No abstract

IoT-based multi-biometric system is a blend of multiple biometric templates that can be used for user authentication/verification using sensors. The leakage of the biometric trait information may cause critical privacy and security issues. It is expected to protect the privacy details of individuals through the irreversibility, unlinkability, and renewability of multi-biometric templates used in the authentication system. This study presents a robust authentication system with secure multi-biometric template protection techniques based on discrete cosine transform feature transformation and Lagrange’s interpolation-based image transformation. Three biometric traits namely iris, fingerprint, and palm print are recorded using sensors to validate the proposed multi-biometric template protection system. The fusion of all traits used is giving an average of 95.42% genuine acceptance rate and an average of 4.57% false rejection rate. Despite any number of biometric templates used for authentication, the proposed image transformation techniques keep the size of the final storage requirement as 8 X 8, which achieves constant space complexity (O(1)). The stored template is not linked with original templates; it is irreversible and renewable as new enrolment of the same individual will produce a new template every time. Overall, the proposed technique provides a secure authentication system with high accuracy, a constant size database, and the privacy preservation of biometric traits.

The insect pests and crop diseases are the most critical factors that affect agricultural production, which reduces the sustainable development of agriculture. While detecting the pest, it is inconsistent to place the surveillance cameras near the target pests and the captured images from the Internet of Things (IoT) monitoring equipment at a constant location that is mostly insufficient for pest detection. IoT is a well-known advanced technology and an analytics system incorporated in diverse industries based on its unique abilities and flexibilities over a particular environment like agriculture. There is a demand for the IoT in agricultural areas to reduce the chemical crop protection agents and fertilizers to manage the efficient crop state and crop production. Hence, the data collection is through the IoT devices in this research model. This research aims to develop a pest identification and classification model to detect and identify the pests in the images. Initially, the IoT platform is created, and IoT devices conduct the data collection. Then, object detection is performed using Yolov3 to detect the pests in the images from the gathered images. The detected images are subjected to the Convolutional Neural Network (CNN) for gathering the deep features, which are then forwarded to the enhanced classifier, termed Convolution Neural Long Short-Term Memory (CNLSTM) for getting the classified outcomes as pest details, in which the optimization of parameters is done by Adaptive Honey Badger Algorithm (AHBA). These results demonstrated that the proposed method shows enhanced performance by rapidly collecting the information in agriculture and ensures the technical indication for population estimation and pest monitoring.

The allocation of resources in the cloud environment is efficient and vital, as it directly impacts versatility and operational expenses. Containers, like virtualization technology, are gaining popularity due to their low overhead when compared to traditional virtual machines and portability. The resource allocation methodologies in the containerized cloud are intended to dynamically or statically allocate the available pool of resources such as CPU, memory, disk, and so on to users. Despite the enormous popularity of containers in cloud computing, no systematic survey of container scheduling techniques exists. In this survey, an outline of the present works on resource allocation in the containerized cloud correlative is discussed. In this work, 64 research papers are reviewed for a better understanding of resource allocation, management, and scheduling. Further, to add extra worth to this research work, the performance of the collected papers is investigated in terms of various performance measures. Along with this, the weakness of the existing resource allocation algorithms is provided, which makes the researchers to investigate with novel algorithms or techniques.

No abstract