Murugan B S
@kalasalingam.ac.in
Professor
Kalasalingam Academy of Research and Education
received the Undergraduate Degree (B.Tech) in Information Technology from Anna University, the Post Graduate degree (M.Tech) in Information Technology from SRM University and Ph.D in Information Technology (Cloud Computing) from Kalasalingam University in June 2017. He has more than 30 publications in National, International Conference and International Journal proceedings. He has 13+ years of teaching experience. His areas of interest include Cloud Computing, Data Science, Operating Systems, and DBMS etc. He is currently working as Deputy Registrar and Senior Associate Professor in the Department of Computer Science and Engineeringat Kalasalingam Academy of Research and Education, Krishnankoil, Tamil nadu, India.
EDUCATION
PhD in Cloud Computing
M.Tech in Information Technology
B.Tech in Information Technology
RESEARCH INTERESTS
Cloud Computing, Data Science, Artificial Intelligence
Scopus Publications
Scholar Citations
Scholar h-index
Scholar i10-index
Scopus Publications
Sethu Vinayaga Vadivu, Palanigurupackiam Nagaraj, and Bagavathi Ammai Shanmugam Murugan
Informa UK Limited
Ramana Rajendran and B. S. Murugan
Springer Science and Business Media LLC
Shanmugapriya. S, Dhanasekaran. S, and Murugan. B. S
IEEE
Citrus is widely cultivated and consumed fruit across the global. It gains its popularity from its nutritional benefits. Citrus fruit has gained its international trading due to freshly consumed, processed products, oil production and other by-products. Citrus production is cut short mainly due its diseases. Identifying the diseases at early stage and stepping forward to cure them is vital role in improving the production. Automated disease identification cuts down the manual effort and also improves the efficiency in identification. With various degrees of accuracy, both machine learning and deep learning algorithms have been successful in identifying citrus leaf and fruit diseases. This research study proposes a novel disease classification method using CNN and also provides a comparative study with models like SVM, k-means clustering, ANN and CNN for disease identification.
D. Balakrishnan, T. Dhiliphan Rajkumar, S. Dhanasekaran, and B. S. Murugan
Springer Science and Business Media LLC
R. Raja Subramanian, P. Vijaya Karthick, S. Dhanasekaran, R. Raja Sudharsan, S. Hariharasitaraman, S. Rajasekaran, and B. S. Murugan
Springer Nature Singapore
N. Siva Balan and B. S. Murugan
Springer Science and Business Media LLC
Ramana Rajendran and B. S. Murugan
IEEE
The lack of appropriate shape representation makes it complex to recognize the 3D shapes accurately and it is a hot topic in the field of Computer Vision (CV). This paper presents a Pelican optimized Convolutional Neural Network (CNN)-Bidirectional Long Short Term Memory (BiLSTM) to recognize the different objects in a particular scene. The CNN-BilSTM architecture is formed by placing two BiLSTM architectures below the CNN network and integrating the outputs via a fully connected layer. The pelican optimization algorithm is mainly incorporated to optimize the different hyperparameters associated with the CNN-BiLSTM architecture such as number of layers, batch size, number of layers, dropout, etc. The experiments are conducted using the ScanNet dataset which comprises both 2D and 3D data along with the labeled voxels. The proposed methodology offers improved results when compared with the existing techniques in terms of confusion matrix, accuracy, precision, and recall.
Sivabalan N. and Murugan B. S.
IGI Global
In recent days, the usage of cloud computing in wireless networks offers more advantages to the users by storing resources with less complexity and ease to control. Data security is considered a critical aspect in a cloud computing environment due to the sensitive and confidential information of users stored in IOT. So, this paper introduces a Lightweight and Privacy-Preserving Two-Factor Authentication (TFA) with S-box based Flipping Module (SBFM) to provide data security for a user. The proposed scheme uses Unclonable Function Key (UFK) to provide a better solution for highly-secured cloud computing. Moreover, Reconfigurable Unpredictable Response Value (RURV) helps to generate the different response values for every clock cycle in IoT. Finally, Spartan 6 Field Programmable Gate Array (FPGA) performances of the proposed TFA-RURV-IoT are compared to existing TFA-URV-IoT protocols, whereas the simulation results show that proposed TFA-RURV-IoT achieves better results in terms of LUT, slices and flip flops.
Prathima. Y and B. S. Murugan
IEEE
Data clustering is a crucial phase in data analysis, widely concentrated by data mining communities. Many previous algorithms based on data clustering are related to the endless models that look for sparsity and higher dimensional issues and try to avoid the sequence of information and the data structural data. The recurrent and convolutional neural networks work on deep learning-based models concerning the data as sequences. Yet, the explanation of outcomes and the supervised signals are lacking. The adaptive data clustering model (ADCM) technique is proposed in this system to incorporate the pre-trained data encoders into data clustering tasks. This model depends on the representation of a sequence that breaks the dependencies on the supervision. The proposed system provides experimental outcomes that perform better than the traditional data clustering algorithm and the modern data model, pre-trained on the complete datasets. Additionally, the clustering result explains the significant understanding of the deep learning technique principles. The clustering approach proposes the description model that assists the users in understanding the quality and meaning of the outcome of the clustering process.
N. Siva Balan and B.S. Murugan
IEEE
On the digital side, data transfer is constantly subject to attack. A cypher strength study is an important part of any corporate or academic security assessment. A strong encryption mechanism is required for data security. For secure information transmission, System-On-Chips (SoCs) for Internet of things (IoT) applications require hardware-based coordinated arbitrary number generators. They don’t have a confined hardware and power spending plan, intends to the use of particular Twisted Edwards Curve (TEC). A symmetric encryption with incorporated TEC is proposed in this paper. A regular 40 nm CMOS innovation is utilized to accomplish the proposed plan. The aftereffects of the post-design recreation uncover that it gives great irregularity at a modest energy-per-bit cost. Moreover, the circuit finished all NIST assessments with no post-handling. When contrasted with the conventional encryption, it has an extremely low region increment of just 0.14 percent. As a proof of idea, a FPGA execution is likewise shown, which checks the simulated results. To show the double utilization of the proposed TEC, the Advanced Encryption Standard (AES) key extension method is also evolved. The AES technique depends solely on the substitute-permute network plan rule, and it functions admired in both programming and hardware. The procedure introduced here utilizes a solitary indistinguishable mystery key for encryption and disentangling. It can't be utilized in public sector or private, business, or different projects. The AES calculation is executed on two FPGAs in this research, and it was found that the Spartan-6 FPGA conveys better throughput and lower time delay than FPGA-based IoT gadgets.
S. Dhanasekaran, K. Pradeep Mohan Kumar, A. Sivanesh Kumar, R. Jeya, S. Rajasekaran, B.S. Murugan, and R. Rajasubramanian
Elsevier
M. Sangeetha, P. Vijayakarthik, S. Dhanasekaran, and B.S. Murugan
Elsevier BV
Bindu Madavi
Institute of Advanced Scientific Research
M. Mohammed Thaha, K. Pradeep Mohan Kumar, B. S. Murugan, S. Dhanasekeran, P. Vijayakarthick, and A. Senthil Selvi
Springer Science and Business Media LLC
G. Sachin, J. Mohammed Ahamed, K. Nagajothi, M. Rana, and B. S. Murugan
Copernicus GmbH
Abstract. Crop Simulation Models (CSM) simulate the growth, development, and yield of crops using various inputs such as soil water, carbon and nitrogen processes, and management practices. DSSAT (Decision Support System for Agrotechnology Transfer) is a software program that comprises dynamic crop growth simulation models for over 42 crops. It incorporates modules for crop, soil, and weather to simulate long-term outcomes of crop management strategies. DSSAT-CSM requires various data for model operation. This includes data on the site where the model is to be operated, on the daily weather during the growth cycle, on the characteristics of the soil at the beginning of the growing cycle or crop sequence, and on the management of the crop. Acquisition of the data and providing the data to the DSSAT model is tedious and time-consuming as each individual value has to be manually entered. Additionally, crop simulation models can only be run for specific points and not for entire locations. Sometimes site-specific data especially weather data cannot be obtained. The output thus produced is difficult to analyze spatially at a large scale. The main purpose of this paper is to take the required dataset directly from spatial data. This is done by dividing locations into grids and taking the data from each grid. Python scripts are then used to convert this data into crop model format which is then run through DSSAT on an individual basis. The output thus obtained is be entered back into their respective grids as spatial data.
B.S. Murugan, Mohamed Elhoseny, K. Shankar, and J. Uthayakumar
Elsevier BV
S. Dhanasekaran, R. Sundarrajan, B. S. Murugan, S. Kalaivani, and V. Vasudevan
IEEE
The Clustering methods have been greatly adopted in various real world data analysis applications, such as customer behavior analysis, medical data analysis, digital forensics, etc. In existing system, MR-mafia subspace clustering algorithm becomes inefficient as well as ineffective because the data size are continuously increasing, and data blocks are overlaying. Big Data environment inherits several knowledge and we extracts the necessary knowledge and K-means clustering algorithm is being designed. This paper focused on K – mean clustering algorithm based on improved map reduce techniques. The algorithm takes advantages to avoid unnecessary input and output data and also used to optimize data storage and also to achieve the out sourcing of data privacy. We have using a medical datasets of this project, and Enhanced map reduce based K – means clustering algorithm have been proposed which work effectively done and that can be outsourced to cloud server.
B. S. Murugan, V. Vasudevan, and B. Ganeshpandi
IEEE
In cloud computing resource allocation should be elastic, within the sense that it must have modification accurate and quickly based on the demand. In cloud computing, Virtual machine allocates the resource for user's needs. Some time workload of service, increase rapidly, the existing approaches solve aggressive resource provisioning tasks using SPRNT, but still some challenges occur in VM allocation. Such as, it may fail when faced with rapidly increasing workload, and the user can't commit with a virtual machine that which one is adapted to the task, then increases in adaptation time and memory access time. In this paper, we proposed agent based resource allocation for intelligent scheduling system. Agent based resource allocation should help users to verify the most effective framework for every individual. It provides a high level of flexibility, balance workload, reduce adaptation time and memory access time. The evaluation result of agent based resource allocation to achieve up to 4.3x speed up in adaptation time.