Abstract
Decision trees are often preferred when implementing Machine Learning in embedded systems for their simplicity and scalability. Hoeffding Trees are a type of Decision Trees that take advantage of the Hoeffding Bound to allow them to learn patterns in data without having to continuously store the data samples for future reprocessing. This makes them especially suitable for deployment on embedded devices. In this work we highlight the features of a HLS implementation of the Hoeffding Tree. The implementation parameters include the feature size of the samples (D), the number of output classes (K), and the maximum number of nodes to which the tree is allowed to grow (Nd). We target a Xilinx MPSoC ZCU102, and evaluate: the design's resource requirements and clock frequency for different numbers of classes and feature size, the execution time on several synthetic datasets of varying sizes (N) and the execution time and accuracy for two datasets from UCI. For a problem size of D=3, K=5, and N=40000, a single decision tree operating at 103MHz is capable of 8.3x faster inference than the 1.2 GHz ARM Cortex-A53 core. Compared to a reference implementation of the Hoeffding tree, we achieve comparable classification accuracy for the UCI datasets.

Abstract
Premature ventricular contractions (PVC), or extrasystoles, represent a type of cardiac arrhythmia that is common among the general population and, notably, among athletes or individuals who exercise frequently. PVC may be asymptomatic and not clinically relevant when their rate is low, up to around 0.5%, or may be symptomatic and clinically relevant when it is high, in the order of or above 10%. ECG analysis in association with a cardiac stress test is important to detect and characterize PVC and to diagnose the heart condition and operation. In this paper, we describe and test a simple signal processing approach that can be used to effectively pinpoint subtle PVC occurrences in various physical effort conditions. In this regard, we discuss i) three important conditions to be met such that PVC are categorized as benign, ii) the design and implementation of a cardiac stress test and ECG data collection, iii) the algorithm analyzing and extracting information from the detected PVC occurrences, and iv) we present and discuss the obtained results, and conclude on their significance. © 2022 IEEE.

Abstract
The fifth generation of mobile communications networks (5G) emerges with the potential to customize the technical parameters of the same physical infrastructure for each application, service, or user, which can compromise the fundamentals that made the Internet the leading platform for dissemi-nating information and a transnational instrument of collaboration of indi-viduals and institutions. In this scenario, the present study intends to ana-lyze this new technological standard, its influence on the informational flow of the Internet, and evaluate the role of information policy for the gover-nance of the multiple interests that permeate the digital ecosystem.

Abstract
Unmanned Aerial Vehicles (UAVs) acting as Flying Access Points (FAPs) are being used to provide on-demand wireless connectivity in extreme scenarios. Despite ongoing research, the optimization of UAVs' positions according to dynamic users' traffic demands remains challenging. We propose the Traffic-aware UAV Placement Algorithm (TUPA), which positions a UAV acting as FAP according to the users' traffic demands, in order to maximize the network utility. Using a DRL approach enables the FAP to autonomously learn and adapt to dynamic conditions and requirements of networking scenarios. Moreover, the proposed DRL methodology allows TUPA to generalize knowledge acquired during training to unknown combinations of users' positions and traffic demands, with no additional training. TUPA is trained and evaluated using network simulator ns-3 and ns3-gym framework. The results demonstrate that TUPA increases the network utility, compared to baseline solutions, increasing the average network utility up to 4x in scenarios with heterogeneous traffic demands.

Abstract
5G has become increasingly popular nowadays, mainly due to its characteristics which enable high data rates and low latency. At the same time, mobile robotic platforms, such as drones and robots, appeared as suitable platforms to carry radio stations, enabling the on-demand placement of 5G communications cells. The main contribution of this paper is an obstacle-aware on-demand 5G network. The proposed solution consists of a 5G radio station (gNB) carried by a mobile robotic platform capable of providing obstacle-aware wireless connectivity to 5G User Equipments (UEs), leveraged by a novel virtual network function - On-Demand Mobility Management Function (ODMMF). ODMMF is designed to integrate the 5G Core network and it allows to monitor the radio conditions provided to the served UEs, while enabling the positioning of the mobile robotic platform remotely by taking advantage of the visual information provided by on-board video cameras. The proposed solution was validated using an experimental prototype, under a representative networking scenario.

Abstract
The ability to move and hover has made rotary-wing unmanned aerial vehicles (UAVs) suitable platforms to act as flying communications relays (FCRs), aiming at providing on-demand, temporary wireless connectivity when there is no network infrastructure available or a need to reinforce the capacity of existing networks. However, since UAVs rely on their on-board batteries, which can be drained quickly, they typically need to land frequently for recharging or replacing them, limiting their endurance and the flying network availability. The problem is exacerbated when a single FCR UAV is used. The FCR UAV energy is used for two main tasks: Communications and propulsion. The literature has been focused on optimizing both the flying network performance and energy efficiency from the communications point of view, overlooking the energy spent for the UAV propulsion. Yet, the energy spent for communications is typically negligible when compared with the energy spent for the UAV propulsion. In this article, we propose energy-aware relay positioning (EREP), an algorithm for positioning the FCR taking into account the energy spent for the UAV propulsion. Building upon the conclusion that hovering is not the most energy-efficient state, EREP defines the trajectory and speed that minimize the energy spent by the FCR UAV on propulsion, without compromising in practice the quality of service offered by the flying network. The EREP algorithm is evaluated using simulations. The obtained results show gains up to 26% in the FCR UAV endurance for negligible throughput and delay degradation.