The Internet of Things (IoT) enables large-scale distributed sensing, improving automation and real-time decision-making across domains such as smart cities and environmental monitoring. However, as IoT deployments increase in scale and complexity, ensuring the reliability of sensor nodes becomes increasingly challenging. A key issue is the timely and resource-efficient detection of faulty behavior in IoT nodes. Most existing solutions focus on a limited set of fault types, such as communication failures or sensor data anomalies, leaving many faults undetected. In this article, we propose VarLogger, a lightweight runtime anomaly detection method. Our key observation is that many IoT faults manifest as sub-normal software behavior, such as delayed function responses or repeated hardware access. VarLogger therefore monitors device-internal software event traces and their temporal and spatial properties. We design and evaluate two unsupervised learning-based methods to capture normal event traces and detect sub-normal behavior directly on the edge. Evaluation using real-world use cases and diverse fault types (software, communication, and others) demonstrates that the time-based method outperforms state-of-the-art approaches. VarLogger achieves an F1 score above 0.9, while memory and CPU analyses confirm its suitability for resource-constrained devices. Most importantly, VarLogger can detect previously unknown faults in a real-time and resource-efficient manner.

The introduction of Collective Perception (CP) services in Vehicular Ad-Hoc Networks (VANETs) aimed to enhance safety by increasing awareness of both V2X and non-V2X objects. Recent research confirms that CP services can significantly boost awareness. However, CP services come with several drawbacks, including the need for additional channel capacity to transmit generated CP messages (CPMs) and the required extra processing power at the CPM receiving vehicle to process all the received CPM objects. While numerous studies address the former drawback, there is a gap in research concerning the latter challenge. This study addresses that gap by proposing a novel, lightweight approach to optimize CPM processing through intelligent filtering of received data. At the core of our approach is a lightweight yet effective utility function, designed to identify which CPM objects are relevant in a given driving context. Our analysis shows that the majority of received CPM objects are not utilized by the receiving vehicle, indicating considerable potential for optimization. Evaluation results demonstrate that the proposed utility function can filter out up to 74% of CPM objects in some scenarios, while the resulting reduction in useful situational awareness remains as low as 0.031% illustrating the effectiveness of the proposed method in conserving computational resources without significantly compromising safety.

This paper presents a robust and dependable monitoring system designed for space missions where reliability is critical. Theoretically, we use the Markov chain model to evaluate the availability of the system, considering multiple instances of backup devices. The results indicate a substantial improvement in system availability with the inclusion of backup devices. The findings highlight the significance of redundancy in enhancing system dependability while considering practical constraints. Based on this insight, we design an algorithm to deploy backup devices in the system that would improve the system's availability without putting strain on the resources (e.g., channel congestion). It uses backup devices at the sensor node and gateway levels. The algorithm involves backup devices that actively monitor and take over in case of failures, enhancing the system's ability to recover lost or faulty data. We test the system in three failure scenarios: hard failures, sensor failures, and gateway failures. We perform the experiments on our hardware prototypes. Experimental results demonstrate that using backup devices significantly improves the Packet Reception Ratio (PRR) in failure scenarios. We observed an improvement of 30 - 34% in packet reception for different failure scenarios.

Vehicular Ad-Hoc Networks (VANETs) were pri-marily introduced to enhance vehicular safety by facilitating communication between vehicles and infrastructure, improving overall situational awareness. In VANETs, vehicles are expected to exchange numerous messages generated by various applications, which can be categorized into safety and non-safety applications. The primary communication technologies for VANETs, Dedicated Short Range Communication (DSRC) and Cellular V2X (C- V2X), are mainly focused on delay-sensitive, safety-related applications. However, using the same bandwidth for both safety and non-safety applications can increase the load on the communication channel, potentially leading to higher latencies. This study examines the feasibility of using LoRa communication for non-safety-related urban VANET applications. Due to the long-range communication capability, we believe LoRa is ideal for VANET applications requiring low data rates and wider coverage. We conducted multiple real-world experiments to evaluate the performance of LoRa in different urban VANET scenarios. Our results indicate that LoRa can handle Doppler shifts caused by urban VANET speeds with both Spreading Factor (SF) 7 and 12, though higher SF is more susceptible to Doppler shifts. Additionally, our findings show that the Line-of-Sight (LoS) condition significantly impacts LoRa communication, especially with lower SF.

To go with the current trend of using smart environments or Internet of Things (IoT) and to make the human life easier, many systems have been developed up to date which use special purpose sensing or direct sensing approaches. However, these approaches are expensive due to the need to affix sensors on each and every equipment and also they lack flexibility of usage due to being battery powered. As a solution for these issues, based on the concept of synthetic sensors introduced by a previous research work, a general purpose sensing approach is discussed by the present study. The system can be placed in a corner of a room and will only be taking a maximum of three sensors into account to detect many events happening across the room making it rather cost-effective. Furthermore, performance of several machine learning (ML) algorithms is analysed by this work. With real time testing of the proposed system, the possibility of applying this approach as a cost-effective solution to be used in the day-to-day work is proven.

Creating new lectures and aligning the contents to the students and their knowledge can be a challenging work for all levels of education: Knowledge dependencies between the different blocks of lectures are not always visible. The problem increases if content from third parties like colleagues or from the Internet is being used. The content curation tool described in this poster aims on this challenge and offers a way of automatic, skill-based lecture creation and dependency management. The objective is that the lecturer can enter the required skills as well as the type of teaching materials and the system will generate a complete course from available contents.

Modeling mobility is a key aspect when simulating different types of networks. To cater to this requirement, a large number of models has emerged in the last years. They are typically (a) trace-based, where GPS recordings are re-run in simulation, (b) synthetic models, which describe mobility with formal methods, or (c) hybrid models, which are synthetic models based on statistically evaluated traces. All these families of models have advantages and disadvantages. For example, trace-based models are very inflexible in terms of simulation scenarios, but have realistic behaviour, while synthetic models are very flexible, but lack realism. In this paper, we propose a new mobility model, called TRAILS (TRAce-based ProbabILiStic Mobility Model), which bridges the gap between these families and combines their advantages into a single model. The main idea is to extract a mobility graph from real traces and to use it in simulation to create scalable, flexible simulation scenarios. We show that TRAILS is more realistic than synthetic models, while achieving full scalability and flexibility. We have implemented and evaluated TRAILS in the OMNeT++ simulation framework.

Volumetric Water Content (VWC) is used for determining the field capacity of any soil. Sensor nodes equipped with VWC and temperature sensor are deployed underground to understand the properties of soil for agricultural activities. The cost of VWC sensors becomes the bottleneck in the deployment of these sensors over a large field. In this paper, we analyze the use of low-cost moisture and temperature sensors to estimate the VWC values and field capacity of any field using machine learning techniques i.e. Artificial Neural Network and Random Forests.

This paper focuses on modeling the charateristics of opportunistic networks (OppNets). The motivation of this work is to develop a model which is closer to the reality to represent the data dissemination in OppNets. This paper shows our preliminary results validating our model in OppNets scenarios which give an opportunity to exchange data among a group of people.

The Zigbee low-power communication standard has established itself as one of the most important wireless standards, enabling thousands of industrial and environmental monitoring applications. At the same time, Bluetooth and newly also Bluetooth Low Energy has captured the gadget and smartphone markets and currently enables various health and personal applications. The border between these two markets becomes thinner and applications would profit significantly from interconnecting these two standards and sharing the information obtained.

Wireless sensor networks (WSNs) penetrated the market mainly as solutions for specific application scenarios. However, this strong specialization limits WSNs reuse both in terms of development as well as in terms of technical results: Every new application scenario requires a new design, development and validation, as well as management skills. This is frustrating for any WSNs user, developer or manager. To reverse this tendency and thus improve the quality of experience and user-friendliness in WSNs, we designed FLEXOR, a sustainable software architecture optimized to support the implementation, rapid prototyping, evaluation, and testing of wireless sensor network applications, that is platform independent and user-friendly. FLEXOR is designed to accommodate many different applications and services for wireless sensor networks and foster code re-usability and cross-platform component re-usability. FLEXOR offers high modularity, well defined interfaces, remote node management functionality as well as run-time module exchange. Finally, the introduction of a unifying way for WSNs development opens to a higher homogeneity and thus to more easy comparison among different solutions. We present here an analysis of FLEXOR from these new angles and show how effective it is for several purposes and in particular for non-experts and in education.

Teaching wireless sensor networks (WSNs) at the undergraduate level is both challenging and rewarding. WSNs include low-level programming and debugging, power-aware operations, novel communication paradigms and interdisciplinary applications in real world environments. These characteristics enable the students to learn how to cope with complete systems, starting with hardware design and low-level programming throughout applications and data processing. They learn to recognize complex software-hardware cross-dependencies and to manage them successfully.In this paper we share our experience from a new course on Introduction to Wireless Sensor Networks at the undergraduate level. The most important characteristic of the course is its hands-on approach. We believe this is crucial for fully understanding the complex nature of WSNs and their interdependencies with the environment in which they operate. The students were able to build a solid set of skills and expertise in WSNs by designing and implementing their own small-size real-world WSN deployment.

Clustering and aggregation inherently increase wireless sensor network (WSN) lifetime by collecting information within a cluster at a cluster head, reducing the amount of data through computation, then forwarding it. Traditional approaches, however, both spend extensive communication energy to identify the cluster heads and are inflexible to network dynamics such as those arising from sink mobility, node failure, or dwindling battery reserves. This paper presents CLIQUE, an approach for data clustering that saves cluster head selection energy by using machine learning to enable nodes to independently decide whether or not to act as a cluster head on a per-packet basis. We refer to this lack of actual cluster head assignment as being role-free, and demonstrate through simulations that, when combined with learning dynamic network properties such as battery reserves, up to 25% less energy is consumed in comparison to a traditional, random cluster head selection approach.

The rapid development of wireless sensor networks (WSNs) and the increasing complexity of the deployed applications force researchers to turn to testbeds to test their communication and application related algorithms and protocols. On the other side, industries' demand for young professionals with WSN expertise is growing too. These two factors motivated us to develop a highly modular and extensible testbed, called Atelier, as part of our bachelor level course on WSNs. In this paper we share our testbed design and describe its main requirements and challenges.

Efficient multi-hop data dissemination is a crucial building block to enable mature wireless sensor network (WSN) applications. Exploiting machine learning for these routing problems has received increasing attention in recent years due to its flexibility and localized mechanisms. However, with such an approach the resulting protocols often have additional memory and processing time requirements. Nevertheless, these requirements \emph {are} within the reach of today's WSN hardware, however few substantial tests have been performed to clearly demonstrate this. This paper evaluates and discusses the results and experiences gained from implementing our reinforcement learning based multicast routing protocol (FROMS) in a testbed of ScatterWeb nodes. A comparison of our results is made to a well-known WSN routing scheme, namely a multicast variation of Directed Diffusion. Our evaluation includes several minor, but practical modifications to both protocols such as transmission backoffs and the use of acknowledgments.This paper offers three main contributions. First, we demonstrate that machine learning algorithms \emph{can be} efficiently implemented on resource restricted devices and that they perform very well in multiple network scenarios. Second, we confirm the validity of simulation results obtained in a previous evaluation of FROMS, and at the same time gather delivery rates under realistic settings. Finally, we offer some general observations on properties and pitfalls of WSN implementations along with potential solutions.

Efficiently moving sensor data from its collection to use points is both the fundamental and the most difficult challenge in wireless sensor networks, as any data movement incurs cost. In this work, we focus on routing data to multiple, possibly mobile sinks. To deal with the dynamics of the environment arising from mobility and failures, we choose a reinforcement learning approach where neighboring nodes exchange small amounts of information allowing them to learn the next, best hop to reach all sinks. Preliminary evaluation demonstrates that our technique results in low cost routes with low overhead for the learning process.