The widespread deployment of Autonomous Electric Taxis (AETs) in smart cities introduces critical challenges in large-scale dispatching and charging coordination under energy and operational constraints. Traditional Multi-Agent Reinforcement Learning (MARL) approaches often struggle to ensure both policy feasibility and system scalability in such complex, dynamic environments. In this paper, we propose a safe and scalable two-stage MARL framework for AET dispatching optimization. The proposed method, named Filter-to-Optimization Pipeline (FTOP), decouples constraint handling from reward maximization through a hierarchical architecture. In the first stage, an Action Classification-based Action Filter (ACAF) employs value decomposition to eliminate infeasible actions, enforcing energy and conflict constraints. In the second stage, a Utility-Prioritized Maximization Policy (UPMP) performs a model-based search within the filtered feasible space to optimize system-level utility. Extensive simulations demonstrate that FTOP achieves significant improvements in total fleet revenue, constraint satisfaction, and scalability across varying urban scenarios. These results highlight the potential of decomposed MARL strategies in solving large-scale, safety-critical optimization problems in autonomous transportation systems.

Using time-difference-of-arrival (TDOA) measurements from several sensors to locate a target is one of the most widely used wireless localization methods. A challenge for TDOA localization is that the measurement equations are highly nonlinear. The weighted least-squares (WLS) based method tends to incur the threshold effect, and the convex relaxation based method has a high computational complexity. This paper presents a new iterative method derived directly from maximum likelihood estimation (MLE), and it does not require any matrix operations in its iterations. In order to reduce the likelihood of reaching a local minimum or a saddle point, multiple initial values are randomly selected from the region of interest, and the one that results in the lowest cost function is chosen as the output. Besides, the theoretical bias and covariance matrix of the proposed estimator are derived. The proposed algorithm outperforms state-of-the-art methods and achieves the Cramer-Rao lower bounds (CRLB).

This paper develops an effective artificial intelligence (AI) based approach for estimating heart rate (HR) and breathing rate (BR) using FMCW radar. The method leverages spectrogram analysis and principal component analysis (PCA) to extract compact and informative features from radar signals. We design and train a feedforward dense neural network to jointly estimate heart and breathing rates using radar signals. The approach is evaluated using a publicly available dataset of radar recordings, with ground truth reference values obtained from a clinical-grade contact sensor. Experimental results demonstrate that the proposed model achieves excellent accuracy, with mean absolute errors below 1% for heart rate and approximately 2% for breathing rate, validating the effectiveness of the proposed approach for accurate, contactless monitoring of vital signs.

This paper investigates the problem of localizing a moving target by using asynchronous time-of-arrival (ATOA) measurements. The source’s moving velocity can be considered to be a constant within a sufficiently small observation period. Based on this assumption, we consider the use of ATOA measurements to jointly estimate the initial location and the velocity of the mobile source. We first formulate the maximum likelihood estimation (MLE) problem for such scenario. Since the MLE problem is difficult to solve analytically because of its highly non-linear and non-convex nature, we then develop a semidefinite programming (SDP)-based algorithm to solve the MLE problem. Because two equality constraints are relaxed into two inequality constraints, the solution provided by the SDP-based algorithm is still suboptimal. We thus develop an improved solution that takes the SDP-based solution as an initial value and iteratively calculates the unknown parameters in the form of fixed-point iteration (FPI). Finally, extensive simulation results are obtained to compare its performance with those of methods as well as the Cramér-Rao lower bounds (CRLB). Results show that the proposed algorithm is optimal and can reach the CRLB.

In this paper, we investigate the problem of improving noisy sensor positions using inter-sensor angle-of-arrival (AOA) measurements, which is highly non-linear. First, we present the Cramér-Rao lower bounds (CRLB) analysis to show that the incorporating of inter-sensor AOA measurements refines the accuracy of sensor positions. Second, we proposed two weighted least-squares (WLS) solutions to solve the problem. The one resorts to the Tikhonov regularization method as the formulated regressor is not a column full rank matrix, and the other one (called improved WLS solution) derived from the maximum likelihood estimator, avoids choosing regularization factor. Finally, simulation results show that the performance of the improved WLS solution is close to the CRLB and better than the regularization-based WLS solution irrespective of the choice of regularization factor.

Three-dimensional (3-D) target localization using one-dimensional (1-D) space angle (SA) measurements from linear arrays has recently gained significant attention. Each 1-D SA measurement defines a conical surface, and the intersection of multiple such surfaces determines the target's location in 3-D space. However, state-of-the-art methods for solving the 1-D SA localization problem are often either suboptimal or computationally intensive. In this paper, we propose a novel iterative weighted least squares (IWLS) algorithm to address the problem. To provide deeper insights, we present a geometric interpretation of the iterative process, highlighting its physical significance. Furthermore, we analyze the computational complexity of the proposed algorithm and compare it with existing methods. Simulation results demonstrate that the proposed algorithm not only achieves higher estimation accuracy but also requires less computational time compared to state-of-the-art approaches.

Recently, joint target and transmitter localization using differential time-delay (DTD) and angle-of-arrival (AOA) measurements has attracted researchers' interest. Due to the fact that three Euclidean norms exist in the DTD equation, the DTD equation is difficult to tackle directly. In this paper, we divide the joint localization problem into three subproblems, respectively, the AOA-only localization problem, the hybrid AOA and time-difference-of-arrival (TDOA) localization problem, and the hybrid AOA and time-delay (TD) localization problem with known transmitter location. Then, a two-stage algorithm is developed. In the first stage, solving the AOA-only localization problem provides initial estimates. In the second stage, alternatively and iteratively solving the problem of hybrid AOA and TDOA localization and the problem of hybrid AOA and TD localization provide the improved solutions. Simulation results validate that the proposed algorithm is superior to the existing constrained weighted least-squares (CWLS) algorithm when AOA noise variance is not sufficiently small.Index Term-Angle-of-arrival (AOA), differential time-delay (DTD), time-delay (TD), time-difference-of-arrival (TDOA), elliptic localization.

A recent research on a fixed point iteration (FPI) algorithm for elliptic localization has shown tremendous promise in terms of reduced implementation complexity than existing algorithms without sacrificing performance [1]. However, no theoretical analysis was provided in this study. In this article, we present a thorough theoretical analysis of the FPI algorithm’s convergence and performance. These analyses show that: 1) the optimum estimate of the maximum likelihood estimation (MLE) problem is found in a sphere; 2) the FPI estimator is unbiased and its covariance matrix achieves the Cramér-Rao lower bound (CRLB) matrix when the measurement noise is small enough.

This work presents a new weighted least-squares (WLS) method for source localization using angle-of-arrival (AOA) and time-difference-of-arrival (TDOA) measurements. Previous studies have shown that combining these two independent types of measurements yields more reliable results, as evidenced by lower Cramér-Rao lower bounds (CRLB) compared to using a single type of measurement. Our newly developed WLS algorithm has excellent performance when the source is close to the sensors but encounters challenges when the source is distant. To address these limitations, we develop an improved solution that incorporates a one-dimensional search process along the distance between the reference node and the source node. This enhancement significantly improves the algorithm’s performance in ‘far-field’ (i.e., when the source is far away from the sensors) scenarios. Our extensive simulation results demonstrate that the proposed algorithms substantially outperform existing methods in terms of root mean-squared error (RMSE), establishing a new benchmark for accuracy and reliability in hybrid AOA-TDOA localization.

This paper develops a simple and efficient method for source localization using signal time-of-arrival (TOA) measurements. There exist many TOA localization algorithms, most of which require matrix inversions. Their complexity often makes them unsuitable for cases when a large number of sources need to be localized simultaneously via many sensor nodes. The proposed algorithm, which requires addition, multiplication and taking root-squares only, has a lower complexity while it still performs better than existing schemes.

This paper investigates the problem of source localization using signal time-difference-of-arrival (TDOA) measurements in the presence of clock synchronization bias and sensor position errors. Our existing work has developed a unified solution for TDOA localization in the presence of sensor position errors but clock synchronization bias was not considered. Clock synchronization bias is a more complex problem often encountered in practical localization networks. This paper further generalizes this framework to include clock synchronization bias. The proposed technique employs multiple calibration emitters to simultaneously alleviate both the sensor position errors and clock synchronization bias. The maximum likelihood estimator (MLE) for this problem is optimal, but too complex to be applied in practice. We develop a semidefinite programming (SDP) based localization algorithm to effectively solve the MLE problem. This SDP algorithm can reach the Cramer-Rao lower bound when sensor position errors are not unrealistically large.

This paper investigates time-of-arrival (TOA) source node self-positioning with unknown clock skews in wireless sensor networks. For the source-to-anchor direction, source node clock skew does not affect the localization performance. When synchronized anchor nodes simultaneously transmit signals to a source node, the source node clock skew will degrade the localization performance. A semidefinite programming (SDP) algorithm that jointly estimates the source position and clock skew is proposed for the latter case. The proposed algorithm is better than the two kinds of existing schemes, namely, asynchronous TOA localization and TDOA localization. We also tune the algorithm to the case of anchor nodes position uncertainties. Simulation results validate the performance of the proposed algorithm.

This work presents a target localization scheme in distributed multiple-input multiple-output (MIMO) radar system using bistatic range measurements. The localization approach consists of two phases. First, measurements are divided into multiple groups based on the various transmitter and receiver elements. For each group, an approximate maximum likelihood (AML) estimator is proposed to estimate the location of a target. Then, the estimation results from these different groups are combined to form the final estimate. The performance of the proposed algorithm is validated by simulation and is shown to reach the Cramer-Rao lower bound (CRLB) in a range of measurement noise levels.

The most state-of-art time-difference-of-arrival (TDOA) localization algorithms are performed under the assumption that all the nodes are synchronized. However, for a widely distributed wireless sensor networks (WSNs), time synchronization between all the nodes is not a trival problem. In this paper, we study the problem of source localization using signal TDOA measurements in the system of nodes part synchronization. Starting from the maximum likelihood estimator (MLE), we develop a semidefinite programming (SDP) approach. Besides, we extend the SDP algorithm to the case of non-accurate sensor position. Simulation results validate the localization performance of the proposed SDP algorithms.

A new joint synchronization and localization method for wireless sensor networks using two-way exchanged timestamps is proposed in this paper. The goal is to jointly localize and synchronize the source node, assuming that the locations and clock parameters of the anchor nodes are known. We first form the measurement model and derive the Cramér-Rao lower bound (CRLB). An analysis of the advantages and disadvantages of a recent scheme on joint synchronization and localization motivates us to develop a maximum likelihood estimator (MLE) that effectively resolves the issues of this existing scheme. A novel semidefinite programming method is then proposed to transform the nonconvex MLE problem into a convex optimization problem. Extensive simulation results are obtained to compare the synchronization and localization performances of proposed scheme and a few state-of-the-art existing schemes.

In this paper, we consider using time-of-arrival (TOA) measurements from single moving receiver to locate a stationary source which emits periodical signal. First, we give the TOA measurements model and deduce the Cramér-Rao lower bounds (CRLB). Then, we formulated the maximum likelihood estimation (MLE) problem. We use the semidefinite programming (SDP) method to relax the nonconvex MLE problem into convex problem. It is shown that the original SDP algorithm can not provide a high-quality solution. We jointly add second-order-cone (SOC) constraints and penalty term to improve the tightness of the original SDP algorithm. Besides, we also consider the presence of receiver position errors, and develop the robust localization algorithm. Numerical simulations are conducted to demonstrate the localization performance of the proposed algorithms by comparing with the CRLB.