Modular caterpillar robots are becoming a universal platform for advances in robotics research, especially when simulated in complex environments. This research utilizes the implementation of the “WEBOT” simulator to investigate a modular caterpillar robot developed based on the “Yamor” prototype. Three bio-inspired approaches with different control methods are compared to assess their impact on the motion function of caterpillar-like robots. Performance is evaluated using parameters such as average speed, stability, and total distance over different media. We tested these three approaches in six environments: Earth (clean soil), Earth (desert), Earth (water), Earth (oil), Mars, and the Moon. We also considered functionality in hazardous conditions where random problems can occur during movement. For this reason, we included the accidental detachment of a module to investigate the behavior of the robot after detachment.

Satellites, high-altitude drones, and 5 G links in space now support critical services such as air traffic control, financial transactions, and weather alerts. However, most of this equipment was not originally designed to face modern cyber threats. Ground stations connected to the cloud can be breached, GPS signals can be jammed, and some components in the supply chain may include hidden malware. There is still no shared list of known vulnerabilities and no secure environment for testing spacerelated cyber defense. This paper identifies eleven key research gaps. These include secure routing, onboard attack detection, automated recovery, trusted supply chains, post-quantum encryption, zero-trust implementation, and real-time service impact monitoring. For each topic, we describe the challenge, explain why it matters, and propose a guiding research question. We also explore how a team of small, task-specific artificial intelligence agents, what we call an agentic (multi-agent) approach, could improve onboard defense without relying on large hardware.The paper ends with a proposed five-year roadmap. This includes flight tests of post-quantum and quantum key distribution (QKD) links, open-access cyber-ranges for practical testing, better rules for sharing vulnerabilities, and initial deployments of multi-agent security on operational spacecraft. Moving toward these proactive, modular defenses will help prevent outages like the KA-SAT incident.

The Onion Router (Tor) relies on path selection algorithms to balance performance and anonymity by determining how traffic flows through its relay network. As Tor scales and usage patterns evolve, default strategies such as bandwidth-weighted random selection and persistent guard nodes face increasing performance limitations. This study presents a comparative evaluation of five path selection strategies: Random, Guard, Congestion-Aware, and two Geographic approaches (Diversity Driven and Latency-Optimized), herein referred to as Geo-Latency and Geo-Diversity, using a high-fidelity simulation model inspired by TorPS (Tor Path Simulator). Experiments were conducted across five network scales, simulating 37,500 circuits under realistic relay conditions. Results show that Geographic (Latency-Optimized) consistently achieved the lowest latency $(40.0 \mathrm{~ms})$ and highest efficiency, while Congestion-Aware strategies delivered the best throughput, outperforming the baseline by up to $\mathbf{4 2 \%}$. Guard nodes maintained stable routing but exhibited latency increases under larger networks. No single method proved optimal across all scenarios, but each revealed clear strengths for specific use cases. These findings demonstrate that targeted path selection can significantly improve Tor’s performance without compromising anonymity, providing guidance for optimizing circuit construction in future development and deployments.

Evaluation of recommendation systems continues evolving, especially in recent years. There have been several attempts to standardize the assessment processes and propose replacement metrics better oriented toward measuring effective personalization. However, standard evaluation tools merely possess the capacity to provide a general overview of a system’s performance; they lack consistency and effectiveness in their use, as evidenced by most recent studies on the topic. Furthermore, traditional evaluation techniques fail to detect potentially harmful data on small subsets. Moreover, they generally lack explainable features to interpret how such minor variations could affect the system’s performance. This proposal focuses on data clustering for recommender evaluation and applies a cluster assessment technique to locate such performance issues. Our new approach, namedgroup validation, aids in spotting critical performance variability in compact subsets of the system’s data and unravels hidden weaknesses in predictions where such unfavorable variations generally go unnoticed with typical assessment methods. Group validation for recommenders is a modular evaluation layer that complements regular evaluation and includes a new unique perspective to the evaluation process. Additionally, it allows several applications to the recommender ecosystem, such as model evolution tests, fraud/attack detection, and the capacity for hosting a hybrid model setup.

This paper presents a heterogeneous federated ensemble model for intrusion detection system, employing a semisupervised novelty detection technique - the baseline K-means. The technique learns normal traffic from baseline data and utilizes the Mahalanobis distance to detect anomalous packets. To mitigate the false-positive rate inherent in anomaly-based intrusion detection system, we propose an ensemble approach that integrates local novelty detection models dedicated to each worker in both weighed and voting-based strategies. The federated design augments each worker’s detection capability without increasing the false positive rate. Our extensive experiments showcase the system’s robustness and adaptability over traditional standalone IDS, with marked improvements in precision, recall, and F1score under varying sampling rates. We made this project’s code publicly available on Github for replicability.

Understanding user behavior in the context of recommender systems remains challenging for researchers and practitioners. Inconsistent and misleading user information, which is often concealed in datasets, can inevitably shape the recommendation results in certain distorted ways despite utilizing recommender models with enhanced personalizing capabilities. Naturally, the quality of data that fuels those recommenders should be extremely reliable and free of any biases that might be invisible to a model, irrespective of its type. In this article, we introduce two modern forms of noise that are intrinsically hard to detect and eliminate; one is malicious in nature and will be termed Burst while the other is unique in that it forms its own category and will be referred to as Opt-out. Additionally, with the aim of segregating the nature of noise behind such threats, we present a distinct case study on Burst and Opt-out to illustrate how the detection of those threats can be challenging compared to that of traditional noise and with the current detection methods. Finally, we expound on the ability of such threats to bias the output of recommenders in their own unique way while primarily retaining data that is not fundamentally erroneous.

Recommender systems have been upgraded, tested, and applied in many, often incomparable ways. In attempts to diligently understand user behavior in certain environments, those systems have been frequently utilized in domains like e-commerce, e-learning, and tourism. Their increasing need and popularity have allowed the existence of numerous research paths on major issues like data sparsity, cold start, malicious noise, and natural noise, which immensely limit their performance. It is typical that the quality of the data that fuel those systems should be extremely reliable. Inconsistent user information in datasets can alter the performance of recommenders, albeit running advanced personalizing algorithms. The consequences of this can be costly as such systems are employed in abundant online businesses. Successfully managing these inconsistencies results in more personalized user experiences. In this article, the previous works conducted on natural noise management in recommender datasets are thoroughly analyzed. We adequately explore the ways in which the proposed methods measure improved performances and touch on the different natural noise management techniques and the attributes of the solutions. Additionally, we test the evaluation methods employed to assess the approaches and discuss several key gaps and other improvements the field should realize in the future. Our work considers the likelihood of a modern research branch on natural noise management and recommender assessment.

Recommender systems are nowadays widely implemented in order to predict the potential objects of interest for the user. With the wide world of the internet, these systems are necessary to limit the problem of information overload and make the user’s internet surfing a more agreeable experience. However, a very accurate recommender system creates a problem of over-personalization where there is no place for adventure and unexpected discoveries: the user will be trapped in filter bubbles and echo rooms. Serendipity is a beneficial discovery that happens by accident. Used alone, serendipity can be easily confused with randomness; this takes us back to the original problem of information overload. Hypothetically, combining accurate and serendipitous recommendations will result in a higher user satisfaction. The aim of this paper is to prove the following concept: including some serendipity at the cost of profile accuracy will result in a higher user satisfaction and is, therefore, more favourable to implement. We will be testing a first measure implementation of serendipity on an offline dataset that lacks serendipity implementation. By varying the ratio of accuracy and serendipity in the recommendation list, we will reach the optimal number of serendipitous recommendations to be included in an accurate list.

In collaborative filtering recommender systems, the users rate items, and this process helps in understanding their preferences. The systems can suffer from the cold-start problem, which refers to the absence or insufficiency of ratings for new users. This can be solved by using active learning strategies, which can be non-personalized or personalized, and which were evaluated and tested previously using different datasets and metrics. In this paper, we present a clearer study by implementing the main non-personalized single-heuristic strategies (random, popularity, co—coverage, variance, entropy, entropy0) on the same dataset, and by evaluating them using the same metrics, in order to have a better comparison. We use the public MovieLens dataset in the experimentations and the results show that the random strategy performs the worst, whereas the entropy0 leads to the best results. All strategies except the random strategy lead to very close results at a certain point, where ratings for almost the same items will have been elicited.

The rapid development of medical sensors has increased the interest in Wireless Body Area Network (WBAN) applications where physiological data from the human body and its environment is gathered, monitored, and analyzed to take the proper measures. In WBANs, it is essential to design MAC protocols that ensure adequate Quality of Service (QoS) such as low delay and high scalability. This paper investigates Medium Access Control (MAC) protocols used in WBAN, and compares their performance in a high traffic environment. Such scenario can be induced in case of emergency for example, where physiological data collected from all sensors on human body should be sent simultaneously to take appropriate action. This study can also be extended to cover collaborative WBAN systems where information from different bodies is sent simultaneously leading to high traffic. OPNET simulations are performed to compare the delay and scalability performance of the different MAC protocols under the same experimental conditions and to draw conclusions about the best protocol to be used in a high traffic environment.

Two Radio Resource Control states have been proposed in LTE and implemented to ensure low UE power consumption and high network resource availability. Transiting between these two states optimizes network performance if tuned properly. Currently, a UE switches from the LTE_ACTIVE state to the LTE_IDLE state after a pre-configured static inactivity duration. This paper seeks to demonstrate that no static timeout is optimal for all users at all times. In addition, a user-level dynamic decision algorithm is proposed to have fine-grain user level optimization. Since achieving better efficiency is related to context awareness, we present a solution that allows the UE to auto-learn its traffic behavior. The dynamic algorithm was applied to five different user load scenarios of combined application and legacy traffic, and the results showed that we are able to attain power savings of up to 30% when compared to the fixed timeout case.

Researchers agree on a basic, general and abstract definition that mobile cloud computing is a combination of mobile and cloud computing. Nevertheless, the location of the computation power is still controversial and varies from centralized servers located somewhere in the internet to a collection of mobile devices interested in processing certain tasks. Interestingly, all mobile cloud architectures tried to meet the expectations of mobile cloud through application layer solutions. Although of its innovation, mobile networks' physical layer is not optimized for similar applications, thus it affects the system's overall performance and cause mobile cloud architectures to fail in satisfying its goals such as increase in delay, increase in battery consumption etc. In this paper, we will discuss mobile cloud from telecommunication perspective to harmonize the physical layer of mobile networks (LTE and UMTS) with the application layer by proposing an innovative architecture. This new architecture keeps the mobile operator at the core of mobile cloud computing and offers revenue-making business model that motivates the operator to invest in this technology.