survey is necessary to record all events over an enough long time interval. Here we collected various parameter readings from all electrical feeders of for different buildings. This paper analyzes the complexities as well as consequences of power quality disruptions. In order to reduce the huge amount of data by recording and analyzing several electrical parameters over certain duration of time, some recording limits as per IEC standards are to be set. For particular feeders where limits exceed the set values, various strategies to enhance the quality of power and minimize the losses are implemented.

Resource provisioning is greatly required to expand the performance in cloud. The hardware technique used for resource allocation methods can be central processing unit (CPU) scaling, which is the frequency of physical cores. The software technique used can be horizontal and vertical elasticity. The resource management affects the evaluation of a system by performance, functionality, and cost. The resource management in cloud environment should have enormous policies and decisions for adhering various objective optimization. Efficient resource management is the process of allocating the resources and handling the workload variations effectively. The computing resources have to be handled efficiently among the users of virtual machines. Service level agreement (SLA) is defined as the contract made between the providers and the customers of cloud for guaranteeing the quality of service (QoS) issues. This paper gives various policies defined in cloud computing related to SLA issues.

This paper proposes a look-up table method for tracing the maximum power generated in the wind energy conversion system (WECS). The look-up table is used to develop the corresponding change in the duty ratio with the change in the current. Self-excited induction generator (SEIG) is the perfect choice for generating power from the wind. A diode bridge rectifier (DBR) is used to convert renewable energy into electrical output and a dc–dc converter maintains the grid voltage. This method of tracking the maximum power from the wind is determined using the dc grid current. The dc grid current is the control variable that tracks the maximum power in the system and it is not dependent on the machine and wind parameters. Simplicity in circuit and control algorithm is the key advantage for supplying maximum power from the WECS to the dc microgrid.

Molodtsov introduced soft set theory in 1999 to handle uncertainty. It has been found that hybrid models are more useful than that of individual components. Yang et al. introduced the concept of interval valued fuzzy soft set (IVFSS) by combining the interval valued fuzzy sets (IVFS) and soft set model. In this paper we extend it by introducing interval valued hesitant fuzzy soft sets (IVHFSS) through the membership function approach introduced by Tripathy et al. in 2015. To illustrate the application of the new model, we provide a decision making algorithm and use it in stock market analysis,

Multilevel Inverter extensively used in medium and high voltage applications, because of their lower harmonic distortion, electromagnetic interference, and higher DC bus utilization. Still some demerits presence in MLIs is more number of power semiconductor devices. Due to this, complexity arises in switching strategies and voltage imbalance occurs across each level. In addition to the above, harmonic spectrum of MLIs depends on the switching strategy. In order to improve the performance of MLIs, novel hybrid switching strategies were developed and implemented in seven-level reduced switch inverter. This paper analysis the potential of above strategies interns of performance parameters like THD, V_RMS, and DF are compared with sub-harmonic pulse width modulation strategy. Simulations were carried out using MATLAB/SIMULINK.

Social network sites (SNS) are open platform that allows users to create an account in public sites. Social network site contains more active users, who share the posts, messages, or valuable information in real time. Even though social network provides timely updating still it is hard, because the current event mentions are overwhelmed by other mentions, which are totally irrelevant with the current affairs. The proposed methods mainly concentrate on text mentions in social networks and avoid the current affairs of the social aspect. Here two main methodologies are proposed that will improve the web content mining in social network sites. GBMAD (Geographically based mention anomaly detection), is a one of the best methods to manage dynamic links (i.e., Mentions: shares, likes, re-tweets, comments, etc.). GBMAD detects current events over the particular geographical location and it determines the importance of mentions over the crowd. The main advantage is it dynamically works in the real-time data rather than defined data. The next method is connected dominating set is used for identifying the closely related structure by using graph theory techniques. In the proposed work, the anomaly users are identified using Mention anomaly detection. In the future work, the implementation of Sybil attacks in the network will be identified and prevent from unsolicited friend requests in social networks using graph-based theory.

The main objective of this paper is to design a controller for control of an Induction motor. In this paper, we have proposed v/f control of induction motor using artificial neural network, the network is trained using back propagation algorithm and Levenberg–Marquardt learning is used faster computation. The main approach is to keep voltage and frequency ratio constant to obtain constant flux over the entire range of operation and thus to have precise control of the machine. The effectiveness of the controller is demonstrated using MATLAB/Simulink simulation.

The aim of this research is to design and analyze a fireproof firefighting robot that can enter into fire environment and navigate itself through the fire and send information about the fire behavior. This robot would help the fire rescue team to better understand the fire behavior and trapped person location and thus would be a critical advantage in term of time saving and rescue teams risk for their own life. In this paper, an image processing system and communication architecture for firefighting robot based on GSM technology and microcontroller is designed. Camera connected to microcontroller using serial cable will capture the image data and store it on a storage device. The processed image are sending to the predefined mobile number using GPRS technology. The encoder is used to improve the efficiency of compressed image. MATLAB software is used for the image processing which uses Fuzzy Coded Means to complete this process.

Due to low-cost and high-reliability, induction motors have discovered significant applications for variable speed applications. The vector control method is used to achieve better dynamic response and use of induction motor for a wide range of speed variations. However, it uses shaft encoder for rotor speed estimation that suffers from added cost, measurement noise, and maintenance associated with them. To overcome these shortfalls, speed is estimated with the help of measured voltages and currents in sensorless vector-controlled drive. The rotor speed estimation method is based on a direct synthesis of induction motor state equations. Simulation of the drive is carried out using PSIM software. In order to validate the method, laboratory prototype is developed using DSP TMS320C2811 for 10 H.P. and 15 H.P. squirrel cage induction motor.

Network with large number of sensor nodes distributed spatially is termed as Wireless Sensor Network (WSN). The tiny devices called as sensor nodes are cheap, consume less power, and the capabilities of computation is limited. The most challenging issue for WSN is protecting the network from misbehavior of intruders or adversaries. One of the major techniques used to prevent from any type of attack in the sensor network is artificial intelligence system (AIS). Intrusion Detection System (IDS) is considered to be the second line of defense, as sensor nodes are first defense line. WSNs are highly vulnerable to intrusions and different types of attacks. In most critical applications of WSN, the human intervention or some physical devices are not sufficient for protecting the network from strong adversaries and attacks. Thus, artificial intelligence techniques are used for intrusion detection and prevention of sensor networks.

With the growing demand of multimedia applications, quality of service (QoS) assured that multicast routing is an important issue in wireless mesh networks. Finding a multicast tree which satisfies the multiple constraints is a NP-Complete problem. In this paper, we propose a discrete particle swarm optimization (DPSO) approach for finding the minimum tree cost from a given source to a set of destination with delay and bandwidth constraint. The concept of relative position base indexing (RPI) is used to convert a continuous space to discrete space for multicast routing. The simulation is carried out in NS-2.26 and the comparison is made with respect to tree cost under two scenarios, namely varying node mobility and increasing number of nodes. The results demonstrate that DPSO has better speed, performance, and efficiency than multicast routing based on genetic algorithm.