Shall Indian Universities Dream Big and Adopt Global Visions & Missions and Become Entrepreneurial so Essential to Spin Off Deep Tech Startups?
PASTOR-DTN: A Predictive, Adaptive, Social-Trust Optimized Routing Algorithm for Delay Tolerant Networks
Performance Comparison of Advanced Neural Networks and Traditional Algorithms for Handwritten Digit Recognition on the MNIST Dataset
Integrating Dynamic Soil Classification with Pattern Recognition- Based Anomaly Detection for Precision Agriculture
A Comprehensive Survey on Intelligent Cluster Head Selection and QoS-Aware Routing Techniques in Vehicular Ad Hoc Networks (VANETS)
The current investigation briefs the need and importance of deep tech start-ups in a rapidly advancing world, especially in the context of India aiming to achieve developed nation status by the year 2047. The initiatives taken by the Government of India for giving birth to deep tech startups are outlined in brief. Deep tech startups essentially require higher education institutes and universities to become entrepreneurial and innovative for generating advanced breakthrough and disruptive technology so essential for spinning off deep tech enterprises. The changing role of universities and their becoming entrepreneurial in the new world is also presented. Academic and political leaders need to take necessary action for shaping them to spin off deep tech startups. It also describes the characteristics of entrepreneurial universities elsewhere in the world, which are known for giving birth to deep tech startups. Universities becoming entrepreneurial demands a change in mindset of Indian academic and political leaders. Will they change their mindset? They ought to realise vast new opportunities in the world for employment, increase in GDP and harvesting them through deep tech startups. The paper presents new visions & missions derived from world-ranking entrepreneurial universities that need to be adopted and implemented in Indian universities. Organisational structures like advanced research labs, research parks, technology incubation centres, and patent & IPR centres would have to be installed forthwith on university campuses. Curriculum in universities would have to be orientated to creativity, discovery, and commercial exploitation of deep technology. Lastly, the paper also brings out in brief problems in the adoption and implementation of deep tech startups in present Indian conditions. Success depends on whether India will be able to make its universities entrepreneurial in global comparison.
Delay-tolerant networks (DTNs) must move data without stable end-to-end paths, forcing routers to trade delivery, delay, and resource use. We present PASTOR-DTN, a Predictive, Adaptive, Social-Trust Optimized routing algorithm that unifies five local signals—encounter-based predictability, per-peer trust, social centrality, buffer headroom (reward free space, penalize congestion), and TTL urgency—into a single utility used to choose the next hop under an explicit transmission budget. PASTOR couples this multi-factor utility with a token-bucket governor that bounds overhead during contact bursts and a short, hard first-hop window with a brief relay head start that deliberately converts many direct paths into efficient two-hop deliveries without copy explosion. Implemented in The ONE and evaluated against representative baselines under identical mobility, radio, traffic, TTL, and buffer settings, PASTOR traces a favorable delivery–overhead Pareto frontier. At a balanced operating point, it achieves 97.13% delivery with ≈2.01 overhead ratio, 695.5 s median latency, 1.88 average hop count, and 18.1 ks median buffer time—i.e., near the delivery and median delay of aggressive schemes at more than 10x lower overhead ratio in our evaluated setting, while outperforming copy-constrained baselines on buffer occupancy at comparable or lower latency. By integrating predictive, social-trust, congestion, and urgency cues under a strict budget and an explicit first-hop bias, PASTOR provides a practical, tunable controller for DTN routing that advances beyond single-signal or unbudgeted designs.
Digit recognition in handwriting is a general issue of computer vision and pattern recognition, which is extensively used in postal systems, banking, and document processing. This paper provides a significant comparison between a developed Multilayer Perceptron (MLP) model and the other common machine learning and deep learning algorithms, such as a Convolutional Neural Network (CNN), Support Vector Machine (SVM), and k-Nearest Neighbors (KNN). Some of the methods that have been included in the advanced MLP to improve its performance and minimize the risk of overfitting include dropout, batch normalization, learning rate scheduling, and early stopping. The models are tested on the MNIST dataset with such measures as accuracy, precision, recall, and F1-score. Experiments prove that CNNs are the most accurate since they have a higher quality of feature extraction, but the advanced MLP significantly performs better than other architectures and classical algorithms, which provides a robust and efficient solution to practical digit classification problems. The results highlight the significance of architectural contrivance and regularization designs to the enhancement of the neural network and give information on the model selection in the real-world implementations
This current investigation is intended to create an overarching state-of-the-art system that integrates unsupervised soil clustering with pattern recognition-based anomaly detection for the intention of revolutionizing precision farming. Most conventional techniques of classifying soil do not involve dynamic variation in the properties of soil and are unable to detect anomalous conditions that affect agricultural productivity. By incorporating the application of adaptive incremental clustering algorithms and pattern-based analysis techniques, this research introduces a better solution that can classify soil dynamically according to various attributes in addition to outlier detection from defined patterns. The new architecture continues prior research in auto-incremental clustering for dynamic soil classification and industrial anomaly detection by adding a two-phase framework: a high-level sophisticated unsupervised learning algorithm for dynamic soil classification that learns to accommodate new soil samples and environmental conditions, and a high- level sophisticated pattern recognition system that detects anomalous soil conditions through temporal changes in soil parameters. This merging is anticipated to enhance classification precision by 15-20% over existing approaches and decrease false positive anomaly detection by over 30%, thus enabling farmers to make more accurate choices in precision agriculture based on more trustworthy data.
The application of vehicular Ad Hoc networks (VANET) is crucial for intelligent transport systems (ITS). This VANET application also facilitates vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communication. Here, maintaining Quality of Service (QoS) and ensuring efficient routing are still difficult tasks because of the following factors: high mobility, frequent topology changes, and variable node densities. For enhanced network stability and resource optimization, cluster- based (CB) routing protocols (RP) with the application of intelligent cluster head (CH) selection mechanisms have become an effective method. For VANET, recent advancements in intelligent CH selection (CHS) algorithms and QoS- aware RP were systematically reviewed in this study. For CHS and Quality of Service provisioning, 20 state-of-the-art (SOTA) approaches that utilize swarm intelligence (SI), fuzzy logic (FL), machine learning (ML), and hybrid methods are analyzed in this study. When evaluating every technique, the following factors have to be considered: selection strategy, routing efficiency, flexibility, scalability, and the impact on important QoS metrics like packet delivery ratio (PDR), throughput (THRPT), and delay (D). Researchers focus on future initiatives to improve network functionality, security, efficiency, and comparative performance analysis. This study also highlights the research gaps, and it may guide the future studies in the development of more robust and intelligent VANET architecture
