In most of the traditional cluster-based hierarchical routing protocols, the cluster head (CH) selection is made on a random basis. As a result, some unlucky sensor nodes (SNs) become dead quickly; thereby, network lifetime reduces drastically. To overcome this problem, in this paper, a new cluster-based routing protocol- Energy-Aware Threshold Sensitive Stable Election Protocol (EATSEP) is presented for wireless sensor networks (WSNs). In the EATSEP protocol, the CH selection is an optimum process where the initial and residual energy of each SNs are considered within a heterogeneous SNs energy environment. Additionally, our proposed EATSEP protocol has managed to reduce long-distance transmission by routing data among CHs to the base station. In our present study, we have simulated the EATSEP protocol through MATLAB to compare its performance with other popular protocols under some well-known performance metrics. The experimental results indicate that the network stability of the EATSEP protocol improves by 80.81, 66.41, and 27.06 %, respectively, compared to the low-energy adaptive clustering hierarchy (LEACH), Stable Election Protocol (SEP), and Threshold Sensitive SEP under a particular setting. In terms of energy consumption and network throughput, the EATSEP is also superior to other protocols.

Large Language Models (LLMs) are increasingly viewed as demonstrating strong reasoning ability due to their performance in math, logic, and planning tasks. However, most evaluations rely on overall accuracy on fixed benchmarks, making it difficult to understand how reasoning changes as task complexity increases. In this study, we present a controlled evaluation framework to assess the robustness of reasoning in Large Reasoning Models (LRMs) under growing complexity. We design nine classical reasoning tasks, including boolean satisfiability, cryptarithmetic, graph coloring, river crossing, tower of hanoi, sudoku, and rubik’s cube, where difficulty can be scaled while preserving the same context. Using deterministic validators, we evaluate open and proprietary LRMs across low, medium, and high complexity levels, ensuring that only fully valid solutions are accepted. Our findings show that models perform well at low complexity but degrade rapidly as difficulty increases. We formalize this phenomenon as reasoning collapse, characterized by a sharp decline in success probability with increasing complexity. Common failure modes include inconsistent reasoning traces, rule violations, state loss, and confidently incorrect answers. These findings suggest that current LRMs rely on fragile heuristic reasoning strategies and highlight the need for evaluation methods that explicitly measure reasoning robustness beyond static benchmarks.

With the explosive rise of internet usage and the development of web applications across various platforms, ensuring network and system security has become a critical concern. Networks and web services are particularly susceptible to targeted attacks, as hackers and intruders persistently attempt to gain unauthorized access. The integration of artificial intelligence (AI) has emerged as a crucial tool for detecting intrusions and constructing effective Intrusion Detection Systems (IDSs) to counter cyber-attacks and malicious activities. IDSs developed using machine learning (ML) and deep learning (DL) techniques have proven to be highly effective in detecting network attacks, offering machine-centric solutions. Nevertheless, mainstream adoption, confidence and trust in these systems have been greatly impeded by the fact that ML/DL implementations tend to be “black boxes,” and thus lacking human interpretability, transparency, explainability, and logical reasoning in their prediction outputs. This limitation has prompted questions about the responsibility and comprehension of AI-driven intrusion detection systems. In this study, we propose four novel architectures that incorporate Explainable Artificial Intelligence (XAI) techniques to overcome the challenges of limited interpretability in ML/DL based IDSs. We focus on the development of ExplainDTC, SecureForest-RFE, RationaleNet, and CNNShield architectures in network security solutions, and inquiry into their potential to convert the untrustworthy architectures into trustworthy. The models are applied to scan network traffic and identify, and report intrusions based on the traits extracted from the UNSW-NB15 dataset. To explain how a decision is made by the models and to add expansibility at every stage of machine learning pipeline, we integrate multiple XAI methods such as LIME, SHAP, ElI5, and ProtoDash on top of our architectures. The generated explanations provide quantifiable insights into the influential factors and their respective impact on network intrusion predictions.