Background: Rice (Oryza sativa L.) productivity is increasingly constrained by climate variability, biotic stresses and limited natural resources, necessitating the effective utilisation of genetically diverse germplasm. Traditional rice landraces possess valuable adaptive, stress-tolerant and yield-related traits; however, their potential remains largely underutilized in breeding programmes. Methods: The present investigation was conducted during the zaid season of 2025 under field conditions using 130 rice (Oryza sativa L.) landraces evaluated in an augmented experimental design. A total of 21 agronomic, physiological and reproductive traits were assessed. Genetic divergence and trait relationships were analysed using principal component analysis (PCA) and hierarchical cluster analysis to identify the major traits contributing to phenotypic variability and yield performance. Result: Eight principal components with eigenvalues greater than one explained a substantial proportion of the total phenotypic variation. The first two components contributed most of the variability and were primarily associated with plant vigour, reproductive efficiency, biomass accumulation and yield-related traits. Plant height (X3), flag leaf length (X4), internode length (X7), grains per panicle (X9), 1000-seed weight (X10) and pollen fertility (X19) were the major contributors to genotype differentiation. Cluster analysis grouped the 130 genotypes into four distinct clusters. Groups I, II and IV exhibited high pollen fertility and superior yield performance, whereas Group III showed lower pollen fertility and poor yield traits. PCA biplot analysis further revealed a close association of high-yielding genotypes with pollen fertility, biomass traits and physiological efficiency. pollen fertility emerged as a key trait influencing genetic divergence and yield performance. The genetically diverse and superior-performing genotypes identified in this study represent valuable resources for rice improvement programmes aimed at enhancing productivity and adaptability.
R. Mahendran, B. Hariharan, K. Satheeskumar et al.· Indian Journal of Agricultur...· 0 citations
Cloud computing has evolved into a mature technology, seamlessly integrating with modern internet services and functioning as utility computing that delivers infrastructure, platforms, and software on a pay-per-use basis. A key challenge in cloud computing is task scheduling, which significantly impacts both user satisfaction and system performance. Due to the NP-hard nature of the scheduling problem, developing efficient solutions remains complex. This paper proposes an orthogonal opposition-based learning partial reinforcement optimizer (OOLPRO) for efficient task scheduling in IoT-cloud environments. The OOLPRO framework integrates orthogonal oppositional functions (OOF) with the partial reinforcement optimizer (PRO) to overcome the drawbacks of conventional PRO, such as insufficient solution exploitation and premature convergence. The proposed framework integrates orthogonal opposition-based learning with reinforcement-driven optimization to enhance exploration–exploitation balance, accelerate convergence, and improve scheduling performance in terms of energy consumption, execution cost, makespan, and resource utilization compared with existing approaches. The proposed algorithm is designed to optimize multiple conflicting objectives, including cost, energy consumption, and makespan, thereby ensuring efficient allocation of tasks to physical machines within cloud data centres. By incorporating OOF functions, the algorithm enhances the exploration–exploitation balance, leading to improved convergence rates and higher-quality solutions. The performance of OOLPRO is evaluated through extensive simulations and benchmarked against existing task scheduling algorithms. Experimental results demonstrate that the proposed method achieves superior efficiency, resource utilization, and scalability, making it a promising approach for optimizing task scheduling in dynamic cloud computing environments.
Ratnakumari Neerukonda, B. Hariharan· Discover Computing· 0 citations
In recent times, industries are expanding globally to meet the needs of growing population. Government and other regulatory bodies monitor industries to ensure compliance with standards particularly discharges that may pose threat to nearby residents. However, regulatory bodies often struggle with limited human resources that hinder the effectiveness of audit. This issue can be addressed by installing Internet of Things (IoT) devices in the industry to collect the data and store it on the servers, enabling regulatory bodies to review it remotely and conduct onsite inspection when anomalies detected. The challenging aspect is that the centrally stored data are vulnerable to cyberattacks and it can be tampered. Blockchain technology can provide a promising solution as it stores the data in the distributed ledger that is inherently tamper proof. However, using blockchain to store the data collected by IoT devices face challenges related to scalability. Further, latency will be also a concern due to the time taken by the consensus mechanism to validate the blocks and add them to the chain. To address these challenges, we introduce a novel consensus algorithm named Proof of Reward (PoR) tailored for applications demanding high throughput. The PoR consensus mechanism is designed to improve resilience against Byzantine failures by enabling reliable pairwise decision making under distributed adversarial environments. Further, the proposed consensus minimizes communication overhead during the block finalization by considering first two third of node responses, thereby enhancing the overall network throughput. The communication overhead of the proposed consensus is 34.78% lesser compared to Practical Byzantine Fault Tolerance (PBFT) and 18.55% lesser compared to Istanbul Byzantine Fault Tolerant (IBFT). The obtained results highlight the effectiveness of the proposed consensus algorithm.
S. P.N., S. Kaliraj, Jaisingh Thangaraj et al.· Journal of Cloud Computing· 0 citations