Rashtreeya Sikshana Samithi Trust

This review initiative had explored the applicability of the extract isolated from various plant parts of Sapindus (S.) species as the potential green corrosion inhibitor (GCI). Additionally, insights on topics like plant extracts in broad, concept of corrosion inhibition by the pool of biomolecules, use of (S.) species plant parts as the extract to retard corrosion of metals/alloys, composition of (S.) species extract, extraction techniques, etc were also covered. This article would form a platform to establish the present global status of (S.) species towards corrosion inhibition filed and related sectors (surfactant based). The favorable functional groups featuring in the biomolecules (having maximum hydrophilic part) along with the dominant inhibitor component (saponins) present in the (S.) extract had assisted the effective inhibition of corrosion in metals/alloys. This behavior of the extract (having the pool of biomolecules) can be attributed to various factors like superior adsorption, firm complexation and impermeable layer development over the surface of metal/alloy. This review initiative can promote the use of (S.) extract in commercial scale to retard the corrosion of various metals/alloys and hence can add financial angle for the cultivators.

Wireless Sensor Networks (WSNs) are fundamental to Internet of Things (IoT) systems but are increasingly vulnerable to adversarial attacks, data anomalies, and the absence of reliable forensic mechanisms. This paper proposes a full-stack, modular simulation framework that integrates lightweight MQTT-based communication, deep learning-powered anomaly detection, and blockchain-backed immutable logging to secure real-time sensor data pipelines. The architecture incorporates a dual-layer security model-comprising RSA-based source authentication and an Autoencoder-based anomaly detection pipeline—combined with majority voting consensus for anomaly resolution. Verified data is recorded in a custom Proof-of-Authority blockchain to guarantee auditability and tamper-evidence. The system includes a web-based dashboard for simulating sensor behavior, launching targeted cyberat-tacks, tracking anomalies, and generating PDF reports for audit trails. Experimental evaluations demonstrate high detection accuracy (> 90%), low transaction latency (<300 ms), and resilience against attacks such as Replay, Sybil, and Constant Value injection. This research lays the groundwork for scalable, secure, and autonomous IoT deployments in domains such as smart infrastructure, industrial monitoring, and environmental sensing.

Hybrid beamforming has emerged as an effective solution for high-frequency multi-antenna communication systems, enabling high spectral efficiency with reduced hardware cost. This work presents an AI-assisted hybrid beamforming framework that integrates traditional signal processing with deep learning for multi-user MIMO scenarios. The proposed system models a patch antenna array, enhanced channel conditions, and realistic line-of-sight/non-line-of-sight propagation to generate training datasets via optimal precoder computation. A neural network is designed to predict complex beamforming weights from estimated angles of arrival, using a correlation-based loss to align with optimal solutions. The framework also incorporates a direction-of-arrival estimator based on the MUSIC algorithm, enabling a hybrid approach where AI-predicted and classical beamforming weights can be compared or combined. Numerical simulations demonstrate that the proposed method achieves high beamforming accuracy and robustness to noise while significantly reducing the computational burden compared to purely optimization-based designs.

This work presents the design and simulation of a scalable enterprise network using Cisco Packet Tracer. The network follows a hierarchical three-layer architecture with core, distribution, and access layers. VLANs are used to logically separate departments-HR (VLAN 10), Finance (VLAN 20), Engineering (VLAN 30), and Management (VLAN 40). Inter-VLAN routing is enabled using a Layer 3 switch configured with Switched Virtual Interfaces (SVIs) for seamless communication. As Multiple Spanning Tree Protocol (MSTP) is not supported in Cisco Packet Tracer, basic Spanning Tree Protocol (STP) is implemented for redundancy and loop prevention. Trunk links using 802.1 Q encapsulation are configured to carry multiple VLANs across switches. Redundancy between core switches and STP priority tuning is applied to simulate root bridge scenarios. Performance metrics such as bandwidth usage, availability, and failover time are measured using Packet Tracer. The network achieved an average throughput of 51.2Kbps and an average latency of 1.5 ms, reflecting efficient data transmission and minimal delay. Additionally, Cytoscape is employed to visualize and analyze network optimization using Prim's and Kruskal's algorithms. This integrated approach demonstrates logical segmentation, redundancy, and scalability, offering an efficient and faulttolerant network solution in a simulated enterprise environment.

In the present scenario of the digital revolution, smart home automation is experiencing rapid growth due to the need for convenience, energy efficiency, and centralized home appliance control. Commercial offerings such as Amazon Alexa and Google Home have some inherent limitations, such as the need for permanent internet connectivity and support for few local languages. These limitations minimize their usability in areas with weak network infrastructure and exclude non-English-speaking clients from efficiently using them. To overcome such limitations, this study proposes a multilingual, offline-enabled, and voiceauthenticated smart home automation system. The system employs the ESP32 microcontroller to enable local voice authentication and command execution, thus allowing only the authorized person to use the system offline. Simultaneously, the Arduino UNO is interfaced to a Bluetooth enabled mobile application that runs offline and enables users to control home appliances using a local wireless network. The system is enabled to support five languages Kannada, English, Hindi, Tamil, and Telugu thus making it inclusive. Users are empowered to perform routine activities, such as light, fan, and door operation, using voice commands in their local language, thus making the smart home environment secure and inclusive and empowering them to become independent.

With the proliferation of connected devices and constantly evolving attack techniques, modern networks require security systems that can see the threat landscape, understand what the events mean and respond autonomously. This paper presents an AI-powered adaptive firewall that combines deep packet inspection via the open-source Suricata IDS/IPS with large-language-model (LLM) reasoning and the FreeBSD/macOS PF firewall. Suricata inspects network traffic in real time, logs suspicious events and supplies structured alerts in JSON form. These alerts are parsed, summarized using an LLM and then translated into precise firewall table commands. The adaptive firewall dynamically blocks or unblocks IP addresses without manual intervention, generates periodic PDF reports for administrators and exposes a web dashboard and CLI for naturallanguage commands. As evaluated in Section VI, by leveraging LLM-based summarization and dynamic PF tables, the system reduces time to mitigation and lowers false positives compared with static rule sets.

Usage-based Storage-as-a-Service (STaaS) platforms require continuous telemetry (e.g. IOPS, latency, capacity) from customer-edge storage for SLA enforcement and billing. However, conventional cloud-centric collectors generate excessive bandwidth overhead, slow SLA violation detection, and impose heavy CPU/RAM load on edge sites. This paper presents Keystone-Edge, a scalable, adaptive telemetry framework that addresses these issues with four key innovations: (1) an Adaptive Sampling algorithm that dynamically adjusts data collection frequency based on workload volatility; (2) a Local SLA Violation Detection module for real-time alerting; (3) a Differential Transmission protocol that sends only significant metric changes; and (4) a suite of Resilience and Security enhancements (e.g., encrypted channels, persistent buffering, retry/backoff) to ensure robust, privacy-preserving operation. We validate the design via custom simulations of ONTAP storage workloads. Experimental results show up to 57% reduction in telemetry data volume, SLA breach detection in <5 seconds, and low resource use (CPU <9%, RAM <600MB ) compared to a naive collector. We also discuss how the architecture scales to 1000+ sites and trades off sampling rate versus resource cost. Keystone-Edge thus provides an efficient, secure solution for edge telemetry in on-premises STaaS, combining intelligent data reduction with rapid responsiveness.

Objectives: Diuretic therapy in liver cirrhosis is traditionally adjusted through a trial-and-error process, characterised by inefficiencies, frequent therapy modifications, prolonged hospital stays and high economic costs. Predictive analytics powered by machine learning (ML) offers a transformative solution, enabling personalised and precise interventions that minimise regimen changes, expedite recovery and enhance healthcare efficiency. This study evaluates the impact of ML-guided diuretic therapy on reducing treatment modifications, hospitalisations and associated costs, ultimately delivering precision-driven outcomes. Materials and Methods: An ambidirectional cohort study analysed 481 cirrhotic patients receiving diuretic therapy. Key clinical data, including liver function markers, serum electrolytes and urinary output, were collected to analyse therapy modifications and identify predictors of suboptimal outcomes. ML classification models were trained to predict optimal diuretic regimens using one-hot encoding for categorical variables. Model performance was assessed through the area under the curve, accuracy, sensitivity and specificity. Hyperparameter optimisation enhanced predictive accuracy, while feature importance analysis identified critical predictors of diuretic response. Results: Under the conventional approach, 71.5% of patients initially received suboptimal therapy, leading to add-on treatments (44%) or complete regimen changes (27%). The ML model reduced therapy modifications by 58%, achieving 81% accuracy in predicting optimal regimens. Key predictors of diuretic response included serum albumin, bilirubin and creatinine levels. Incorporating ML reduced the time to optimal therapy from 7–8 days to 4 days, shortened hospital stays by 40% and significantly lowered healthcare costs. Conclusion: ML-guided diuretic management in liver cirrhosis represents a paradigm shift, delivering precision-driven care that minimises therapy changes, shortens hospital stays and reduces economic burden. By leveraging key clinical predictors, this approach accelerates recovery and enhances treatment precision. Further validation in larger datasets and real-world settings is critical to establishing ML as a cornerstone of fluid management in cirrhosis, enabling cost-effective, patient-centred care.

Tomato (Solanum lycopersicum L.) is a globally important horticultural crop; however, inconsistent germination and weak early seedling vigour remain major bottlenecks to productivity. Seed nanopriming using zinc oxide nanoparticles (ZnO NPs) represents an emerging, eco-compatible strategy for enhancing seed performance by modulating early biochemical and physiological processes. This study investigated the influence of ZnO NPs seed priming on seed quality traits and physio-biochemical responses in three genetically distinct tomato varieties: Pusa Rohini, H-81, and Pusa Prasanskrit. The optimal concentration (50ppm) of ZnO NPs significantly enhanced germination percentage, speed of germination, seedling vigour indices, shoot and root elongation, and biomass accumulation while reducing mean germination time. Biochemically, ZnO nano-primed seedlings exhibited elevated levels of total chlorophyll, proline, and total phenol along with enhanced antioxidant enzyme activities (superoxide dismutase, catalase and peroxidase). A concurrent reduction in malondialdehyde content indicated lower lipid peroxidation and oxidative damage. These effects collectively suggest improved homeostasis and metabolic efficiency. The findings confirm that 50ppm ZnO nanopriming modulates germination-linked signaling and antioxidant defense pathways, accelerating early seedling establishment and enhancing seed quality through improved enzyme activity and seedling development in tomato, while offering a cost effective and environmentally sustainable approach.