Abstract The evolution toward 6G will continue to leverage massive multiple-input multiple-output and millimeter-wave systems, which demand accurate angle-of-arrival (AoA) and angle-of-departure (AoD) estimation. While several deep learning models have demonstrated strong performance for this task, their accuracy, like that of most estimation methods, is often degraded by hardware non-idealities, which can be further exacerbated by time-varying operational factors such as component aging and adverse weather, among others. Building on a pre-trained U-Net architecture with demonstrated competitive performance for AoA/AoD estimation, we first propose an adaptation mechanism based on fine-tuning with impairment-augmented data. Specifically, we simulate hardware imperfections by introducing random phase errors in the antenna elements, ranging from mild fluctuations to severe signal distortions. The U-Net model with adaptation capabilities is then implemented on an NVIDIA Jetson Orin Nano device, a compact edge platform with heterogeneous computing resources. To this end, we design a co-execution strategy that performs AoA/AoD estimation (inference) on the CPU while simultaneously fine-tuning the model on the GPU, thus enabling continuous model adaptation to changing environmental or hardware conditions while preserving real-time inference performance. Experimental results show that impairment-aware fine-tuning effectively counters hardware degradation, particularly under significant phase impairments. In such scenarios, the fine-tuned model consistently preserves or even improves estimation accuracy, reducing the Root Mean Square Error (RMSE) by approximately 3.6% and increasing the Probability of Detection ( $$P_D$$ P D ) by up to 1 percentage point compared to the base model. Furthermore, a detailed energy-performance analysis demonstrates that while maximum frequency settings reduce training time by over 11 $$\times $$ × , they also increase power consumption by more than 5 $$\times $$ × , with optimal energy efficiency achieved at mid-range CPU and high GPU frequencies. This work establishes the feasibility of concurrent training and inference on resource-constrained heterogeneous hardware, paving the way for resilient and autonomous edge intelligence in future 6G systems.

Abstract Advances in object tracking and acoustic beamforming are driving new capabilities in surveillance, human-computer interaction, and robotics. This work presents an embedded system that integrates deep learning–based tracking with beamforming to achieve precise sound source localization and directional audio capture in dynamic environments. The approach combines single-camera depth estimation and stereo vision to enable accurate 3D localization of moving objects. A planar concentric circular microphone array constructed with MEMS microphones provides a compact, energy-efficient platform supporting 2D beam steering across azimuth and elevation. Real-time tracking outputs continuously adapt the array’s focus, synchronizing the acoustic response with the target’s position. By uniting learned spatial awareness with dynamic steering, the system maintains robust performance in the presence of multiple or moving sources. Experimental evaluation demonstrates significant gains in signal-to-interference ratio, making the design well-suited for teleconferencing, smart home devices, and assistive technologies.

In the context of upcoming sixth-generation (6G) wireless communication systems, the use of millimeter wave (mmWave) frequencies is a key technology for achieving high-throughput communications. Accurate parametric estimation of mmWave channels is critical for effective beamforming design and configuration, requiring sophisticated models to capture the directional characteristics of these channels. This work considers an innovative artificial intelligence (AI) approach for accurate estimation of angle-of-arrival (AoA) and angle-of-departure (AoD) parameters from frequency-domain channel observations. Our approach is based on the implementation of two convolutional neural networks (CNNs): a residual CNN (ResNet) and a U-Net CNN. Specifically, this work focuses on the efficient implementation of both schemes in an embedded system suitable for edge AI. We performed the experiments in a low-power NVIDIA Jetson Orin Nano platform and evaluated the effect of modifying the frequencies of its CPU and GPU on the performance of the inference process, both in terms of execution time and energy consumption. Experimental results showed that the U-Net model is more power consuming, but as it is faster, it consumes less energy per channel.

In this paper, the performance of Low-Cost Sensors (LCS) for Air Quality (AQ) monitoring is analyzed. These sensors are unstable, unreliable and cannot replace the official ones. They suffer from deviations, drifts and errors that make them invalid for direct use, but through Machine Learning (ML) techniques, they can increase their overall accuracy. These sensors can be used in smart cities, increasing the sampling density of the AQ monitoring network, which could enable the development of new applications for citizens to plan healthy routes. After a review of the different AQ LCS, we focused on the Winsen’s ZPHS01B multisensor module because it embeds 11 different sensors in the same module. Since the information and the experimental data from field tests for this module is limited, we carry out different experiments and perform a thorough analysis, evaluating the sensor reading differences that could appear in a batch of them. We note that some of the integrated sensors are more reliable than others, but in practice they can improve their reading by using the ML models mentioned above, as they show correlation. Of all the modules tested, we observed that most of the sensors showed similar performance. However, a certain percentage of sensors in certain modules performed worse than their counterparts, which shows that at least 80% of the tested multi-sensor modules had similar levels of performance.

The creation and implementation of a system that bridges visible light communication (VLC) with selected identifiers, envisioned as informational beacons, are outlined. This system is adept at controlling LED panels and orchestrating communication between LED lights and the receiving unit, laying the groundwork for vehicle-to-vehicle (V2V) communication systems tailored for precise positioning. The design and assessment of test-bed requirements cater to the anticipation of future V2V optical wireless communication scenarios. Through analysis, it is demonstrated that commercial outdoor lighting can be effectively utilized for V2V positioning, capable of operating under daylight with a data transmission rate of 100 Kbps. Crucially, the system’s ability to accurately detect positioning beacons facilitates the localization of vehicles within designated areas of interest, paving the way for the advancement of V2V communication technologies. This comprehensive approach not only validates the feasibility of using VLC for vehicle positioning but also sets a precedent for future developments in V2V communication systems.

In this work, we propose using a deep learning method for parametric millimeter-wave (mmWave) channel estimation, specifically focusing on angle-of-arrival (AoA) and angle-of-departure (AoD) parameters in the frequency domain. Channel estimation is fundamental in mmWave to efficiently implement beamforming techniques. Our approach involves adapting a residual convolutional neural network (ResNet) to the task, incorporating a technique from topological data analysis to accurately estimate angular frequencies. Additionally, we enhance the model’s performance by including a posterior model fitting to improve the probability of detection. Through simulations, we compare our ResNet-based approach with existing signal processing methods and the Crámer-Rao lower bound. Our findings demonstrate significant enhancements in system robustness, increasing the probability of detection while minimizing estimation errors.

Parametric millimeter-wave (mmWave) channel estimation involves modeling the channel matrix by combining direction-dependent signal paths, exploiting the sparse nature of mmWave channels. In our study, we propose a deep learning-based approach to estimate angle-of-arrival (AoA) and angle-of-departure (AoD) parameters from input observations in the frequency domain. To address this challenge, we have adapted a residual convolutional neural network (ResNet) to this specific problem and incorporated a technique from topological data analysis, enabling us to accurately retrieve the angular frequencies. Furthermore, we have extended this basic architecture by incorporating a posterior model fitting to enhance the system performance in terms of probability of detection. In our research, we compare the ResNet and extended ResNet approaches with state-of-the-art signal processing techniques and the Crámer-Rao lower bound through simulation. Our results indicate significant improvements in system robustness by increasing the probability of detection while maintaining a reduced estimation error.

Abstract Hybrid platforms combining multicore central processing units (CPU) with many-core hardware accelerators such as graphic processing units (GPU) can be smartly exploited to provide efficient parallel implementations of wireless communication algorithms for Fifth Generation (5G) and beyond systems. Massive multiple-input multiple-output (MIMO) systems are a key element of the 5G standard, involving several tens or hundreds of antenna elements for communication. Such a high number of antennas has a direct impact on the computational complexity of some MIMO signal processing algorithms. In this work, we focus on the channel estimation stage. In particular, we develop a parallel implementation of a recently proposed MIMO channel estimation algorithm. Its performance in terms of execution time is evaluated both in a multicore CPU and in a GPU. The results show that some computation blocks of the algorithm are more suitable for multicore implementation, whereas other parts are more efficiently implemented in the GPU, indicating that a hybrid CPU–GPU implementation would achieve the best performance in practical applications based on the tested platform.

The use of Multi-Processor System-on-Chip (MPSoC) is becoming widespread in a huge number of signal processing systems, including wireless communications and vehicular technology applications. In those scenarios, when Multiple-Input Multiple-Output (MIMO) communication schemes are considered, the system usually has to deal with a high number of communication links that involve sensors and antennas from different vehicles and users. The use of MIMO systems with a high number of antennas increases the complexity of many signal processing algorithms which could benefit from computationally efficient implementations. The Xilinx Zynq UltraScale+ EG Heterogeneous MPSoC is a well-positioned platform to manage computationally-demanding communication systems. This platform holds a dual-core ARM Cortex-R5, a quad-core ARM Cortex-A53, a graphics processing unit (GPU) and a high-end Field Programmable Gate Array (FPGA). In particular, this work aims to evaluate the computational performance of the Transformed Spatial Domain Channel Estimation (TSDCE), a novel millimeter-wave MIMO channel estimation algorithm, on the proposed embedded platform. This work focuses firstly on developing an efficient sequential implementation that runs on the ARM Cortex-A53, so that we can afterwards leverage the use of the multi-core system to accelerate the sequential performance.

Industry 4.0 is currently viewed as the future era in which a new generation of wireless communications will enable the widespread connectivity between machines, objects, and users, promising significant advancements in industrial automation. Industrial scenarios are highly dynamic environments that require the ability to adapt to changes in operating conditions. Reliable wireless communications can be challenging in industrial environments, where the communications signal is often blocked due to the presence of industrial machinery, with possible intermittent and/or unpredictable blockages. This paper analyzes the wireless coverage provided by a low-cost IoT system in an industrial environment. A measurement campaign in different points of an industrial warehouse were carried out, showing substantial differences in terms of coverage between a totally empty area and an area with machinery and workers. The results show that the IoT system is able to provide enough coverage around the whole area, even with mobility or blockages.

Enhanced vehicle-to-everything (eV2X) communication is one of the key challenges to be addressed by the fifth generation (5G) of cellular mobile communications. In particular, eV2X includes some 5G vehicular applications targeting fully autonomous driving which require ultra-high reliability. Although vehicular communications are by default assumed between single antennas located on the roof of the transmitter and receiver vehicles, prior art has shown that there are other antenna positions more suitable for V2X communication, depending on the specific communication context. Antenna selection can be used in this case to select one specific antenna or a subset of them better suited for a certain communication link. In this work, we propose a context-aware antenna selection procedure able to enhance the communication with multi-antenna vehicles. To enable such scheme in 5G systems, we discuss the necessary signaling to extend current 5G radio resource control and radio resource management mechanisms, which are mainly focused on single-antenna communication. The signaling overhead caused by context exchange for antenna selection is analyzed and compared to the overhead when reference signals are exchanged for that purpose instead. Finally, simulation results for a 5G platooning use case are presented to show the advantages of antenna selection.

Abstract The emerging Multi-Processor System-on-Chip (MPSoC) technology, which combines heterogeneous computing with the high performance of field programmable gate arrays (FPGA), is a promising platform for a large number of applications, including wireless communications and vehicular technology. In this specific application context, when multiple-input multiple-output (MIMO) scenarios are considered, the system usually has to manage a large number of communication links among sensors and antennas involving different vehicles and users. Millimeter wave (mmWave) communications are one of the key technology enablers toward achieving high data rates in beyond 5G systems (B5G). Communication at these frequency bands usually involves the use of large antenna arrays, often requiring high computational resources. One of the candidate platforms able to manage a huge number of communications is the Xilinx Zynq UltraScale+ EG Heterogeneous MPSoC, which is composed of a dual-core Cortex-R5, a quad-core ARM Cortex-A53, a graphics processing unit (GPU) and a high-end FPGA. This work analyzes the computational performance that requires a recent mmWave MIMO channel estimation algorithm in a platform of this kind. As a first approach, we will focus our work on the performance that can be achieved via the quad-core ARM Cortex-A53. To this end, we will use the libraries for numerical algebra (BLAS and LAPACK). The results show that our reference implementation is able to manage a large MIMO communication system with 256 antennas without exhausting platform resources.

High-accuracy angle of arrival (AoA) and angle of departure (AoD) estimation is critical for cell search, stable communications and positioning in millimeter wave (mmWave) cellular systems. Moreover, the design of low-complexity AoA/AoD estimation algorithms is also of major importance in the deployment of practical systems to enable a fast and resource-efficient computation of beamforming weights. Parametric mmWave channel estimation allows to describe the channel matrix as a combination of direction-dependent signal paths, exploiting the sparse characteristics of mmWave channels. In this context, a fast Transformed Spatial Domain Channel Estimation (TSDCE) algorithm was recently proposed to perform parametric channel estimation with low complexity, which in turn results in a full characterization of the transmitting and receiving angles for dominant signal paths. In this paper, we analyze the AoA/AoD estimation capability and accuracy of the TSDCE algorithm in detail. We find that the TSDCE algorithm has a significant performance advantage with respect to the traditional approach, which is based on frequency domain processing, in complexity-constrained environments, especially at high signal-to-noise ratios.

Mobile network operators are interested in providing Vehicle-to-Vehicle (V2V) communication services using their cellular infrastructure. Regional split of operators is one possible approach to support multi-operator infrastructure-based cellular V2V communication. In this approach, a geographical area is divided into non-overlapping regions, each one served by a unique operator. Its main drawback is the communication interruption motivated by the inter-operator handover in border areas, which prevents the fulfillment of the maximum end-to-end (E2E) latency requirements of fifth generation (5G) V2V services related to autonomous driving. In this work, we enable a fast inter-operator handover based on the pre-registration of the users on multiple operators, which substantially reduces the handover time to guarantee maximum E2E latency values of 100 ms in non-congested scenarios. To further reduce the latency of time-critical services to always less than 70 ms, even with the handover interruption time, while providing a latency around 20 ms in the majority of locations, we propose to complement the former technique with a mobile edge computing approach. Our proposal consists in the localization of application servers and broadcasting entities in all the base stations, to avoid the communication through the core network, together with the use of a new set of nodes in the base stations of cross-border areas called inter-operator relays, to minimize the communication latency between operators. Based on analytic and simulation results, it is demonstrated that the proposed techniques are effective to support low-latency infrastructure-based cellular V2V communications in multi-operator environments with regional split.

Fast channel estimation in millimeter-wave (mmWave) systems is a fundamental enabler of high-gain beamforming, which boosts coverage and capacity. The channel estimation stage typically involves an initial beam training process where a subset of the possible beam directions at the transmitter and receiver is scanned along a predefined codebook. Unfortunately, the high number of transmit and receive antennas deployed in mmWave systems increase the complexity of the beam selection and channel estimation tasks. In this work, we tackle the channel estimation problem in analog systems from a different perspective than used by previous works. In particular, we propose to move the channel estimation problem from the angular domain into the transformed spatial domain, in which estimating the angles of arrivals and departures corresponds to estimating the angular frequencies of paths constituting the mmWave channel. The proposed approach, referred to as transformed spatial domain channel estimation (TSDCE) algorithm, exhibits robustness to additive white Gaussian noise by combining low-rank approximations and sample autocorrelation functions for each path in the transformed spatial domain. Numerical results evaluate the mean square error of the channel estimation and the direction of arrival estimation capability. TSDCE significantly reduces the first, while exhibiting a remarkably low computational complexity compared with well-known benchmarking schemes.

Wireless communications for vehicular applications in fifth generation cellular systems (5G) are required to be of low latency and high reliability. Among other factors, the amount of control information to be exchanged between each vehicle and the base station can penalize the communication latency. Several 5G vehicular use cases involve communications within groups of vehicles, for instance, vehicle platooning. This work is focused on exploiting the structure and characteristics of this particular group-based vehicular service to decrease the control information exchange related to the channel estimation stage necessary for vehicle to infrastructure cellular communications. A scheme based on channel spatial interpolation is proposed, where the real channel is only available at a subset of vehicles, and subsequent spatial interpolation of the channel provides estimates of the large-scale channel parameters for the rest of vehicles in the group. In the paper, communication cost expressions are derived for centralized and distributed topologies, considering a scenario with a macro base station serving a vehicle platoon. The evaluation results show that the distributed topology is more efficient in terms of communication cost, while the centralized architecture is more robust against inter-vehicle distance variations.

With the advent of automated driving functions the need for cooperation among vehicles becomes increasingly necessary. The availability of high quality connectivity is an important factor for successful and uninterrupted cooperative and autonomous services. Flexible selection of available connectivity options, increases the probabilities to achieve and maintain the required Quality of Service (QoS). Future vehicles will support multiple 5G communication interfaces e.g., cellular (Uu), sidelink (PC5). A 5G RAN-based solution is proposed in this paper to dynamically select and change the communication interface that is suitable for a V2X service, considering radio and road conditions. The combination of different interfaces, for data packets duplication or splitting, increases the multi-connectivity capabilities and expected QoS benefits. The simulation results show the improvement that multi-connectivity could bring to V2X services and specifically to packet reception ratio (PRR) in an urban environment.

The long-term evolution multimedia broadcast multicast service (MBMS) infrastructure has been shown to efficiently fulfill the requirements of classical vehicle-to-vehicle services, such as cooperative awareness message and decentralized environmental notification message delivery, which involve the transmission of the same message to a set of destination receivers in close proximity. When targeting autonomous driving applications, related 5G V2X services, such as cooperative collision avoidance (CCA), impose challenging latency and reliability requirements that cannot be generally guaranteed using the current MBMS architecture. In this paper, we propose a low-latency multicast scheme aimed at decreasing the end-to-end (E2E) communication latency. The proposed approach concentrates the multicast network functionalities into the layer-2 of the base stations (BSs) protocol stack, including a message passing between neighboring BSs, to achieve a fast management of multicast groups and a reduction of the communication path through the network infrastructure. The results show that the proposed technique can substantially decrease the E2E latency of conventional MBMS and ensure the correct operation of demanding services, such as CCA moving toward 5G communication systems.

In order to shed some useful light on the design of multi-connectivity (MC) solutions for vehicular-to anything (V2X) communications, this work evaluates the performance of different connectivity alternatives based on LTE-Advanced for the support of key intelligent transportation systems (ITS) services as CAM, DENM and platooning in a realistic highway scenario. In particular, downlink unicast, downlink multicast based on Single-Cell Point-to-Multipoint (SC-PTM), and V2X sidelink connectivity modes have been benchmarked for each ITS service separately. Results show that V2X sidelink transmission exhibits excellent performance for all the services when its subchannelization is properly configured. Downlink multicast is a valid option for DENM and CAM although the optimum modulation and coding scheme (MCS) is different for each service. Finally, unicast is unable to provide a successful performance for CAM while it is valid for platooning and DENM.

This paper presents Grassmannian signaling as a transmission scheme that can be integrated in Long Term Evolution (LTE) to support higher user speeds and to increase the throughput achievable in the high Signal to Noise Ratio (SNR) regime. This signaling is compared, under realistic channel assumptions, with the diversity transmission modes standardized in LTE, in particular, Space-Frequency Block Coding and Frequency-Switched Transmit Diversity for two and four transmit antennas, respectively. In high-speed scenarios, and even with high antenna correlation, Grassmannian signaling outperforms the LTE diversity transmission modes starting from four transmit antennas. Furthermore, in the high SNR regime, Grassmannian signaling can increase the link data rate up to 10% and 15% for two and four antennas, respectively.

The coexistence of multiple air interface variants in the upcoming fifth generation (5G) wireless technology remains a matter of ongoing discussion. This paper focuses on the physical layer of the 5G air interface and provides a harmonization solution for the joint implementation of several multicarrier waveform candidates. Waveforms based either on cyclic prefix-orthogonal frequency division multiplexing (CP-OFDM) or on filter bank multicarrier (FBMC) are first presented through a harmonized system model. Complexity comparisons among five different waveforms are provided. Then, the complexity of a proposed configurable hardware implementation setup for waveform transmission and reception is evaluated. As a result, the harmonized transmitter and receiver exhibit 25–40% and 15–25% less complexity in floating-point operations, respectively, in comparison to two standalone implementations of the most complex waveform instances of the CP-OFDM and FBMC families. This highlights the similarities between both families and illustrates the component reuse advantages associated with the proposed harmonized solution.

Among the many proposals to meet the fifth generation (5G) data rate demands, current spectrum scarcity has motivated the use of millimeter-Wave (mmW) bands for cellular communication. It is well known that mmW communication involves many propagation challenges, which can be compensated through massive multi-antenna techniques. However, implementing fully digital precoding schemes with massive arrays entails huge complexity and costs, which have recently boosted the interest for hybrid beamforming solutions. Although hybrid schemes have shown good performance in simplified setups, their suitability for realistic cellular systems with many interfering base stations and users is still unclear. In this sense, this paper assesses the performance of hybrid beamforming in a dense urban cellular system with a realistic mmW channel model and shows that it can reach the performance of fully digital maximum ratio transmission under line of sight conditions and with a sufficient number of parallel radio-frequency chains. Another important result is that multi-user hybrid beamforming provides a substantial capacity increase with respect to its single-user (SU) counterpart. This finding complements the views on mmW communication usage, which has been so far mainly intended for SU communication to provide high beamforming gains.

Current cellular technologies are based on the concept of coherent communication, in which the channel matrix used for demodulation is estimated via reference or pilot signals. Coherent systems, however, involve a significant increase of the signalling overhead, especially when the number of transmission points is increased or when the mobile channel changes rapidly, which motivates the use of non-coherent techniques. This letter extends the use of non-coherent communications to a multi-user (MU) multiple-input multiple-output (MIMO) framework by combining superposition coding with a reduced-complexity detection method. Numerical results confirm that our scheme achieves higher user rates than non-coherent MU transmission based on time multiplexing. In addition to the well-known sum-rate gain of MU systems, an extra performance gain given by downlink non-coherent MU communication is shown and qualitatively justified.

Fast parallel processing capability of general purpose Graphic Processing Units (GPU) can be exploited to accelerate the precoding calculation needed in spatially multiplexed wireless communication systems. In this paper, a GPU-based implementation of the well-known multiuser Tomlinson-Harashima precoding (THP) scheme combined with a lattice-reduction (LR) stage is presented. The proposed approach allows the LR stage to be switched off when user requirements are achieved by using only THP. Moreover, our GPU implementation provides scalability in the number of sub-carriers per symbol, which is a key factor in LTE and 4G wireless standards. Simulation results show that the GPU-based THP implementation performs up to 7 times faster than its CPU-equivalent whereas the LR stage implementation only achieves a speedup of 3. Despite the fact that the LR cannot be as efficiently parallelized as the THP, a speedup of nearly 6 is achieved when both are combined.

Tree search detection algorithms can provide Maximum-Likelihood detection over Gaussian MIMO channels with lower complexity than the exhaustive search. Furthermore, the performance of MIMO detectors is highly influenced by the channel matrix condition number. In this paper, the impact of the 2-norm condition number in data detection is exploited in order to decrease the complexity of already proposed algorithms. A suboptimal tree search method called K-Best is combined with a channel matrix condition number estimator and a threshold selection method. This approach leads to a variable-breadth K-Best detector with predictable average performance and suitable for hardware implementation. The results show that the proposed scheme has lower complexity, i.e. it is less power consuming, than a fixed K-Best detector of similar performance.