Location data has become central to many services, requiring effective modeling of movement patterns and detection of irregular behaviors. This paper presents a system for behavior modeling and anomaly detection based on georeferenced movement data, evaluated in security scenarios. Points of interest are identified using an adapted Time- and Distance-Based Clustering (TDBC) algorithm, and behavior patterns are extracted to build a historical model. Deviations from this model are then detected through statistical inference, with anomaly types and degrees defined by a dedicated taxonomy. Evaluation in a synthetic drug investigation scenario shows that the proposed clustering method outperforms classical density-based approaches for moderate and high-frequency data, while anomaly detection accuracy improves with larger historical datasets and clearly defined anomalies, yet remains effective in less obvious cases.

Abstract Object detection in military operations faces critical challenges, including camouflaged targets and occlusion, where traditional RGB-based systems often fail. This study proposes a systematic framework for optimizing multimodal late fusion in object detection by integrating color space transformations with depth information. We contribute three key elements: (1) the development of the “militar-VALID” dataset, a specialized, defense-oriented collection of 6,054 images curated for challenging detection scenarios; (2) a comprehensive statistical evaluation framework comparing four late-fusion algorithms across 247 unique configurations; and (3) the hyperparameter optimization of the optimal fusion configuration through Bayesian search. Leveraging the YOLOv8-small architecture trained on eight parallel color representations (RGB, BGR, Grayscale, HSV, CIELab, YUV, YCrCb, and Depth), we establish that Weighted Boxes Fusion combining RGB, Depth, and HSV modalities delivers statistically significant improvements ( $$p \le 0.001$$ , Cliff’s $$\delta \ge 0.8$$ ). Specifically, this configuration achieves a 1.30% increase in mean Average Precision (mAP@50-95), a 3.20% improvement in precision, and a 3.22% enhancement in Average Precision (AP@50) for small objects compared to RGB-only baselines. Statistical analysis highlights the depth maps as the most impactful modality and HSV as the optimal color space complement. This work provides a quantitative framework for multimodal color space fusion in military object detection and discusses its implications for deployment in high-stakes operational environments.

Critical infrastructures face demanding challenges due to natural and human-generated threats, such as pandemics, workforce shortages or cyber-attacks, which might severely compromise service quality. To improve system resilience, decision-makers would need intelligent tools for quick and efficient resource allocation. This article explores an agent-based simulation model that intends to capture a part of the complexity of critical infrastructure systems, particularly considering the interdependencies of healthcare systems with information and telecommunication systems. Such a model enables to implement a simulation-based optimization approach in which the exposure of critical systems to risks is evaluated, while comparing the mitigation effects of multiple tactical and strategical decision alternatives to enhance their resilience. The proposed model is designed to be parameterizable, to enable adapting it to risk scenarios with different severity, and it facilitates the compilation of relevant performance indicators enabling monitoring at both agent level and system level. To validate the agent-based model, a literature-supported methodology has been used to perform cross-validation, sensitivity analysis and test the usefulness of the proposed model through a use case. The use case analyzes the impact of a concurrent pandemic and a cyber-attack on a hospital and compares different resiliency-enhancing countermeasures using contingency tables. Overall, the use case illustrates the feasibility and versatility of the proposed approach to enhance resiliency.

Information Processing and Telecommunications Center, Universidad Politécnica de Madrid, [email protected] Wearable (headset-enabled) Augmented Reality has great potential to facilitate understanding and interaction with digital data-driven scenes. In this article, a design concept for a monitoring service addressing the needs of a specific data-intensive use case, building occupancy monitoring, is presented. From specific system requirements, a set of guidelines grounded on previous literature are identified. On these, the resulting layout and implementation is presented, to substantiate the final wearable-AR service. The guidelines presented enable to identify and organize visualization and interactive components, such as the map and the spherical notifications along with the notification stack.

In this paper we address the adaptation of applications for conventional platforms (web, tablets and smartphones) to Augmented Reality (AR) settings. We do this on a specific use case, a tool to help users to define drone missions, which has been implemented with equivalent functionalities both for web-enabled devices and AR wearable devices (in particular, HoloLens). First, a brief description of the application workings is presented as well as the creation process for the mission and the manipulation of its components. Then, we present a preliminary validation with 8 users to analyze the degree of acceptance of the AR version, taking the web user interface as a baseline. Despite the technical limitations of the AR version (mainly related to the device visualization constraints and weight) and the reduction in efficiency (tasks need more time to be completed), several users have expressed their preference for this version. However, with the problem of precision in AR Head Mounted Display context always present, the need to establish new interaction techniques is once more highlighted.

Nowadays, the Bluetooth Low-Energy (BTLE) technology is integrated in numerous smartphones and beacons, but there is still an open challenge on delivering a presence-into-zone positioning system that facilitates the design and reconfiguration of the service zones and offers stable target tracking. In this paper, we report the experience gained from the deployment of a BTLE zone-based positioning system in a retail setting with defined service objectives. The algorithm powering the positioning system requires previous calibration, which comes to be a tedious and long invasive process when it is accomplished in a space with high people flow. The paper describes our work to minimize the calibration effort, by analyzing the effect of calibration samples reduction and proposing the application of a on-the-move calibration strategy that facilitates the calibration dynamics preserving the system's correct recognition rate.

Pointing is one of the more meaningful gestures in human expression. In this article, deictic and head pose pointing methods are evaluated as interaction means, with the objective of comparing their efficiency and delivered comfort, usability, and user experience to apply them to the deployment of services in smart spaces. Standard tools and questionnaires are applied in a user study with 20 individuals who have tried a Kinect V2-enabled pointing system to a) complete a multidirectional International Organization for Standardization (ISO) 9241-9 pointing task and b) test a service to command smart objects. The results show that both deictic and head pose interaction perform in a similar way, providing a throughput of 2.14 and 2.04 bits/s, respectively (comparable to joysticks or touchpads in equivalent ISO tasks). The system being quite robust to the user position in the Kinect coverage area, target size influences performance, but also learnability and the overall response when cursor visual cues are not available. From comfort, usability, and user experience responses, it can be said that deictic is perceived to perform better than head pose in terms of accuracy, efficiency, and organization, although head pose is preferred for its speed.

The estimation of the 3D position and orientation of moving objects (‘pose’ estimation) is a critical process for many applications in robotics, computer vision or mobile services. Although major research efforts have been carried out to design accurate, fast and robust indoor pose estimation systems, it remains as an open challenge to provide a low-cost, easy to deploy and reliable solution. Addressing this issue, this paper describes a hybrid approach for 6 degrees of freedom (6-DoF) pose estimation that fuses acceleration data and stereo vision to overcome the respective weaknesses of single technology approaches. The system relies on COTS technologies (standard webcams, accelerometers) and printable colored markers. It uses a set of infrastructure cameras, located to have the object to be tracked visible most of the operation time; the target object has to include an embedded accelerometer and be tagged with a fiducial marker. This simple marker has been designed for easy detection and segmentation and it may be adapted to different service scenarios (in shape and colors). Experimental results show that the proposed system provides high accuracy, while satisfactorily dealing with the real-time constraints.

Gestures contain enough information to distinguish a person from another. This `biometric' feature is the basis of the Kinect-enabled system described in this paper, which is designed to identify a given person through a short chain of in-air gestures (2-3 ones). The user-dependent (trainable) system relies on a traditional Dynamic Time Warping classification algorithm. This core algorithm enhances its performance thanks to two filters (duration and sensitivity) that make the system more robust against the variability of the gestures from the same user among iterations (e.g. gestures may differ in duration and shape). The system includes a `point-based' score strategy designed to rapidly discard the non-probable users and to speed up identity confirmation among the most probable candidates. Additionally, the system includes a specific component to guarantee the training quality. In order to integrate the identification process into a real service, the gesture chain is retrieved through a sequence of double-option questions that aim at better adapting a smart space (the lighting infrastructure and the available multimedia contents) to a given user depending on his specific needs. This way, the identification task remains in second plane, wrapped in functional questions. Ten users have tested the service, providing their positive feedback.

This paper describes an improved version of our intent-related hierarchy of formal languages designed to describe aircraft trajectories. These languages allow a complete or partial specification of aircraft trajectories at different levels, and the extension of its features described in this paper allows their applicability to define more complex kind of missions, such as those of unmanned vehicles or military aircraft. This paper provides a complete description on lexicon, syntax, and graphical representation details for each language and shows their applicability through a set of clarifying examples of flight specifications with different granularity. The described language hierarchy has been proven as a suitable framework for describing aircraft trajectories with different levels of detail and for different applications. Its versatility and flexibility are demonstrated through a set of scenarios identifying characteristic operational examples.

Nowadays, mobile and sensing technologies may facilitate ubiquitous disease prevention, making the user aware of his health status and providing useful tips to change non-adequate habits. This paper presents the FLES framework, a tool designed to simplify the deployment of medical check-ups and specific risk assessment components to prevent different diseases (e.g. ictus, heart disease, obesity, Type II diabetes, osteoporosis, mental problems, etc.). The platform enables easy integration of wireless health sensors, Health Parameter Control components and Disease Risk Assessment components, ready to be deployed in Android-equipped devices (e.g. smartphones, tablets, smart TVs or mini-computers). This enables to easily set home-oriented and on-the-move deployments, that may adapt to different user settings (e.g. a person in his daily routines, an elderly at home or a traveller in a hotel). FLES permits end-user in-session data control, providing full control of personal health data after each medical check-up.

This paper presents a mobile-instrumented Platform to interact with smart objects by using gestures. The system relies on a Dynamic Time Warping algorithm, which serve to train personalized gestures and also to recognize them real-time. The trained set of gestures is available for consumer applications to include this type of interaction in their functionalities. In particular, we describe a consumer application that enables to configure and orchestrate different controllable smart objects. Each object provides a list of configurable actions that it can perform, thus the user may match each action with his preferred gesture. In execution time, a user may control the desired object just by pressing the object name; additionally, the user can retrieve the existing configuration, reconfigure it and train new gestures. The result is an easy-to-scale system, in which developers may include new object modules.

This paper presents a light inference system for context management, ready to perform reasoning tasks in resource-constrained devices. The inference system is part of a service-oriented mobile software framework which runs on Javaenabled handheld devices. This framework serves to facilitate the creation of context-aware applications, as it decouples sensor's data acquisition and context processing from the application logic itself. The inference architecture is composed by different modules, some of them encapsulate existing tools (ìJena and Bossam) that have been adapted to the mobile working environment. The (rule-based) reasoning is prepared to use a general ontology. Both applications and the framework's components may configure the query set in order to retrieve the information they need from the inference system. In the paper, a validation example shows how this process is done.

Due to the random characteristics of the indoor propagation channel, received signal strength-based localization systems usually need to be manually calibrated once and again to guarantee their best performance. Calibration processes are costly in terms of time and resources, so they should be eliminated or reduced to a minimum. In this direction, this paper presents an optimization algorithm to automatically calibrate a propagation channel model by using a Least Mean Squares technique: RSS samples gathered in a number of reference points (with known positions) are used by a LMS algorithm to calculate those values for the channel model's constants that minimize the error computed by a hyperbolic triangulation positioning algorithm. Preliminary results on simulated and real data show that the localization error in distance is effectively reduced after a number of training samples. The LMS algorithm's simplicity and its low computational and memory costs make it adequate to be used in real systems.

In this contribution, a real experience of rapid design and deployment of context-aware services for an exhibition hall is detailed. The prototype has been built on a combination of a commercial system (which has been customized and improved to satisfy the prototype needs) with an in-home developed context acquisition framework. In order to partially overcome device fragmentation issues, we have focused on the development of web-based context-aware applications. The whole system has been deployed from scratch under real constraints of time and environment. The objective has been to test the integration problems of context-aware systems, in order to infer some conclusions on what it is needed to generalize them.

The development of ambient intelligence (AmI) applications usually implies dealing with complex sensor access and context reasoning tasks, which may significantly slow down the application development cycle when vertically assumed. To face this issue, we present CASanDRA, a middleware which provides easily consumable context information about a given user and his environment, retrieving and fusing data from personal mobile devices and external sensors. The framework is built following a layered service oriented approach. The output data from every CASanDRA’s layer are fully accessible through semantic interfaces; this allows AmI applications to retrieve raw context features, aggregated context data and complex ‘ images of context’, depending on their information needs. Moreover, different query modes -subscription, event-based, continuous and on-demand- are available. The current ‘mobile-assisted’ version of CASanDRA is composed by a CASanDRA Server, developed on an applications container and hosting the system intelligence, and CASanDRA Lite, a mobile client bundling a set of sensor level acquisition services. How an AmI application may be effortlessly built on CASanDRA is described in the paper through the design of an ‘ Ambient Home Care Monitor’.

Ageing societies raise new social and economic challenges which are expected to be partially solved by the use of advanced technology. Ambient home care systems (AHCS) aims at alleviating both elderly and caregivers daily tasks, thanks to their control of environmental, health and physical sensors and their capability of reasoning on the acquired information. In this paper, we design and develop a prototype of an AHCS using and adapting an open software framework, the context toolkit (CTK). The CTK here is customized to handle our proper acquisition platform. On our work, we infer some conclusions about design practices and context management for AHCS.

In this paper we consider a signal strength based location- aware network and evaluate how to optimize energy consumption while achieving a certain accuracy in the localization result. Obviously, the greater the accuracy, the greater energy will be consumed, and, although different strategies can yield similar localization accuracy, their energy efficiency can be different. The relationship between these parameters is evaluated and the optimum strategy is found with respect to the transmission power of the packets, the packet transmission rate, the signal to noise ratio at the receiver and the packet reception rate. From this analysis, some guidelines for the efficient management of energy for localization in an ad hoc or sensor network are extracted.

Energy efficient management and routing is one of the issues which affect the practical deployment of wireless ad hoc sensor networks, as it directly impacts network reliability and durability. Saving energy in a sensor network depends to a large extent on the percentage of time that its nodes are in the idle state, i.e. not transmitting or receiving. Thus power aware routing algorithms have to be used. In this paper we propose a low duty cycle Zone Routing-like communication scheme, conceived to efficiently manage transmission-reception resources of the nodes in path discovery and routing, to ultimately optimize energy consumption. The algorithm is based on the "social" cooperation of nodes in neighbourhoods guided by the heuristic objective of keeping as small as reasonably possible any transmission operation and avoiding the redundant sending of messages, while pursuing the route which locally minimizes the transaction delay

Energy efficient management is one of the issues which affect the practical deployment of wireless ad hoc sensor networks, as it directly impacts on network reliability and durability. In this paper we propose and analyse a power aware zone routing-like communication scheme to be used in ad hoc sensor networks. The algorithm is conceived to efficiently manage the transmission-reception resources in the network nodes during path discovery and routing. This will ultimately lead to optimizing the average energy consumption in the nodes. The algorithm is based on each node having a two-hop list of contacts and a simple distributed reasoning mechanism guided by two heuristic objectives: on the one hand, avoiding redundant sending of messages; on the other hand, cooperatively finding the route which locally minimizes the delay in completing the required communication transaction. We describe the basics of the algorithm, and include some performance figures in comparison with the "optimal" and with a pure "flooding" scheme, for a given reference network topology. Finally we include a preliminary qualitative discussion about the optimal size of the list of contacts as a function of the size of the network