The widespread adoption of Unmanned Aerial Vehicles (UAVs) has raised critical security and safety concerns, particularly in sensitive areas and air traffic management. Modern counter-drone systems integrate multiple sensing modalities, but their development is hindered by the lack of comprehensive, publicly available datasets. To address this, the Drone-vs-Bird Detection Grand Challenge provides a manually annotated UAV dataset to advance research in drone detection. Since its inception in 2017, the competition has attracted global interest, fostering the development of advanced detection methods. This paper presents an overview of the 8th edition as data competition hosted at the International Joint Conference on Neural Networks (IJCNN) 2025. The data competition generated high engagement with 16 competing algorithms successfully submitted. The variability of the results underscores the complexity of the task and the need for future research. Over almost a decade, this data competition has been bridging the domains of signal processing, computer vision, and deep learning, paving the way for next-generation counter-drone solutions.

Unmanned aerial vehicles (UAVs) combined with intelligent reflective surfaces (IRSs) represent a cutting-edge technology for improving the channel capacity of wireless communications, by capitalizing on UAVs’ 3-D mobility coupled with the IRSs’ smart radio capabilities. This work envisions a scenario in which a swarm of UAVs equipped with IRSs serves multiple Internet of Things (IoT) ground nodes (GNs) concurrently transmitting to a single base station (BS) via OFDMA. The huge number of passive elements composing the IRSs introduces a significant complexity in the mission design. Therefore, each IRS is divided into patches that can be simultaneously used to serve different nodes. Considering general Rician fading, a comprehensive channel model for IRS-assisted UAV-aided networks is derived. Then, a multiobjective mixed-integer nonlinear programming problem is conceived to maximize the sum-rate of the GNs and, at the same time, minimize the difference among the users’ data rates, by jointly optimizing the trajectories and the phase shift matrices. This nonconvex problem, reformulated in terms of scheduling (i.e., patch-GN assignment), is challenging to solve. Hence, it is rearranged as a Markov Decision Process and a quasi-optimal solution is obtained via Deep Reinforcement Learning. Extensive simulation analysis is performed to validate the results and the accuracy of the proposed model.

Characterization of the exceedance probability (EP) of the energy statistic (ES) plays a fundamental role in several signal processing applications, including radar (e.g., probability of false alarm) and communications (e.g., outage probability). However, manageable closed-form expressions are not available for general non-Gaussian models such as the$\eta$-$\mu$distribution. In this letter, simple formulas for predicting the EP of the ES are provided, based on second- and third-order cumulant series expansion of the tightest Chernoff bound, coupled with low-complexity approximations of Hoyt moments. Results show that the proposed method significantly improves over earlier work based on different bounds, and outperforms the asymptotic approximation via the central limit theorem as well as the Generalized Pareto Distribution fitting of the distribution tail.

This paper presents the 6th edition of the "Drone-vs-Bird" Detection Grand Challenge, organized within the 48th IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP). Taking video samples recorded by commercial RGB cameras as input, the challenge stimulates the design of advanced approaches capable of detecting the presence of small drones flying in a given area under surveillance. Successful methods should ensure high detection rates while limiting the number of false alarms erroneously triggered in presence of very similar false targets (birds). The paper summarizes the novel methods proposed by the four participating teams that succeeded in providing satisfactory detection performance on the 2023 challenge dataset.

We tackle the problem of estimating the spatial distribution of mobile phones from Mobile Network Operator (MNO) data, namely Call Detail Record (CDR) or signalling data. The process of transforming MNO data to a density map requires geolocating radio cells to determine their spatial footprint. Traditional geolocation solutions rely on Voronoi tes sellations and approximate cell footprints by mutually disjoint regions. Recently, some pioneering work started to consider more elaborate geolocation methods with partially overlapping (non- disjoint) cell footprints coupled with a probabilistic model for phone-to-cell association. Estimating the spatial density in such a probabilistic setup is currently an open research problem and is the focus of the present work. We start by reviewing three different estimation methods proposed in literature and provide novel analytical insights that unveil some key aspects of their mutual relationships and properties. Furthermore, we develop a novel estimation approach for which a closed-form solution can be given. Numerical results based on semi-synthetic data are presented to assess the relative accuracy of each method. Our results indicate that the estimators based on overlapping cells have the potential to improve spatial accuracy over traditional approaches based on Voronoi tessellations.

This paper presents the 4-th edition of the “drone-vs-bird” detection challenge, launched in conjunction with the the 17-th IEEE International Conference on Advanced Video and Signal-based Surveillance (AVSS). The objective of the challenge is to tackle the problem of detecting the presence of one or more drones in video scenes where birds may suddenly appear, taking into account some important effects such as the background and foreground motion. The proposed solutions should identify and localize drones in the scene only when they are actually present, without being confused by the presence of birds and the dynamic nature of the captured scenes. The paper illustrates the results of the challenge on the 2021 dataset, which has been further extended compared to the previous edition run in 2020.

MmWave multiple-input single-output (MISO) systems using a single-antenna receiver are regarded as a promising solution for the near future, before the full-fledged 5G MIMO will be widespread. However, for MISO systems synchronization cannot be performed jointly with user localization unless two-way transmissions are used. In this paper we show that thanks to the use of a reconfigurable intelligent surface (RIS), joint localization and synchronization is possible with only downlink MISO transmissions. The direct maximum likelihood (ML) estimator for the position and clock offset is derived. To obtain a good initialization for the ML optimization, a decoupled, relaxed estimator of position and delays is also devised, which does not require knowledge of the clock offset. Results show that the proposed approach attains the Cramér-Rao lower bound even for moderate values of the system parameters.

This paper proposes a decision scheme based on the$k$-nearest neighbors rule to detect coherent signals in non-Gaussian noise modeled as the sum of K-distributed clutter plus thermal noise. The analysis is conducted also on real data recordings and shows that the proposed detector can outperform natural competitors.

5G mmWave MIMO systems enable accurate estimation of the user position and mapping of the radio environment using a single snapshot when both the base station (BS) and user are equipped with large antenna arrays. However, massive arrays are initially expected only at the BS side, likely leaving users with one or very few antennas. In this paper, we propose a novel method for single-snapshot localization and mapping in the more challenging case of a user equipped with a single-antenna receiver. The joint maximum likelihood (ML) estimation problem is formulated and its solution formally derived. To avoid the burden of a full-dimensional search over the space of the unknown parameters, we present a novel practical approach that exploits the sparsity of mmWave channels to compute an approximate joint ML estimate. A thorough analysis, including the derivation of the Cramér-Rao lower bounds, reveals that accurate localization and mapping can be achieved also in a MISO setup even when the direct line-of-sight path between the BS and the user is severely attenuated.

The problem of robust radar detection is addressed from a machine learning inspired perspective. In particular, a novel interpretation of the well-known Kelly's and adaptive matched filter (AMF) detectors is provided in terms of decision region boundaries in a suitable feature space. Then, a new detector based on a feature vector that combines the two detection statistics is obtained by exploiting the k-nearest neighbors (KNN) approach. The resulting receiver possesses the constant false alarm rate (CFAR) property and can achieve the same benchmark performance of Kelly's detector under matched conditions while being almost as robust as the AMF (which instead experiences a loss under matched conditions).

The problem of position estimation of a mobile user equipped with a single antenna receiver using downlink transmissions is addressed. The advantages of this setup compared to the classical MIMO and uplink scenarios are analyzed in terms of achievable theoretical performance (Cramér-Rao bounds) considering a realistic power budget. Based on this analysis, a low-complexity two-step algorithm with improved localization performance is proposed, which first performs a (coarse) angle of departure estimation and then precodes the down-link signal to introduce beamforming towards the user direction. Results demonstrate that position estimation in downlink can be potentially much more accurate than in uplink, even in presence of multiple users in the system.

Cellular localization generally relies on time-difference-of-arrival (TDOA) measurements. In this paper, we investigate a novel scenario where the mobile user estimates its own position by jointly exploiting TDOA and angle of departure (AOD) measurements, which are estimated from downlink transmissions in a millimeter-wave (mmWave) multiple-input single-output (MISO) setup. We first perform a Fisher information analysis to derive the lower bounds on the estimation accuracy, and then propose a novel localization algorithm, which is able to provide improved performance also with few transmit antennas and limited bandwidth.

This paper presents the second edition of the “drone-vs-bird” detection challenge, launched within the activities of the 16-th IEEE International Conference on Advanced Video and Signal-based Surveillance (AVSS). The challenge's goal is to detect one or more drones appearing at some point in video sequences where birds may be also present, together with motion in background or foreground. Submitted algorithms should raise an alarm and provide a position estimate only when a drone is present, while not issuing alarms on birds, nor being confused by the rest of the scene. This paper reports on the challenge results on the 2019 dataset, which extends the first edition dataset provided by the SafeShore project with additional footage under different conditions.

A novel maximum likelihood trajectory estimation algorithm for targets in mixed stationary/moving conditions is presented. The proposed approach is able to estimate position and velocity of the target over arbitrary complex trajectories, while explicitly taking into account the possibility of stop&go motion. Moreover, a novel trajectory reconstruction method based on the theory of Bézier curve is developed for online smoothing of the trajectory, which keeps the advantages of Bayesian smoothing while introducing only a fixed lag in the estimation process. The performance assessment, conducted on both simulated and real data, shows that the proposed approach can outperform classical Kalman filter and Rauch-Tung-Striebel smoother techniques.

This paper addresses the problem of determining the unknown position of a mobile station for a mmWave multiple-input single-output (MISO) system. This setup is motivated by the fact that massive arrays will be initially implemented only on 5G base stations, likely leaving mobile stations with one antenna. The maximum likelihood solution to this problem is devised based on the time of flight and angle of departure of received downlink signals. While positioning in the uplink would rely on angle of arrival, it presents scalability limitations that are avoided in the downlink. To circumvent the multidimensional optimization of the optimal joint estimator, we propose two novel approaches amenable to practical implementation thanks to their reduced complexity. A thorough analysis, which includes the derivation of relevant Cramér-Rao lower bounds, shows that it is possible to achieve quasi-optimal performance even in presence of few transmissions, low signal-to-noise ratio (SNRs), and multipath propagation effects.

The limited localization capabilities provided by global navigation satellite systems (GNSS) is one of the main obstacles toward the development of reliable road safety applications in urban scenarios. In order to improve GNSS accuracy, a number of approaches have been proposed which exploit additional position-related information, for instance provided by local inertial sensors. However, such solutions cannot meet the very stringent accuracy requirements of safety applications, which call for advanced processing and the fusion of position-related signals and data from heterogeneous sources. In this paper, we aim at combining the potential of antenna array processing with a suitably-designed cooperation strategy that exploits vehicle-to-vehicle and vehicle-to-infrastructure communications. Particularly, we define a novel tracking algorithm with asynchronous updates triggered by beacon packet receptions, from which angle of arrival estimates are opportunistically obtained. A dynamic setting of relevant parameters allows the resulting cooperative positioning algorithm to adapt to the different operating conditions found in urban vehicular contexts. Simulation results under realistic environment conditions show that the proposed algorithm can achieve high position accuracy even in sparse scenarios, outperforming a natural competitor while keeping lightweight communication and low computational complexity.

Distance estimation, which arises in many applications and especially in range-based localization, is addressed for joint received signal strength (RSS) and time of arrival (TOA) data. A statistical characterization of the joint maximum likelihood estimator, which is unavailable in closed-form, is provided together with a full performance assessment in terms of the actual mean squared error (MSE), in order to establish when hybrid estimation is superior compared to RSS-only or TOA-only estimation. Furthermore, a novel closed-form estimator is proposed based on an ad-hoc relaxation of the likelihood function, which removes the need to adopt iterative methods for hybrid TOA/RSS ranging and strikes a better bias-variance tradeoff for improved performance. A thorough theoretical analysis, corroborated by numerical simulations, shows the effectiveness of the proposed approach, which outperforms state-of-the-art solutions.

Small drones are a rising threat due to their possible misuse for illegal activities, in particular smuggling and terrorism. The project SafeShore, funded by the European Commission under the Horizon 2020 program, has launched the “drone-vs-bird detection challenge” to address one of the many technical issues arising in this context. The goal is to detect a drone appearing at some point in a video where birds may be also present: the algorithm should raise an alarm and provide a position estimate only when a drone is present, while not issuing alarms on birds. This paper reports on the challenge proposal, evaluation, and results.

Motivated by safety applications in urban vehicular scenarios, where GPS does not typically provide the required positioning accuracy, a GPS-free localization technique that exploits vehicle-to-infrastructure communications is proposed. In particular, it provides for a vehicle to opportunistically use the beacon packets received from a roadside unit (RSU) in order to obtain estimates of their angle of arrival. Such estimates, together with the RSU's position information within beacon packets, are fed to a weighted least squares algorithm that aims at localizing the vehicle. The algorithm tries to take advantage of reliable measurements typically collected closer to the RSU-where a very high signal-to-noise ratio yields an accurate angular resolution-while keeping robustness against multipath phenomena. Simulation results show the effectiveness of the proposed technique.

This paper deals with estimation of gain and phase errors possibly present in frequency modulated continuous wave radars. In particular, the Cramér-Rao lower bound of unbiased estimators of gain and phase errors in presence of nuisance parameters is computed. It is used as a reference for the performance of already proposed estimators (computed by Monte Carlo simulation).

An Empirical Bayes formalization of the regularized covariance estimation problem is proposed for (possibly high-dimensional, low-sample) normal variates. A simple iteration is provided to automatically adjust the shrinkage level, which provably converges to the maximum likelihood hyperparameter estimation for any choice of the starting point. The proposed approach is effective and can outperform both MSE-optimized diagonal loading and the Rao-Blackwell Leidot-Wolf estimator in terms of covariance-matrix-specific metrics.

Adaptive radar detection of point-like targets in presence of disturbance with unknown spectral properties is addressed. By relaxing the assumptions of the W-ABORT, a tunable detector with improved detection vs rejection trade-off is proposed. To this aim, the presence of a fictitious signal under the null hypothesis is modeled probabilistically, so that an additional degree of freedom is introduced in the statistic of the detector. The resulting parametric GLRT shows a range of possible behaviors, from Kelly's detector to the plain W-ABORT. Monte Carlo simulations reveal that an improved compromise can be obtained. Simple closed-form approximations are also given with near-optimal performance, so that the ultimate computational complexity remains comparable to that of the W-ABORT.

This paper proposes a procedure based on time-difference of arrival measurements to localize a blind node in an asynchronous network where a set of nodes with known position is present. The proposed method computes the time-difference of arrival of each transmitted signal to any pair of receiving nodes in order to get rid of the unknown transmission time. A range-based localization procedure is implemented: first a least-squares estimator is used to compute a set of pseudo-ranges involving the blind node, then an iterative least-squares method is used to localize the blind node. The effectiveness of the proposed scheme is illustrated via simulations.

In this paper, we deal with the problem of RSS-based self-localization of a wireless blind node using a statistical path loss model for the measurements. The considered environment is non-homogeneous, i.e., the attenuation factors of the various links are different. We propose a two-stage procedure: the first stage exploits measurements between anchors to estimate transmitted powers and attenuation factors. Then, a ML localization algorithm, fed by the measurements at the blind node only, is used to estimate the unknown position. In this second stage, the attenuation factors between the blind node and the anchors are modeled as IID RVs ruled by a Gaussian distribution with mean and variance to be computed based on the estimated attenuation factors of the first stage. The performance assessment shows that the proposed approach could be a viable means to handle localization in non-homogeneous environments.

In this paper we consider a network of agents monitoring a spatially distributed traffic process. Each node measures the number of arrivals seen at its monitoring point in a given time-interval. We propose an asynchronous distributed approach based on a hierarchical Bayes model with unknown hyperparameter, which allows each node to compute the minimum mean square error (MMSE) estimator of the local arrival rate by suitably fusing the information from the whole network. Simulation results show that the distributed scheme improves the estimation accuracy compared to a purely decentralized setup and is reliable even in presence of limited local data. An ad-hoc algorithm with reduced complexity is also proposed, which performs very closely to the optimal MMSE estimator.

In this paper, we use the tools of statistical hypothesis testing to determine whether or not the different links of a WSN are homogeneous. At the design stage we use a statistical path loss law to model the RSS measurements. More precisely, in the homogeneous case all links share one and the same attenuation factor while in the non-homogeneous one the attenuation factors of the various links are different. We thus derive a GLRT-based decision rule for the considered problem and compute its distribution. Some numerical examples are finally presented to evaluate the potential to discriminate between the two hypotheses.

We propose a new approach to adaptive detection of coherent signals backscattered by possible point-like targets in the context of electronic warfare; in fact, the possible target signal is buried in thermal noise, clutter, noise-like interferers and, possibly, coherent (i.e., deceptive ECM) interferers. We assume a network of radars: for a given cell under test only a subset of the radars receives ECM signals. Training data containing thermal noise, clutter, and noise-like interferers are available. The problem at hand is solved resorting to a two-stage detection strategy: first, the subset of radars under ECM is estimated; then, a proper detection strategy resorting to W-ABORT based detectors for radars under ECM is implemented. The performance assessment shows that the proposed solution is effective in presence of ECM systems.

The paper deals with the problem of distance estimation (ranging) between nodes of a wireless system, relevant e.g. to range-based localization. The case of Received Signal Strength (RSS) measurements is addressed, where the Path Loss model (PLM) is adopted to infer the unknown distance via Maximum Likelihood (ML) estimation. In the paper it is shown that, although the ML-based approach can provide unbiased estimates when the PLM parameters are known, it may be severely biased in the real case of self-calibration via estimated parameters. The bias is characterized in detail, and nonlinear effects depending on system aspects are highlighted through the analysis. Novel reduced-bias estimators with low complexity are then derived, and their effectiveness is demonstrated via Monte Carlo simulations and illustrative experimental results by GNU Radio IEEE 802.15.4 receiver and COTS ZigBee nodes.

In this paper the feasibility of an alternative approach to construct stream ciphers is discussed by revisiting an old friend, i.e. the popular one-time pad. The idea is that the rationale underlying the one-time pad --- which is notoriously unpractical in pure form due to the need of massive secret key exchange --- might be translated into practical cryptosystems that are different from conventional stream ciphers. In alternative to the usual pseudo- random keystream generation approach (expansion), a ``dual'' approach based on sampling of a much longer sequence (extraction) could be conceivable nowadays due to the ready availability of sufficiently large memory resources, even in mobile devices such as PDAs, smartphones and tablets. The paper presents this idea, analyzing its pros and cons versus the classical one-time pad and conventional stream ciphers.

This paper presents an open source Software-Defined Radio tool compliant with IEEE 802.15.4, which incorporates features for the collection and processing of Received Signal Strength (RSS) measurements from incoming packets. The implementation includes RSS Indicator (RSSI) feature, data handling and application code for channel estimation, ranging and localization. The tool can be used for experimenting with RSSI measurements from over-the-air IEEE 802.15.4 packets. To illustrate the tool usage, we present experimental results on packets sniffed from commercial ZigBee nodes. Moreover, we highlight some issues in the RSSI calculation, showing how different aspects of the RSS computation can be investigated at the finest granularity, hence allowing researchers and practitioners to experiment down to the PHY layer.

Current Radio-Frequency Identification (RFID) technology involves two types of physical devices: tags and reader. The reader combines in a single physical device transmission (to the tags) and reception (from the tags) functions. In this paper we discuss an alternative approach, where receive functions are performed by a separate device called listener. This allows distributed tag-sensing schemes where one transmitter coexists with multiple listeners. We discuss pros and cons of both approaches and present our implementation of a passive RFID listener on GNU Radio. Our implementation is a basis for experimenting with future distributed listener-based systems, but it can be also used as a cheap and flexible protocol analyzer for currently available commercial RFID readers.

In this work we present a novel scheme for statistical-based anomaly detection in 3G cellular networks. The traffic data collected by a passive monitoring system are reduced to a set of per-mobile user counters, from which time-series of unidimensional feature distributions are derived. An example of feature is the number of TCP SYN packets seen in uplink for each mobile user in fixed-length time bins. We design a change-detection algorithm to identify deviations in each distribution time-series. Our algorithm is designed specifically to cope with the marked non-stationarities, daily/weekly seasonality and longterm trend that characterize the global traffic in a real network. The proposed scheme was applied to the analysis of a large dataset from an operational 3G network. Here we present the algorithm and report on our practical experience with the analysis of real data, highlighting the key lessons learned in the perspective of the possible adoption of our anomaly detection tool on a production basis.

In this work we present a simple toy-model that is able to explain certain empirical observations reported in a set of previous papers by Hohn et al. about the wavelet spectrum of real traffic traces. Therein, the authors found that the wavelet spectrum is substantially invariant to flow scrambling and truncation. Such finding suggested that super-flow structures above the transport layer - i.e., sessions - can be ignored for modeling the packet arrival process. Based on the proposed toy-model, we offer an interpretation framework that goes in the opposite direction, indicating that sessions, not transport-layer flows, should be taken as the main structural entities in simplified on/off models.

In this paper we investigate the dynamics of one-way delays in an operational mobile core network. Our final goal is to develop anomaly detection schemes for the packet delay process in order to reveal network and equipment problems. This requires a preliminary exploration of the delay distribution in the core network, which we undertake in this study. We present one-way delay measurements extracted from passive traces captured at an operational General Packet Radio System (GPRS)/Universal Mobile Telecommunications System (UMTS) network. We find that queuing is the only source of delay at Gateway GPRS Support Nodes (GGSN), while for Serving GPRS Support Nodes (SGSN) the mobility and flow-control further complicate the characterization of the ldquonormalrdquo delay behavior. Moreover, the presence of unwanted traffic has an impact on the delay statistics and should be taken into account. We find that information about actual bandwidth conditions in the Radio Network may be inferred by investigating packet delays in the core network. Our explorative results are promising about the possibility of leveraging one-way delay measurement for troubleshooting in such networks.