An Intrusion Detection System is crucial for securing Internet of Things. The conventional Intrusion Detection System technologies face challenges in detecting minority-class intrusions due to class imbalance and limited ability to model intricate dependencies in network traffic. To address these challenges, this research provides an intelligent framework to detect network intrusion in an Internet of Things network named Self-Supervised Multi-Residual Attention Graph Network (SS-MRAGNet). Specifically, SS-MRAGNet integrates node and edge aggregation with a multi-residual attention mechanism to learn complex structural dependencies within network traffic data. The framework incorporates self-supervised pre-training with pseudo-labeling and edge augmentation to address the class imbalance challenges and detect minority-class intrusions accurately. A minimal Random Under Sampling strategy is applied only within training folds to stabilize optimization, without serving as the primary imbalance-handling mechanism. Then, the model performs iterative self-training to generate high-confidence predictions, which are then used to pseudo-label unlabeled data. After the iterative process, predictions are refined. Rademacher complexity framework and convergence proofs are utilized for theoretical analysis to analyze its performance in noisy and imbalanced data. On validation, CIC-Darknet2020, CSE-CIC-IDS2018, UNSW_NB15, and Ton-IoT datasets result in 98.25
Precision agriculture is one of the major breakthroughs which has been adopted to counteract some of the major challenges facing contemporary agriculture, such as maximizing crop yield, maximizing resource efficiency, and sustainability. The application of Artificial Intelligence (AI) and the Internet of Things (IoT) technologies can go a long way in enhancing crop yield prediction through data obtained from multiple sources, including satellite images, ground sensors, and weather stations. This research paper proposes a next-generation crop yield prediction system called Geo-Optimized Deep Learning (GODL) algorithm. The GODL algorithm integrates geospatial information with deep learning algorithms for crop yield analysis and prediction. This combination of deep learning and geospatial data allows accurate and scalable crop yield predictions using geospatial and sensor information. In order to ensure model optimization in terms of accuracy and efficiency, the GODL Algorithm makes use of advanced hyperparameter optimization methods that include Adam and Stochastic Gradient Descent (SGD). The research paper highlights the efficiency of the GODL Algorithm by conducting extensive experiments that compare the algorithm’s performance with other crop yield prediction models like Recurrent Neural Network (RNN) and other machine learning techniques. The evaluation takes into consideration important metrics such as accuracy, Jaccard Index, Fowlkes-Mallows Index (FMI), and G-mean, as well as IoT data delay and its effects on the prediction accuracy. In future research, the use of blockchain technology to facilitate data sharing and edge computing to process the data will be considered.
The use of Orthogonal Frequency Division Multiple Access (OFDMA) was first introduced to Wireless Local Area Networks (WLANs) in the IEEE 802.11ax standard. Under the OFDMA mechanism, users can access different channel bandwidths to transmit their packets. As the bandwidth demands of users using the Internet increase, how to allocate channel resources for the OFDMA mechanism to improve the channel utilisation and fairness of the network has been a challenge. This paper proposes a modified Dynamic Programming algorithm combined with the Timsort (MDPT) to solve the resource allocation problem for downlink traffic under the OFDMA mechanism. The experiments compare the channel utilisation, Jain’s Fairness index (JFI), R-value and computation time of MDPT with other optimisation techniques. According to the simulation results, MDPT can generate an RU allocation scheme that improves the efficiency and fairness of the network, preserves packet ordering and can generate an optimal RU allocation scheme within an RU allocation interval.
Recent demand for Internet of things (IoT) based sensing devices has increased in many significant areas, such as greenhouse monitoring, remote patient monitoring, military applications, wild-life tracking and habitat monitoring, precision agriculture, etc. Moreover, some critical issues are present in the perpetual sensor network like node energy consumption rate, coverage/connectivity, quality of service, network throughput and charging cycle capacity of sensor node batteries. Thereby, various network issues can arise, especially regarding coverage holes, proper connectivity, and network life-time. In this paper, we have studied these problems well and have shown the necessary constraints for the connectivity of the IoT sensor network. Moreover, we have also presented various helpful IoT applications and energy harvested IoT sensor networks. Any network can perform well for a long time because of energy management which is very crucial. This paper presents various feature scopes used in wireless sensor networks (WSNs). It also illustrates distinct approaches like relay node placement, efficient communication, machine-learning systems i.e. reinforcement learning, and heuristic techniques. It also discusses real-world deployments to highlight practical connectivity challenges and failures.
Decentralized applications increasingly require eligibility checks based on Web2-held attributes, yet provider-side signing and centralized attestations limit deployability and can expose sensitive user data. This paper presents Scribe, a distributed eligibility verification service that bridges TLS-derived Web2 evidence to decentralized applications without any modification to Web2 servers. In Scribe, a client engages in a standard TLS session with a Web2 service and, using a DECO/TLSNotary-style provenance workflow, obtains an attestation node signature on a commitment to an extracted value. The client then generates a transparent zero-knowledge proof that verifies the attestation and proves compliance with an application policy without revealing the underlying value. Each proof outputs an application-scoped spend tag (nullifier) to provide replay resistance. An aggregator verifies proofs off-chain, maintains a Merkle-indexed spent-tag set, and publishes batched state roots to an on-chain anchor. Root publications are authorized using a post-quantum signature scheme, providing post-quantum integrity for the on-chain state without requiring on-chain proof verification. We specify the protocol and establish provenance binding, policy soundness, privacy, and replay resistance under stated trust assumptions.
