Proceedings of the 7th International Conference on Communications and Cyber Physical Engineering

Table of Contents

1. Privilege Escalation Attack Detection and Mitigation in Cloud Using Machine Learning

   T. Kavitha, E. Poojitha, M. Bhavani, K. Sravanthi

   Abstract

   With the expansion of technology, the frequency and complexity of attacks are also increasing, and cybersecurity competition is also strong. While centralized cloud computing has changed for businesses, it has faced problems when using decentralized security systems [1]. Due to the large amount of inconsistent and weak data exchange between businesses and cloud service providers, this can lead to data leakage. Insider threats, especially from malicious people with significant access, pose a significant risk. To solve this problem, a machine learning-based application for detecting and classifying insider threats focuses on suspicious situations that indicate increased privilege [2].Integrated learning technology is used to improve prediction performance by combining multiple models. Although previous studies have examined the vulnerability and vulnerability in the network, most of them did not provide information about the attack and its distribution [3]. This study uses a data set from the CERT dataset and uses four machine learning algorithms: Random Forest, Adaboost, XGBoost, and LightGBM. Results show that LightGBM outperforms the other algorithms, demonstrating its efficacy in identifying and classifying insider attacks.

2. IOT Based Smart Home Using Alexa

   Y. Pavan Kumar Reddy, K. Shankar, Y. Sunanda, M. Ravi Kishore, K. Raj Mohan

   Abstract

   Home monitoring is steadily replacing omnipresence because of its many advantages. Given the current state of affairs and advancements, the house motorization structure can be appropriately controlled by SMS, messaging, or other apps. Regardless, industry players and conventional experts have recently shown a great deal of interest in the Internet of Things (IoT). The advent of devices such as the Samsung Sharp Things, Google Home, and Amazon Resonation, among others, has brought particular attention to the Splendid Home industry. Creative, incisive, best-in-class plans are produced by an industry's augmentation. The objective of this work is to design a conservative, dependable setup that may be widely applied in households with little to no experience. Amazon Cloud, Amazon Discourse Administrations, and Amazon Reverberation power our framework. The Arduino ESP32 is the hardware that is utilized to add wise characteristics to unappealing houses. Understand the different components of our system and demonstrate how successfully it tries to turn on and side-street our devices. Voice commands should be used to operate any devices or equipment that you have at home. On the other hand, because of this, automated homes will not match conventional residences as well. “Use IoT and Alexa to turn your house into a smart heaven. You can effortlessly use your voice to manage the lighting, thermostat, and security. Take pleasure in a house that adapts to your requirements, providing convenience and energy savings at your fingertips.”.

3. Advancements in Image Dehazing: Enhanced Techniques with Non-uniform Atmospheric Light, Improved Dark Channel Prior, and Combined Window Filter

   K. Shankar, M. Ravi Kishore, Y. Pavan Kumar Reddy, S. Gnanamurugan, Y. Sunanda, R. Mahesh Kumar

   Abstract

   Haze is when fine particles like smoke, dust, and water droplets in the air make it difficult to see clearly and distort light. This can affect the sharpness, contrast, and colors in pictures. To fix this, we've come up with a new way to remove haze from both remote sensing and regular images. We start by breaking the image into patches (small groups of pixels) and figuring out the atmospheric light for each patch. Then, we smooth it using a guided filter. After that, we calculate the transmission map using an improved dark channel prior, which includes a dark channel prior and a combined window filter to reduce unwanted effects. We tested this method on images from the ground and remote sensing, and it worked well. So, our method can restore images by dealing with uneven atmospheric light, minimizing unwanted effects, and producing clear, detailed, and natural-looking results.

4. Bird Genus Classification and Identification Using Deep Learning Approach

   P. Prashanth Kumar, V. Supraja, K. Pranathi, E. Naveena

   Abstract

   These days, it's hard to classify and forecast which bird species will be around since some of these species are so scarce. Birds in diverse settings naturally seem different to humans in terms of size, shape, color, and angle. Beyond that, visual cues are more useful than auditory cues for identifying bird species. The fact that people can identify the birds in the photos is also easier to grasp. Therefore, the Caltech-UCSD Birds 200 [CUB-200-2011] dataset is used for both training and testing purposes by this technique. To create an autograph using tensor flow, a picture is gray scaled and then fed into a deep convolutional neural network (DCNN) algorithm, which generates many comparison nodes. The testing dataset is used to compare these various nodes, and a score sheet is then generated. After looking at the score sheet, it may use the highest score to predict which bird species are needed.

5. Leveraging RACDE-Net for Advanced Channel Estimation in Orthogonal Frequency-Division Multiplexing

   Thanneeru Durga Rao, T. J. Nagalakshmi

   Abstract

   This study aims to improve symbol detection accuracy in the field of Orthogonal Frequency-Division Multiplexing (OFDM) by introducing the Recurrent Attention Channel Detection and Estimation Network (RACDE-Net). RACDE-Net is a deep learning model that combines the capabilities of recurrent neural networks with attention mechanisms. RACDE-Net is specifically designed for the complexities of OFDM. It utilizes temporal dependencies and focuses on important data sequences to accurately identify and estimate channel conditions. Compared to traditional Least Squares and Minimum Mean Square Error methods in a simulation framework that includes Wideband Rician Fading and Additive White Gaussian Noise, the model’s effectiveness is measured by its Symbol Error Rates (SER), which are significantly lower as the signal-to-noise ratios increase. This innovative method not only exceeds current methods but also establishes a new benchmark for improving the dependability and effectiveness of OFDM systems using deep learning.

6. Machine Learning for Image and Video Analysis: Recent Advances and Challenges

   D. Rahul, Venkataswamy Gutam, Kathoju Navya, A. Shiva Prasad, RadhaKrishna Karne, Kallem Niranjan Reddy

   Abstract

   These last few years have seen major advancements in the area of image and video analysis brought about by machine learning. These advancements have made it possible to construct models that are both highly accurate and efficient for a broad variety of applications. The purpose of this study is to present a complete review of these improvements, with a particular emphasis on important approaches such as convolutional neural networks (CNNs), recurrent neural networks (RNNs), and generative adversarial networks (GANs). We investigate its applicability in a variety of tasks, including image classification, object identification, semantic segmentation, and action recognition in films. Even though many breakthroughs have been made, there are still a number of issues that need to be addressed. Many challenges include data quality and annotation, computational complexity, resilience to fluctuations, and ethical considerations. In addition, we talk about the most recent developments in hardware acceleration, self-supervised learning, and explainable artificial intelligence. The purpose of this article is to give insights into the present status of machine learning in image and video analysis as well as future directions by emphasising both the successes and the continuing obstacles that are currently being faced.

7. Distribution and Transmission Line Security Monitoring and Protection System

   K. Shankar, Y. Sunanda, K. MadhaviPriya, Y. Pavan Kumar Reddy, M. Ravikishore, K. Chandrahasa Reddy

   Abstract

   An electrical system that is becoming more complicated in all areas, including distribution, transmission, monitoring, and load systems, is the Distribution and Transmission Line Security Monitoring and Protection System. In order to quickly respond to any thefts, this system combines real-time data processing with sophisticated monitoring tools to identify irregularities. A high voltage transmission is needed. Electricity is transmitted by high voltage transmission. Electricity is crucial for comprehending and tracking the system’s behavior.

8. Agricultural Innovation Through AI: Implementing an Enhanced XGBS Model for Smart Crop Recommendations

   G. Thapaswini, M. Gunasekaran

   Abstract

   The research presents the development and validation of an Enhanced XGBoost-Support Vector Machine (XGBS) Model for crop recommendation in Karnataka based on soil and environmental parameters. By obtaining a dataset encompassing variables such as soil nutrients (N, P, K), temperature, humidity, pH, and rainfall, the study embarked on pre-processing, including median imputation for missing values and outlier correction through IQR, followed by feature normalization. Exploratory data analysis revealed distinct phosphorus, potassium, and rainfall requirements across various crops, with relational plots indicating diverse nutrient tolerances. The Enhanced XGBS Model, integrating the strengths of XGBoost and SVM, demonstrated superior predictive performance, with an overall accuracy of 99.31%, outshining conventional models in precision, recall, and F1-score metrics across all crops. This ensemble model serves as a robust framework, offering significant improvements over traditional methods and showcasing potential for precise agricultural decision-making, optimized resource utilization, and enhanced crop yield outcomes.

9. An Integrated Framework for Early Detection of Diabetic Cardiomyopathy Using CARDIO-VGTS-Net Model

   Punganuru Swathi, M. Gunasekaran

   Abstract

   Diabetic Cardiomyopathy (DCM) is a serious form of cardiovascular disease in diabetic patients, characterized by structural and functional changes in the heart that frequently lead to heart failure. Coronary artery plaque formation is an important aspect of DCM, which exacerbates the condition by restricting blood flow and adding to the complexity of diabetes-related heart disease. To address this issue, we present CARDIO-VGTS-Net, a novel framework that combines the deep learning power of the VGG-19 Convolutional Neural Network with the nuanced texture analysis capabilities of the Gray-Level Co-occurrence Matrix (GLCM). This integrated approach efficiently extracts and combines intricate image features, allowing for the precise detection and characterization of plaque formations and other DCM-related abnormalities in microscopic images of blood vessels. Our methodology, by providing a more nuanced understanding and accurate diagnosis of DCM, promises to improve early detection and inform targeted treatment strategies, representing a significant advancement in the management of cardiovascular complications in diabetic patients.

10. Improving Network Reliability and Energy Efficiency in WSNs Through ARC-Net Trust Management

    Bojja Nagabhushana Babu, M. Gunasekaran

    Abstract

    Trust management plays a crucial role in wireless sensor networks to ensure secure and dependable communication between nodes. This study presents a new ARC-Net model that utilizes adaptive penalty coefficients to dynamically modify trust levels based on node behavior. The ARC-Net model combines direct, indirect, and energy trust assessments, along with advanced optimization and machine learning methods, to effectively handle network energy efficiency and trust. The model’s efficacy is emphasized by its capacity to preserve network integrity until the initial node failure occurs at 8100 rounds, showcasing the system’s resilience and its potential to improve network longevity.

11. Automatic Driverless Tram

    N. Malla Reddy, P. Nikitha, D. Nitya Spoorthy, P. Jayanthi, R. Vaishnavi, A. Sneha

    Abstract

    The automatic driver less tram concept presents an innovative solution for urban transportation, designed for efficient and autonomous operation in smart or metro cities. Utilizing tracks arranged alongside roads, this tram occupies minimal road space while integrating seamlessly with normal traffic, thereby reducing congestion. Its primary power source is solar energy, harvested through rooftop solar panels, ensuring eco-friendly and sustainable operation. The generated energy is stored in rechargeable batteries, allowing continuous operation even in the absence of sunlight. Automatic features such as auto doors, headlights, horn, and object detection enhance safety and convenience. Specially designed grooved wheels maintain alignment with the track, while a DC motor and spur gears ensure smooth propulsion. Destination points are precisely identified using Hall Effect sensors, enabling accurate halting. Passenger alerts, including halting time displays and countdown timers for door closure, ensure safety and efficiency. Overall, this automatic driver less tram offers a comprehensive solution addressing traffic congestion, environmental concerns, and passenger comfort, making it an ideal choice for modern urban transit systems.

12. An Assessment on the Performance of a Composite Learning System for Prediction Using Precision, Recall, Accuracy, and F1-Score for Rainfall

    G. Ravi Kumar, V. Venkataiah, Borra Sivaiah, B. Kavitha Rani, Kanthi Murali, G. Swathi

    Abstract

    Several countries’ whole economies rest on rain. When rain is predicted early, the results on several fields can be seen. Predictions of rain that are accurate and made on time can help the building industry, transportation, agriculture, flying, and people who are worried about flooding. Every year, heavy rains cause a lot of destruction to both structures and people's lives. Plenty of investigations are being done to figure out how to predict rain based on where it is and how the weather is. Every year, rain-related disasters cause a lot of damage and loss of life, both to structures and to people. This study suggests using the Indian dataset and the Composite Learning Algorithm (CLA) to predict rainfall using global characteristics. Using data from the UCI repository, the method is used to predict how much rain there will be. The meteorological factors in the dataset are used to forecast the rain more accurately. Precision, memory, and accuracy, as well as the F1-score, are used to measure performance.

13. GNITS-NaviGlow: College Navigation Chatbot

    Sripriya Nittala, Nidhi Bagade, Sowmya Konkala, P. Dharani, B. Sasidhar

    Abstract

    For new students and their parents, the task of finding their way about and seeking particular places within the campus is sometimes very difficult due to the nature of the campus and the identity of specific destinations. Asking manual directions to people for help often causes misunderstandings, making it even harder to adjust to a new environment. To tackle this problem, the authors propose a sophisticated chatbot that acts like a navigation tool in the GNITS campus. The chatbot will have a speech recognising system and will use Artificial Intelligence chatbot to give efficient navigation to newcomers in the campus.

14. Smart Control of Traffic Light Using Artificial Intelligence

    N. Sujata Kumari, K. Kavya, Bh. Sirisha, S. Bhoomika

    Abstract

    The proliferation of both people and cars in urban areas has led to a number of serious problems, one of the most pressing being traffic congestion. Not only can traffic congestion add stress and waste time in drivers’ lives, but it also increases pollution and fuel consumption. It is a widespread issue, particularly affecting megacities. As urban population grows, the real-time assessment of road traffic density becomes vital for better signal control and efficient traffic management. The effectiveness of traffic flow largely depends on the traffic controller, highlighting the need for advancements in traffic management to cope with increasing demands. The proposed system leverages image processing and AI to analyze live images from traffic junction cameras to determine traffic density accurately. Additionally, it zeroes in on the algorithm that controls the traffic lights according to the number of vehicles, which speeds up transportation for people and cuts down on pollution.

15. Machine Learning Approaches for Anomaly Detection: A Comprehensive Review

    S. Gopalakrishna, B. Kishore, K. Haripalreddy, V. Sumathi, PradeepKumar, G. Archana

    Abstract

    Anomaly detection is an important job in many different fields, such as cybersecurity, healthcare, finance, and industrial systems. In these areas, identifying deviations from typical behaviour may help avert large undesirable effects. Another important domain is industrial systems. When confronted with the complexity and high dimensionality of current datasets, traditional approaches for anomaly identification sometimes fail to identify anomalies. In order to solve these issues, machine learning methodologies, and more specifically deep learning techniques, have emerged as formidable tools. The purpose of this work is to present a complete review of machine learning algorithms for anomaly identification. These techniques include supervised, unsupervised, and semi-supervised methods. We investigate the recent developments in deep learning models, including Autoencoders, Generative Adversarial Networks (GANs), and Recurrent Neural Networks (RNNs), as well as the applications of these models in anomaly detection. In order to successfully evaluate the performance of these models, evaluation criteria and procedures are given. In addition to this, we investigate a wide range of real-world applications and case studies, focussing on the effect that machine learning-based anomaly detection has had in a variety of industries.

16. A Comprehensive Investigation of Energy-Efficient Wireless Sensor Networks in Consideration of Optimization and Scheduling Techniques

    Amit Kumar, Deepak Prasad

    Abstract

    This study presents a comprehensive investigation into energy-efficient wireless sensor networks (WSNs), focusing on optimization and scheduling techniques. This paper evaluates the key factors influencing energy consumption in WSNs, such as network architecture, data collection, security threats, and deployment techniques. The energy-efficient scheduling routing protocols are explored to enhance energy efficiency and security in sensor networks. The research uses MATLAB simulations to model optimal scheduling, showing a significant reduction in energy consumption compared to traditional methods. This exploration suggested that the approach to improve energy efficiency using energy-efficient scheduling, particularly in high-data-rate WSN environments.

17. Mitigating Driver Fatigue: A Multisensory Approach with Advanced Machine Learning Techniques

    Arika Asha Susmitha, Gopu Akshay, Marri Sahasra Reddy, Pavan Kumar Pagadala, Chanda Raj Kumar, Sree Lakshmi Pinapatruni

    Abstract

    The Drowsy Driver Detection system is about generating warnings/notifying danger by providing some sound to drivers and allowing them to take corrective action such as pulling over and resting. These systems are an important component of road safety and can help to reduce the risk of accidents caused by driver extreme tiredness or exhaustion. This system employs a variety of sensors and technologies to monitor driver behavior and includes infrared cameras tracking eye movement and facial expressions, and lane departure warning systems. With the help of some modern techniques like OpenCV (Computer vision), SVM (Support vector machine), and Gradient descent models of ML (Machine Learning), we will be working on this project.

18. Unlocking Customer Insights: Analyzing Factors Influencing Term Deposit Subscriptions in the Banking Industry

    Danam Sree Mahima, Raj Kumar Chanda, Pavan Kumar Pagadala, Sree Lakshmi Pinapatruni

    Abstract

    The Term Deposit Analysis project aims to uncover the essential factors that influence customer subscription to term deposit products in the banking industry. Term deposits are a crucial financial instrument for both banks and customers, providing a secure and predictable way to save and invest money. Understanding the determinants of customer decision-making in subscribing to term deposits is vital for banks to tailor their marketing strategies, optimizing the resources allocation, and enhancing the satisfaction of customers. The project leverages a comprehensive dataset containing customer attributes, banking interactions, economic indicators, and historical subscription data. We employ advanced data analysis techniques, including data preprocessing, exploratory data analysis, feature engineering, and predictive modeling, to extract valuable insights from the data. The significance of the project lies in its holistic approach, encompassing customer-centric attributes, broader economic influences.