We introduce the definition ofm-to-1 mappings between two finite sets, which unifies and generalizes the definitions of 2-to-1 andn-to-1 mappings in recent literature. We also characterize thesem-to-1 mappings in terms of the generalized local criterion and thus provide three generic constructions ofm-to-1 mappings, which unify and generalize the previous known constructions. Using these constructions, the problem whetherxrh(xs) ism-to-1 on the multiplicative groupF∗qis converted into that whether an associated polynomialxr1h(x)s1ism2-to-1 on the order ℓ subgroupUℓ ofF∗q, wherem2=m/(r,s) and ℓ = (q− 1)/s. Furthermore, them2-to-1 property ofxr1h(x)s1onUℓ is studied in detail in five different cases. In addition, a recursive construction ofm-to-1 mappings fromm-to-1 mappings is proposed.

Compared with real-time multi-object tracking (MOT), offline multi-object tracking (OMOT) has the advantages to perform 2D-3D detection fusion, erroneous link correction, and full track optimization but has to deal with the challenges from bounding box misalignment and track evaluation, editing, and refinement. This paper proposes “BiTrack”, a 3D OMOT framework that includes modules of 2D-3D detection fusion, initial trajectory generation, and bidirectional trajectory re-optimization to achieve optimal tracking results from camera-LiDAR data. The novelty of this paper includes threefold: (1) development of a point-level object registration technique that employs a density-based similarity metric to achieve accurate fusion of 2D-3D detection results; (2) development of a set of data association and track management skills that utilizes a vertex-based similarity metric as well as false alarm rejection and track recovery mechanisms to generate reliable bidirectional object trajectories; (3) development of a trajectory re-optimization scheme that re-organizes track fragments of different fidelities in a greedy fashion, as well as refines each trajectory with completion and smoothing techniques. The experiment results on the KITTI dataset demonstrate that BiTrack achieves the state-of-the-art performance for 3D OMOT tasks in terms of accuracy and efficiency.

Orthogonal time frequency space (OTFS) modulation is a promising technology for mitigating severe Doppler effects in high-mobility scenarios. However, existing OTFS channel estimation methods neglect the Doppler Squint Effect (DSE), which incurs serious performance loss. In this paper, we propose a channel estimation algorithm based on Newton's method to accurately estimate fractional delay and Doppler in OTFS systems with DSE. In particular, we obtain the maximum delay and Doppler grid spacing for codebook design to guarantee the convergence of the algorithm. Additionally, we derive the Cramér-Rao lower bound (CRLB) for testing channel parameter estimation performance of our proposed algorithm. Simulation results demonstrate that our proposed algorithm outperforms the orthogonal matching pursuit (OMP) algorithm in terms of normalized mean square error (NMSE), surpassing the Newtonized OMP algorithm with traditional dictionary matrix and approaching the CRLB performance.

Most autonomous driving (AD) datasets incur substantial costs for collection and labeling, inevitably yielding a plethora of low-quality and redundant data instances, thereby compromising performance and efficiency. Many applications in AD systems necessitate high-quality training datasets using both existing datasets and newly collected data. In this paper, we propose a traffic scene joint active learning (TSceneJAL) framework that can efficiently sample the balanced, diverse, and complex traffic scenes from both labeled and unlabeled data. The novelty of this framework is threefold: 1) a scene sampling scheme based on a category entropy, to identify scenes containing multiple object classes, thus mitigating class imbalance for the active learner; 2) a similarity sampling scheme, estimated through the directed graph representation and a marginalize kernel algorithm, to pick sparse and diverse scenes; 3) an uncertainty sampling scheme, predicted by a mixture density network, to select instances with the most unclear or complex regression outcomes for the learner. Finally, the integration of these three schemes in a joint selection strategy yields an optimal and valuable subdataset. Experiments on the KITTI, Lyft, nuScenes and SUScape datasets demonstrate that our approach outperforms existing state-of-the-art methods on 3D object detection tasks with up to 12% improvements.

Federated Learning (FL) is widely applied in communication networks. The performance of clients in FL can vary due to various reasons. Assessing the contributions of each client is crucial for client selection and compensation. It is challenging because clients often have non-independent and identically distributed (non-iid) data, leading to potentially noisy or divergent updates. The risk of malicious clients amplifies the challenge especially when there’s no access to clients’ local data or a benchmark root dataset. In this paper, we introduce a novel method called Fair, Robust, and Efficient Client Assessment (FRECA) for quantifying client contributions in FL. FRECA employs a framework called FedTruth to estimate the global model’s ground truth update, balancing contributions from all clients while filtering out impacts from malicious ones. This approach is robust against Byzantine attacks and incorporates a Byzantine-resilient aggregation algorithm. FRECA is also efficient, as it operates solely on local model updates and requires no validation operations or datasets. Our experimental results show that FRECA can accurately and efficiently quantify client contributions in a robust manner.

Federated Learning (FL) Services enable customers (requesters) to outsource their FL tasks to the FL service provider, who will recruit a group of clients with appropriate datasets to complete the FL task. For a given FL task, how to select appropriate clients fairly becomes a challenging problem due to budget restrictions and client heterogeneity. In this paper, we propose a new client selection, scheduling, and rewarding scheme to ensure fairness through a three-stage process: 1) multicriteria initial client pool selection, 2) data quality-oriented perround client scheduling, and 3) performance-based rewarding. Specifically, we first define a client selection metric with multiple criteria, such as client resources, data quality, and client behaviors. Then, we formulate the initial client pool selection problem into an optimization problem that aims to maximize the overall scores of selected clients within a given budget and propose a greedy algorithm to solve it. Furthermore, we formulate the per-round client selection problem into a data quality-oriented scheduling problem that aims to improve model quality and guarantee fairness. We propose a heuristic algorithm to divide the pool into several subsets such that the federated dataset in a subset is close to an independent and identical distribution (iid) while guaranteeing each client is selected at least once in a scheduling period. In addition, we propose a performance-based payment adjustment protocol with a bonus and punishment mechanism, such that the final payment reflects the actual performance of each selected client. Our fairness analysis and experimental results show that our scheme not only can guarantee fairness but also can improve the model quality especially when data are non-iid.

Simulation data can be accurately labeled and have been expected to improve the performance of data-driven algorithms, including object detection. However, due to the various domain inconsistencies from simulation to reality (sim-to-real), cross-domain object detection algorithms usually suffer from dramatic performance drops. While numerous unsupervised domain adaptation (UDA) methods have been developed to address cross-domain tasks between real-world datasets, progress in sim-to-real remains limited. This paper presents a novel Complex-to-Simple (CTS) framework to transfer models from labeled simulation (source) to unlabeled reality (target) domains. Based on a two-stage detector, the novelty of this work is threefold: 1) developing fixed-size anchor heads and RoI augmentation to address size bias and feature diversity between two domains, thereby improving the quality of pseudo-label; 2) developing a novel corner-format representation of aleatoric uncertainty (AU) for the bounding box, to uniformly quantify pseudo-label quality; 3) developing a noise-aware mean teacher domain adaptation method based on AU, as well as object-level and frame-level sampling strategies, to migrate the impact of noisy labels. Experimental results demonstrate that our proposed approach significantly enhances the sim-to-real domain adaptation capability of 3D object detection models, outperforming state-of-the-art cross-domain algorithms, which are usually developed for real-to-real UDA tasks.

2D&3D object detection always suffers from a dramatic performance drop when transferring the model trained in the source domain to the target domain due to various domain shifts. In this paper, we propose a Joint Self-Training (JST) framework to improve 2D image and 3D point cloud detectors with aligned outputs simultaneously during the transferring. The proposed framework contains three novelties to overcome object biases and unstable self-training processes: 1) an anchor scaling scheme is developed to efficiently eliminate the object size biases without any modification on point clouds; 2) a 2D&3D bounding box alignment method is proposed to generate high-quality pseudo labels for the self-training process; 3) a model smoothing based training strategy is developed to reduce the training oscillation properly. Experiment results show that the proposed approach improves the performance of 2D and 3D detectors in the target domain simultaneously; especially the superior accuracy of 3D detection can be achieved on benchmark datasets over the state-of-the-art methods.

Wireless powered communication (WPC) has been considered as one of the key technologies in the Internet of Things (IoT) applications. In this paper, we study a wireless powered time-division duplex (TDD) multiuser multiple-input multiple-output (MU-MIMO) system, where the base station (BS) has its own power supply and all users can harvest radio frequency (RF) energy from the BS. We aim to maximize the users’ information rates by jointly optimizing the duration of users’ time slots and the signal covariance matrices of the BS and users. Different to the commonly used sum rate and max-min rate criteria, the proportional fairness of users’ rates is considered in the objective function. We first study the ideal case with the perfect channel state information (CSI), and show that the non-convex proportionally fair rate optimization problem can be transformed into an equivalent convex optimization problem. Then we consider practical systems with imperfect CSI, where the CSI mismatch follows a Gaussian distribution. A chance-constrained robust system design is proposed for this scenario, where the Bernstein inequality is applied to convert the chance constraints into the convex constraints. Finally, we consider a more general case where only partial knowledge of the CSI mismatch is available. In this case, the conditional value-at-risk (CVaR) method is applied to solve the distributionally robust system rate optimization problem. Simulation results are presented to show the effectiveness of the proposed algorithms.

Abstract Conventional algorithms fail to obtain satisfactory background segmentation results for underwater images. In this study, an improved K-means algorithm was developed for underwater image background segmentation to address the issue of improper K value determination and minimize the impact of initial centroid position of grayscale image during the gray level quantization of the conventional K-means algorithm. A total of 100 underwater images taken by an underwater robot were sampled to test the aforementioned algorithm in respect of background segmentation validity and time cost. The K value and initial centroid position of grayscale image were optimized. The results were compared to the other three existing algorithms, including the conventional K-means algorithm, the improved Otsu algorithm, and the Canny operator edge extraction method. The experimental results showed that the improved K-means underwater background segmentation algorithm could effectively segment the background of underwater images with a low color cast, low contrast, and blurred edges. Although its cost in time was higher than that of the other three algorithms, it none the less proved more efficient than the time-consuming manual segmentation method. The algorithm proposed in this paper could potentially be used in underwater environments for underwater background segmentation.

In this article, a new relay-aided secure communication system is investigated, where a transmitter sends signals to a destination via an amplify-and-forward (AF) relay in the presence of an eavesdropper. We consider a general system configuration, where the source, relay, destination, and eavesdropper are all equipped with multiple antennas. In the practical scenarios of unknown eavesdropper's channel state information (CSI) and uncertainty of the eavesdropper's location, we aim to maximize the expected value of the system secrecy rate over the presumed distribution of the eavesdropper's channels, by exploiting the artificial noise (AN) transmitted by the source and relay nodes. The system design issue is formulated as a nonconvex stochastic optimization problem with a source transmission power constraint and a nonconvex relay transmission power constraint. A novel computational method is proposed to solve this challenging problem. The new method is developed based on an exact penalty function method together with a parallel stochastic decomposition algorithm. Numerical simulations are performed to study the effectiveness of the proposed scheme at various locations of the eavesdropper. Simulation results show that for most cases, secure communication can be achieved without the CSI knowledge of eavesdropper's channels, and the achievable secrecy rate follows the trend of a benchmark system where the eavesdropper's full CSI is available. In particular, the achievable system secrecy rate increases with the number of antennas at the legitimate users. Moreover, the optimal power allocated for the transmission of the AN increases with the system signal-to-noise ratio. The proposed computational method achieves a higher system secrecy rate than a conventional penalty function based approach.