Achieving reliable multidimensional Vehicle-to-Vehicle (V2V) channel state information (CSI) prediction is both challenging and crucial for optimizing downstream tasks that depend on instantaneous CSI. This work extends traditional prediction approaches by focusing on four-dimensional (4D) CSI, which includes predictions over time, bandwidth, and antenna (TX and RX) space. Such a comprehensive framework is essential for addressing the dynamic nature of mobility environments within intelligent transportation systems, necessitating the capture of both temporal and spatial dependencies across diverse domains. To address this complexity, we propose a novel context-conditioned spatiotemporal predictive learning method. This method leverages causal convolutional long short-term memory (CA-ConvLSTM) to effectively capture dependencies within 4D CSI data, and incorporates context-conditioned attention mechanisms to enhance the efficiency of spatiotemporal memory updates. Additionally, we introduce an adaptive meta-learning scheme tailored for recurrent networks to mitigate the issue of accumulative prediction errors. We validate the proposed method through empirical studies conducted across three different geometric configurations and mobility scenarios. Our results demonstrate that the proposed approach outperforms existing state-of-the-art predictive models, achieving superior performance across various geometries. Moreover, we show that the meta-learning framework significantly enhances the performance of recurrent-based predictive models in highly challenging cross-geometry settings, thus highlighting its robustness and adaptability.

In massive MIMO systems, acquiring accurate channel state information (CSI) incurs significant overhead due to the need for extensive pilot signaling. This paper proposes a novel learning-based framework for pilot reduction and CSI reconstruction, inspired by image inpainting techniques. Pilot overhead is reduced through lightweight pilot pattern learning. At the receiver, a deep neural network reconstructs the full CSI matrix from partial observations. The pilot pattern design and CSI reconstruction can be jointly optimized in an end-to-end fashion. Unlike prior works that focus on feedback compression after full pilot transmission, our approach reduces pilot usage before transmission, enhancing efficiency while maintaining interpretability, which can enable hybrid signal processing and deep learning methods in practical systems.Simulation results on 3GPP channel models show consistent improvements of over 2 dB in normalized mean square error (NMSE) compared to baselines using uniform or random pilot patterns at only 25% pilot usage. Performance gains persist under low SNR conditions and higher compression ratios, making the approach promising for scalable 5G and beyond systems, such as intelligent CSI-RS configuration.

Indoor localization is a challenging problem that - unlike outdoor localization - lacks a universal and robust solution. Machine Learning (ML), particularly Deep Learning (DL), methods have been investigated as a promising approach. Although such methods bring remarkable localization accuracy, they heavily depend on the training data collected from the environment. The data collection is usually a laborious and time-consuming task, but Data Augmentation (DA) can be used to alleviate this issue. In this paper, different from previously used DA, we propose methods that utilize the domain knowledge about wireless propagation channels and devices. The methods exploit the typical hardware component drift in the transceivers and/or the statistical behavior of the channel, in combination with the measured Power Delay Profile (PDP). We comprehensively evaluate the proposed methods to demonstrate their effectiveness. This investigation mainly focuses on the impact of factors such as the number of measurements, augmentation proportion, and the environment of interest impact the effectiveness of the different DA methods. We show that in the low-data regime (few actual measurements available), localization accuracy increases up to 50%, matching non-augmented results in the high-data regime. In addition, the proposed methods may outperform the measurement-only highdata performance by up to 33% using only 1/4 of the amount of measured data. We also exhibit the effect of different training data distribution and quality on the effectiveness of DA. Finally, we demonstrate the power of the proposed methods when employed along with Transfer Learning (TL) to address the data scarcity in target and/or source environments.

Indoor localization is a challenging task. Compared to outdoor environments where GPS is dominant, there is no robust and almost-universal approach. Recently, machine learning (ML) has emerged as the most promising approach for achieving accurate indoor localization. Nevertheless, its main challenge is requiring large datasets to train the neural networks. The data collection procedure is costly and laborious, requiring extensive measurements and labeling processes for different indoor environments. The situation can be improved by Data Augmentation (DA), a general framework to enlarge the datasets for ML, making ML systems more robust and increasing their generalization capabilities. This paper proposes two simple yet surprisingly effective DA algorithms for channel state information (CSI) based indoor localization motivated by physical considerations. We show that the number of measurements for a given accuracy requirement may be decreased by an order of magnitude. Specifically, we demonstrate the algorithms' effectiveness by experiments conducted with a measured indoor WiFi measurement dataset: As little as 10% of the original dataset size is enough to get the same performance as the original dataset. We also showed that if we further augment the dataset with the proposed techniques, test accuracy is improved more than three-fold.

We propose machine learning (ML) based optimization methods and new high order modulations for reliable and high-capacity communications. The widely adopted square quadrature amplitude modulations (QAM) fundamentally exhibit a shaping loss of up to 1.53 dB to the Shannon capacity bound. The proposed modulations obtained through the ML based optimization outperform the square QAMs and other state of-the-art ones by about 1.2 dB and 0.3 dB, respectively, for 1024-ary modulation with LDPC coding. We construct the neural network architecture and training methods to reflect the desired properties of well-performing modulations. This significantly helps in the training convergence of the ML models to a desired optimal state and leads to the modulation constellation and bit to-symbol mapping that reduces the shaping loss to the Shannon capacity bound to a large extent. Moreover, the ML methods enable the development of new optimal modulations for a wide range of target SNR and modulation orders.

Estimating vehicle locations is important for realizing Intelligent Transportation Systems (ITS). This paper considers utilizing vehicle-to-vehicle (V2V) communication for relative vehicular localization. In particular, we develop a machine learning (ML) solution that uses the Channel State Information (CSI) from multiple-antenna transceivers for vehicular localization. We develop suitable pre-processing to obtain a compact CSI representation as an input feature to the ML solution. The proposed solution is then based on feed-forward neural networks. Training and evaluation are done on measured real-world data in the 5.9 GHz band. The performance on two routes shows that the proposed feature may improve the performance while reducing the number of trainable parameters. Furthermore, the paper raises a number of interesting observations regarding the learnability in V2V ML-based localization solutions.

Many wireless networks, including 5G NR (New Radio) and future beyond 5G cellular systems, are expected to operate on multiple frequency bands. This paper considers the band assignment (BA) problem in dual-band systems, where the basestation (BS) chooses one of the two available frequency bands (centimeter-wave and millimeter-wave bands) to communicate with the user equipment (UE). While the millimeter-wave band might offer higher data rate, there is a significant probability of outage during which the communication should be carried on the (more reliable) centimeter-wave band. With mobility, the BA can be perceived as a sequential problem, where the BS uses previously observed information to predict the best band for a future time step. We formulate the BA as a binary classification problem and propose supervised Machine Learning (ML) solutions. We study the problem when both the BS and the UE use (i) omnidirectional antennas and (ii) both use directional antennas. In the omnidirectional case, we derive analytical benchmark solutions based on the Gaussian Process (GP) assumption for the inter-band shadow fading. In the directional case, where the labeling is shown to be complex, we propose an efficient labeling approach based on the Viterbi Algorithm (VA). We compare the performances for two channel models: (i) a stochastic channel and (ii) a ray-tracing based channel.

Delay dispersion is one of the key propagation channel characteristics that impact system design and performance. In particular, for OFDM-based systems such as Wi-Fi, it determines both the amount of available frequency diversity and the minimum required cyclic prefix, which in turn impacts the spectral efficiency. For this reason, delay spread and power delay profiles have been analyzed for a long time. However, recent upgrades in Wi-Fi, in particular the addition of a new frequency range (6-7.1 GHz), and the introduction of adaptive beamforming, require a re-assessment based on new measurements. This paper presents extensive measurement results of RMS delay spread and interference quotient that take these developments into account. Results were measured in the 2.4-2.5, 5-6, and 6-7 GHz bands. Furthermore, we compare the “omni-directional” delay spread and interference quotient (i.e., when measured with omni-antennas at transmitter and receiver), to those that occur when beamformed antennas are used. We found that for both outdoor and indoor environments, beamforming typically improves the interference quotient (for a given window size) by about 3-5 dB. The 2.4 and 5-7 GHz bands show a significant difference, while we could not observe statistically significant differences between the 5-6 and 6-7 GHz bands.

Motivated by an increasing interest in wireless networking in mission-critical applications, and a recent amendment of the time slotted channel hopping to IEEE 802.15.4, the multichannel delay sensitive scheduling is investigated in the many-to-one network, which is also known as the convergecast network. In such a network, each node has data to be transmitted to a gateway through multi-hop communications. As a realistic setting, packet release time at each node is not assumed to be uniform. Under this assumption, the goal of this work is to design a scheduling scheme that minimizes the schedule length and maximum end-to-end delay, in which the former is essential for repetitive data acquisition, whereas the later improves the freshness of the acquired data. To achieve the scheduling goal, the problem is formulated as a multi-objective integer programming. To obtain a feasible solution and gain an insight into the problem, a lower bound on the schedule length is derived. Based on that, a new scheduling scheme is designed to minimize the two objectives simultaneously. Link level simulations verify the performance improvement of the proposed scheme over the existing schemes.

Future wireless networks are expected to operate in both the centimeter-wave and millimeter-wave bands. It is thus necessary to provide an accurate characterization of the {\em joint} channel properties in the two bands. Shadow fading is an important characteristic of wireless channels as it impacts achievable data rates and in particular outage probabilities. In this paper, we propose cross-band correlation models that can be used to capture the correlation over space and frequency. The models are designed to reduce to existing correlation models in the case of a single-band, which enables reuse of the large existing literature for (single-band) shadowing models. We extract and report the parameters of the proposed models in a micro-cellular environment. The important finding from our proposed models is that the cross-band spatial correlation function may not be derived just by scaling a single-band correlation function.

Device-to-Device (D2D) communications has been embraced as a novel approach for extending coverage in fifth-generation (5G and beyond) wireless networks. Throughput and reliability assessment of such wireless networks are therefore of utmost importance for D2D communication systems design. In this paper, we consider the capacity results of wireless propagation channels in two 5G modes namely the sub-6 GHz centimeter-wave (cm-wave) and 60 GHz millimeter-wave (mm-wave) bands for a multiple-input-multiple-output (MIMO) D2D fading channel. Results presented in this paper were obtained from propagation channel measurements jointly conducted in both cm-wave (2-6 GHz) and mm-wave (59-63 GHz) bands using an 8 × 8 MIMO virtual array setup in an outdoor environment. These measurements were conducted at the exact same locations (with constituents unchanged) in both cm-wave and mm-wave bands for comparability of results. Capacity values were computed using two power policies (with and without channel state information (CSI)). We also investigated the eigenmode spectral structure to understand the spatial correlatedness of the channel and the Rician K-factor to analyze small-scale fading in the channel. The results presented in this paper can be used by wireless systems designer for assessing the performance of D2D systems operating in the cm-wave and/or mm-wave bands for this type of environment.

Neighbor discovery (ND) is an essential prerequisite for any peer-to-peer communication. In general, minimizing the discovery time is the goal for ND schemes. In this paper, we study the average discovery time for directional random ND when nodes have prior information about their set of possible neighbors, which also helps identify the performance limits of random ND schemes. Typically, discovery time analysis is done for assumptions that simplify the network structure, such as uniform neighbor relations for all nodes. However, with prior information the directional transmission probabilities depend on the node and the direction. This complicates the analysis of the expected discovery time. We first provide a closed-form expression for the expected discovery time based on the non-uniform coupon collector problem. Next, we identify directional transmission probabilities of each node that achieve a small discovery time. Due to the mathematical complexity, we provide a lower and an upper bound on the expected discovery time, which allows us to write the problem as a convex optimization problem. Through simulations, we demonstrate the performance gain due to prior knowledge with the proposed methods as compared with when no prior information is available, as well as the impact of uncertainty in the prior knowledge.

Neighbor discovery is an essential prerequisite for any device-to-device (D2D) communication. Unlike ad-hoc networks, the base station (BS) in D2D networks may facilitate the neighbor discovery process. However, device-to-BS channel states or device locations are not enough to provide BS with information on the channel quality between the devices. Thus, due to the inherent uncertainty of link quality between devices, the BS-assisted neighbor discovery cannot be treated as typical scheduling problem. In this paper, we investigate the assisted directional neighbor discovery in D2D networks. We first formulate the scheduling problem as an integer optimization problem that captures the uncertainty. Then we propose a greedy based centralized scheduling to determine directional pilot transmission instances. We also propose a one-way randomized discovery, where we choose the directional transmission probabilities based on two techniques, an intuitive and an optimized methods. Finally, we provide simulation results that assess the performance the schemes.

We consider the problem of acquiring channel state information (CSI) in base-station (BS) controlled device-to-device (D2D) networks. Obtaining high-quality CSI requires a tradeoff between interference, outdatedness of CSI, and noise. Thus, the goal is to find an efficient pilot scheduling scheme that minimizes errors in the estimates. In this paper, we present the location aware training scheme (LATS) as simple yet efficient training technique. Assuming that the devices are aware of their location, LATS groups the devices into geographical segments and assigns a frequency reuse pattern to them. To identify the parameters of the scheme (segmentation and guard parameters, Snand Gn, respectively), we use the average normalized mean square error (NMSE) as a metric, which combines the effects of outdatedness, noise, and interference. We derive an approximation of the average NMSE based on statistics of the devices and LATS structure. We present simulation results that illustrate LATS behavior and show that it outperforms TDMA- and CSMA-based schemes. Furthermore, we consider some practical challenges in using location information and evaluate their effects on the scheme.