Prompt engineering is crucial for optimizing the performance of Large Language Models (LLMs), yet it remains a manual and resource-intensive process that requires multiple iterations of trial and error. Current automated prompt enhancement approaches face key challenges in preserving component relationships, managing computational requirements, and maintaining optimization traceability. This paper introduces CARE (Comprehensive Analyzer & REfiner), an LLM-based dual-agent framework that models prompt enhancement as a staged transformation pipeline with explicit validation constraints. CARE tackles three fundamental challenges: prompt decomposition with interleaved dependencies, component interference in LLM processing, and semantic drift during refinement. The framework enables reliable prompt enhancement in a single iteration through systematic component extraction and rule-based transformations. Evaluation using established benchmarks demonstrates consistent improvements across diverse LLM architectures.

Attack paths are the potential chain of malicious activities an attacker performs to compromise network assets and acquire privileges through exploiting network vulnerabilities. Attack path analysis helps organizations to identify new/unknown chains of attack vectors exposing critical assets, as opposed to individual attack vectors in signature-based attack analysis. Timely identification of attack paths enables proactive mitigation of threats. Nevertheless, manual analysis of complex network configurations, vulnerabilities, and security events to identify attack paths is rarely feasible. This work proposes a novel transferable graph neural network-based model for shortest path identification. The shortest path, integrated with a novel holistic model for identifying potential network vulnerabilities interactions, is then utilized to detect network attack paths. Our framework automates the risk assessment of attack paths indicating the propensity of the paths to enable the compromise of highly-critical assets (e.g., databases). The proposed framework, named SPGNN-API, incorporates automated threat mitigation through a proactive timely tuning of the network firewall rules and Zero-Trust (ZT) policies to break critical attack paths and bolster cyber defenses. Our evaluation process is twofold; evaluating the performance of the shortest path identification and assessing the attack path detection accuracy. Our results show that SPGNN-API largely outperforms the baseline model for shortest path identification with an average accuracy$\geq95$% and successfully detects 100% of the potentially compromised assets, outperforming the attack graph baseline by 47%.

Micro-segmentation is an emerging security technique that separates physical networks into isolated logical micro-segments (workloads). By tying fine-grained security policies to individual workloads, it limits the attacker’s ability to move laterally through the network, even after infiltrating the perimeter defences. While micro-segmentation is proved to be effective for shrinking enterprise networks attack surface, its impact assessment is almost absent in the literature. This research is dedicated to developing an analytical framework to characterise and quantify the effectiveness of micro-segmentation on enhancing networks security. We rely on a twofold graph-feature-based framework of the network connectivity and attack graphs to evaluate the network exposure and robustness, respectively. Tracking the variations of formulated metrics values post the deployment of micro-segmentation reveals exposure reduction and robustness improvement in the range of 60% – 90%.

In the recent years, intelligent vehicles became the most common everyday task that attracts a great interest from the industry. This work is dedicated to enhance the vehicles' intelligence through an adaptable four dimensional model capable of impersonating the vehicle driver by accurately predicting the time and location of destinations. Existing research efforts have addressed this challenge through prediction models that consider one or two aspects of human behavior, such as the social network or location semantics. However, the accuracy of these models remains rather limited. Accordingly, this paper addresses the shortfalls of the existing work by incorporating most aspects of human mobility. Moreover, we study the impact of each of the model dimensions on the prediction accuracy, individually and combined. We further propose an optimization algorithm to calculate the best blend of the model dimensions, that maximizes the prediction accuracy. The performance of the model is evaluated using a dataset that is collected from users in the city of UIm, Germany, as well as through computer simulations. The results show that the model can achieve a prediction accuracy of 95%, outperforming the state-of-the-art counterparts.

Ubiquitous computing is the growing trend of establishing a place for technology in the background, rather than the foreground, of our lives through embedding computational capability into everyday objects to make them effectively communicate and perform useful tasks. Hence, the goal of pervasive computing is to make devices "smart". Standing on this ground, the model subject of this work enhances the vehicles intelligence through enabling them to accurately predict the drivers destination without their intervention through an optimized four dimensional destination prediction model. Moreover, it takes the lead in leveraging the semantic properties of both the locations of interest and the social relations affecting the visiting activities of the drivers. In line with this, our proposed model simultaneously utilizes an optimization algorithm to blend the different dimensions and to accommodate the evolving human mobility behavior. The experimental results demonstrate that the optimized model significantly outperforms the basic model achieving an average prediction accuracy of 95% when compared to the real trace. Show the XML Only

Simulation is a key feature in validating the protocols dedicated for VANET applications. Since the vehicles mobility is driven by the human mobility characteristics and is controlled by the geographical restrictions of the roads, the work presented in this paper is aimed to having a realistic mobility model that incorporates both the social aspects of human mobility together with the geographical restrictions that governs the movement of the mobile nodes. This mobility model is dedicated for testing the protocols related to the delay tolerant applications where unicasting/multicasting is used for data dissemination rather than broadcasting. Thus location awareness and destination prediction is essential. The model is based on using realistic data sets rather than randomly generated data. Evaluating the technique using the SUMO simulator have proven a range of 80% to 90% of match between the predicted traces and real life data available for evaluation.