The surge in urban population has led to an imbalance between the demand of residents for travelling and available taxi resources in some specific spatial-temporal contexts. This paper delves into the utilization of reinforcement learning technology to enhance taxi dispatching, with a particular emphasis on optimizing passenger and driver satisfaction. The optimization objective is to maximize revenue while simultaneously minimizing waiting times. We introduce a novel dual-objective optimization system for taxi dispatching, employing reinforcement learning techniques. This system comprises three core modules of the traffic environment simulation module, the mathematical modeling module, and the RL-based dispatching optimization module. Employing a comprehensive approach, we specifically design reward models in reinforcement learning to ensure thorough optimization of taxi scheduling. Stability plays a pivotal role in addressing the intricacies of urban taxi scheduling, given the extensive variations in state and action spaces amidst dynamic environmental conditions. Our reinforcement learning model, based on A3C, streamlines strategy adaptation by learning a unified approach, thus bolstering algorithmic stability through gradient averaging across all agents.
The integration of LiDAR and camera data for 3D object detection in autonomous vehicles requires advanced fusion techniques that encompass both spatial and temporal dimensions. Our innovative framework, TempoFusion, implements a Temporal-Spatial Fusion Strategy to enhance the integration of multimodal data. This approach improves detection accuracy by leveraging the geometric precision of LiDAR and the detailed visual information from cameras. TempoFusion processes LiDAR and camera inputs separately, employing sophisticated fusion mechanisms to facilitate effective modality interaction. A significant innovation of our framework is the incorporation of a Temporal Fusion Layer (TFA) module. This module utilizes historical data to augment the contextual understanding of detected objects, overcoming the constraints of traditional fusion methods in dynamic settings. When evaluated on the nuScenes dataset, TempoFusion exhibits exceptional performance, especially in complex scenarios. It adeptly merges multi-frame data across spatial and temporal domains without diminishing the distinct benefits of each sensor type. Our work significantly advances 3D object detection accuracy, setting a new standard in autonomous vehicle technology.
Reference-based Super-Resolution (RefSR) has garnered significant attention for its capacity to leverage external a priori information. RefSR involves the intricate process of transferring texture details from a Reference (Ref) image to a LowResolution (LR) image, relying on corresponding pixel or patch relationships. Despite the proliferation of Convolutional Neural Network (CNN) or Transformer-based models aimed at enhancing RefSR performance, a substantial number of approaches neglect the distinct roles played by LR and Ref images within the reconstruction process. Specifically, LR images inherently contain fundamental structural information corresponding to High-Resolution (HR) images, while Ref images encapsulate potential high-frequency details. In this paper, we introduce a novel module for channel-based adaptive fusion, specifically designed to integrate features extracted from LR and Ref images. The proposed module adeptly combines LR and Ref image features along the channel dimension, enabling the efficient utilization of Transformer long-range modeling capabilities across the width and height dimensions of the feature map. The incorporation of this innovative module results in superior performance compared to state-of-the-art Transformer-based methods, concurrently demonstrating an improvement in inference speed. Rigorous experimental results validate the efficacy of our approach, Channelbased Adaptive Fusion with Transformer for Reference-based Super-Resolution (CAFformer), outperforming existing methods in both quantitative and qualitative evaluations. This contribution holds promise for advancing the field of superresolution reconstruction by comprehensively addressing the inherent distinctions between LR and Ref images.
Underwater wireless optical communication (UWOC) has been regarded as one of the promising solutions to underwater wireless communication systems due to its advantages of high bandwidth, fast transmission and good confidentiality. In UWOC systems, after emitting from light source, photons will be scattered with random deviated angles before propagating to the receiver plane. Hence, the light beam suffers a spatial angle spread at the receiver side. In order to improve the system performance with an appropriate receiver design, it is of significance to figure out the angle of arrival (AOA) distribution of the received light beam. In addition, the UWOC channel is susceptible to other effects such as absorption, turbulence and bubbles, especially when it is exposed to the complex ocean environment. Therefore, analyzing the impacts of these factors on AOA distribution is also essential to evaluate UWOC system performance. The existing studies of AOA distribution only considered single scattering component, which is not practical for turbid water. Furthermore, there are few studies focusing on multi-source scenarios. In this paper, we first present a simple expression for AOA distribution with single light source, taking both single and multiple scattering components into account. Then we extend the work to multi-source scenarios and derive the corresponding closed-form expression of AOA distribution. Compared to the traditional single scattering case, numerical results show that the proposed AOA distribution can fit well with Monte Carlo simulation results with various water types, link distances, and the characteristics of actual light sources.
Underwater wireless optical communication (UWOC) highly depends on the alignment between the receiver and transmitter in realization. In practice, the facts that laser sources have narrow divergence angles, water body suffers from fluctuation, and precise positioning is very difficult in the underwater environment bring great challenges to the link alignment as well as practical implementation. However, due to the intrinsic optical properties of seawater, photons will be scattered during the transmission process, resulting in the diffusion of light, which relaxes the requirements for strict alignment. Inspired by the uniform spatial distribution of the multi-source arrays with a close spacing, in this paper, we investigate the bit error rate (BER) performance of the multi-input single-output (MISO) laser links in the presence of receiver offset. Based on a system model by considering scattering and absorption effects and noises including background noise and blackbody radiation as well as OOK signaling and an ideal photon counter, we derived a closed-form expression of relationship among BER, transmit power, link range, receiver aperture and offset distance, which is verified by Monte Carlo simulations. Numerical results suggest that, regardless of water types, linear light source arrays parallel to the offset direction can improve the tolerance of receiver offset with specific transmit power and link range for reliable communications. On this basis, we compared the anti-offset performance in the case of different inter-spacings of light source arrays such as dual-source and three-source schemes and also derived the optimal inter-spacing for light source to maximize the acceptable offset distance with reliable communication for dual-source links.
In this paper, a 4×4 fixed-scale multi-input multi-output (MIMO) underwater optical wireless communication (UOWC) system is implemented with spatial multiplexing to achieve 20 Mbps rate over 2-m link distance which consists of 1-m free space and a 1-m water-filled tank. Experimental results suggested that the fixed-scale MIMO UOWC system shows more robustness than the conventional counterpart against our experimental setup.
KEYWORDS: Orthogonal frequency division multiplexing, Adaptive optics, Wireless communications, Berkelium, Receivers, Radio optics, Data communications, Telecommunications, Signal to noise ratio, Modulation
Orthogonal frequency division multiplexing (OFDM) has been applied to optical wireless communication to achieve high data rates and wide bandwidth and solve the issues of inter-symbol interference (ISI). Due to the non-negative and real-valued characteristics of intensity modulation/direct detection (IM/DD) signaling for optical wireless communication systems, some optical OFDM schemes have been proposed to reach the requirements. Among these optical OFDM schemes, adaptively biased OFDM (ABO-OFDM) reserves 1/4 of the subcarriers and utilizes Hermitian symmetry to generate non-negative and real values after invert fast Fourier transformation (IFFT) and adding bias in time domain. In this paper, we extend the original ABO-OFDM scheme and proposed a generalized ABO-OFDM scheme in which 1/m subcarriers are reserved for any positive integer m. It is demonstrated that the bias added in time domain can be counteracted exactly in frequency domain, which means it has lower implementation complexity at the receiver than most of the other optical OFDM schemes. This generalized ABO-OFDM scheme has higher frequency efficiency and lower peak-to-average power ratio (PAPR) than asymmetrically clipped optical OFDM (ACO-OFDM) and higher power efficiency than direct current biased optical OFDM (DCO-OFDM). We also evaluate the impact of parameter m on system performance in terms of PAPR and bit error rate (BER). Consider the comparison with other optical OFDM schemes and the tradeoff above-mentioned, the generalized ABO-OFDM is a potential scheme to facilitate optical wireless communications with a flexible parameter m.
Underwater wireless optical communications (UWOC) could transmit data using blue or green light beams with high data rate and safety in a relatively short range. OFDM based UWOC systems are able to further increase data rate, however, highly dependent on the accuracy of channel estimation. In this paper, we consider the channel estimation problem for OFDM based UWOC systems. We firstly apply Monte Carlo simulation to obtain the channel impulse response (CIR) of UWOC links under different conditions to facilitate the design of the subsequent OFDM systems. Secondly, we evaluate the pilot-based least squares (LS), and two types discrete Fourier transform (DFT) channel estimation methods and compare their performance. Numerical results have suggested that the temporal pulse spread strongly degrades the performance of the channel estimation. These two DFT methods especially DFT channel estimation with noise threshold method achieved the best performance among these prior works. While for the signal-to-noise ratio (SNR) less than 10 dB, the performance of DFT with noise threshold method is still poor. To solve this problem, we propose a new channel estimation approach of DFT with adaptive noise threshold (DFT-ANT) which adaptively adjusts the noise threshold based on SNR, and analyze its complexity and normalized mean square error (NMSE) performance in underwater environment. Numerical results have validated the proposed approach which outperforms existing channel estimation methods especial DFT with noise threshold method in terms of accuracy for various water types.
In order to evaluate the performance of underwater wireless optical communication (UWOC) systems, it is of significance to fully understand the impact of spatial diffusion of light beams. Meanwhile, simple and highly adaptable spatial channel modeling is also necessary and essential for performance evaluation and system design. In this paper, we focus on the spatial channel modeling and, in particular, quantify the photon spatial distributions for different water types, link distances, and transmitter/receiver characteristics. Via using the Gaussian distribution to complete the fitting, we have proposed a simple expression to describe the spatial irradiance distribution. The numerical results have shown that the proposed spatial channel model for UWOC systems agrees well with the Monte Carlo simulation results in terms of mean square error (MSE) with or below the order of 10−7 in both turbid coastal and harbor water and demonstrates a high adaptability to the link conditions. Furthermore, on this basis, we extend the study from single source to multi-source scenario and derive the corresponding expression of spatial channel model. Considering the integrity of closely spaced multi-source array, the multi-source model has been further simplified by two-dimensional Gaussian fitting.
In this paper, we have fabricated and packaged a blue micro-LED with a diameter of 50-μm based on a single layer of InGaN QD micro-LED and present a new method to calculate the junction capacitance of micro-LEDs under forward voltage using the forward AC small-signal method. The results confirm that QD micro-LEDs, like commercial LEDs, show obvious negative capacitances at low frequencies and large voltages. The values of negative capacitance at high frequency and low voltage are so small and can be ignored, or there is no negative capacitance. We have also concluded the empirical expressions for negative capacitance, voltage, and frequency.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.