The rapid expansion of low earth orbit (LEO) satellite networks exposes critical limitations in conventional resource allocation schemes, which are unable to simultaneously optimize spectral utilization and adapt to heterogeneous traffic patterns under extreme mobility, necessitating a joint beam-frequency dynamic coordination framework. To address the challenges of multi-beam LEO systems, this paper introduces a hybrid framework that synergizes adaptive beam activation patterns and spectrum reuse optimization, enhanced by a deep reinforcement learning (DRL)-driven coordination mechanism for resource allocation in time. By dynamically adjusting beam activation patterns and frequency allocation, the framework optimizes spatial-temporal resource utilization while mitigating co-channel interference. Angular-constrained multi-criteria clustering achieves dynamic beam-user mapping with low-complexity adaptation for LEO mobility. The DRL-based component further coordinates multi-dimensional parameters, including transmit power and sub-channel assignment, to balance throughput and latency under time-varying channel conditions. Extensive simulations validate the framework’s capability to maintain high spectral efficiency and coverage performance across diverse scenarios, outperforming conventional static allocation methods. The results highlight its adaptability to dynamic traffic patterns and scalability for large-scale deployments, providing a robust foundation for next-generation LEO systems.

This paper presents a novel beamforming design that seamlessly integrates sensing, communication, and over-the-air computation (AirComp), enabling a critical multi-purpose functionality for next-generation wireless networks. Firstly, we formulate an optimization problem with the objective of minimizing the mean squared error of AirComp, subject to constraints that ensure the performance of both sensing and communication. The optimization problem is then parameterized and solved using unsupervised learning, employing real-valued and complex-valued deep neural networks (DNNs), respectively. For the complex-valued DNNs, we introduce its mechanism and then apply it for an intelligent beamforming design. Numerical results validate the convergence, ergodic rate, and ergodic mean square error of the proposed algorithms for the integrated sensing, communication, and computation. Also, our findings show that complex-valued DNNs outperform real-valued DNNs.

This paper proposes a coordinated beamforming scheme for a multi-cell integrated sensing and communication (ISAC) system. A target-centric graph is constructed, with the coordinate system centered at the detection target. Specifically, base stations (BSs) and users are represented as nodes, with their coordinates serving as node features, while channel realizations between nodes are modeled as edge features. To evaluate sensing performance, the Neyman-Pearson detector is employed to compute the detection probability for a fixed false alarm probability. The optimization problem is formulated to maximize the detection probability of the target at the origin while ensuring QoS communication requirements and satisfying the transmit power budget. This sensing-centric problem is addressed using graph neural networks (GNNs), which generate parameterized policies for coordinated beamforming. The GNNs are trained via primal-dual approach, leveraging a small duality gap for efficient convergence. Additionally, specific layers are utilized to generate user association policies for communication links, enabling efficient processing of the graph-structured data after sparsity enhancement. Simulation results validate the feasibility and effectiveness of the proposed GNN-based approach in achieving high detection probability.

The emerging low-altitude economy leverages airspace below 1000 meters for intensive commercial and social aerial activities, where integrated sensing and communication (ISAC) service in 6G network for aircraft is critical to ensuring safe and efficient operations. However, aircraft often operate under constrained wireless resources, particularly in areas with limited or no network coverage. In such settings, exhaustive sensing and data communication among neighboring nodes lead to uncoordinated competition and prohibitive overhead. This paper investigates the ISAC service modeling and scheduling in aerial-vehicle-to-everything (Aerial-V2X) networks, which provides continuous high-accuracy sensing without compromising communication throughput under constrained resource budgets. First, we design a reconfigurable ISAC waveform tailored for aircraft, enabling flexible resource partitioning for both cellular communication (aerial vehicle-to-infrastructure, A-V2I) and cooperative sensing (aerial vehicle-to-vehicle, A-V2V). Second, we formulate a joint sensing-communication service model under this waveform and cast the optimization problem in a partially observable Markov decision process (POMDP). Third, a multi-agent deep reinforcement learning (MADRL) approach is developed to perform decentralized scheduling of sensing actions for each aircraft, minimizing time-frequency resource consumption. Experiments and trace-driven evaluations demonstrate that the proposed method can reduce sensing overhead by up to 70% compared to benchmark policies, while maintaining satisfactory communication and sensing performance.

The possibility of jointly optimizing location sensing and communication resources, facilitated by the existence of communication and sensing spectrum sharing, is what promotes the system performance to a higher level. However, the rapid mobility of user equipment (UE) can result in inaccurate location estimation, which can severely degrade system performance. Therefore, the precise UE location sensing and resource allocation issues are investigated in a spectrum sharing sixth generation network. An approach is proposed for joint subcarrier and power optimization based on UE location sensing, aiming to minimize system energy consumption. The joint allocation process is separated into two key phases of operation. In the radar location sensing phase, the multipath interference and Doppler effects are considered simultaneously, and the issues of UE’s location and channel state estimation are transformed into a convex optimization problem, which is then solved through gradient descent. In the communication phase, a subcarrier allocation method based on subcarrier weights is proposed. To further minimize system energy consumption, a joint subcarrier and power allocation method is introduced, resolved via the Lagrange multiplier method for the non-convex resource allocation problem. Simulation analysis results indicate that the location sensing algorithm exhibits a prominent improvement in accuracy compared to benchmark algorithms. Simultaneously, the proposed resource allocation scheme also demonstrates a substantial enhancement in performance relative to baseline schemes.

The benefits of mobile edge computing (MEC) systems cannot be fully exploited when the communication link is blocked or the communication signal is weak. Intelligent reflective surface (IRS) technology is introduced to build an IRS-assisted MEC system and to solve this issue. In this paper, devices offload part of their computing tasks to MEC through the multi-antenna access point with the help of the IRS, thereby reducing the completion time of computing tasks. We consider the weighted sum-latency minimization for single-device and multi-device scenarios in the uplink, which are constrained by computing offload allocation, edge node computational capability, IRS practical phase shift, beamforming, and device transmitting power. Firstly, block coordinate descent technology is used to decouple latency minimization problem into two subproblems of computation and communication. Secondly, in single-device scenario, the original problem is simplified and solved by the continuous refinement scheme. In multi-device scenario, an algorithm that combines alternating optimization and the Jaya algorithm is proposed for the first time to solve the weighted sum-latency minimization problem. In addition, compared with the conventional MEC systems without IRS, the effectiveness and high-performance gain of the proposed algorithm are proved through simulations.

Deep learning (DL) models for change detection (CD) are affected by the changed/unchanged and hard/easy sample imbalance during the training process. Most of the loss functions for solving the sample imbalance problem are a static loss which is difficult to adapt to the variation of data distribution. In this paper, we propose a dynamic method termed dynamic adaptive focal loss function (DAFL). Specifically, we first statistically count the number of changed/unchanged samples in different batches of training data, and a dynamic weighting factor is constructed to dynamically and adaptively balance their proportions. Furthermore, a dynamic modulation factor is proposed to suppress the hard/easy sample imbalance. In addition, we employ a CD model based on Progressive Scale Expansion Network (PSENet), which is trained by using DAFL for remote sensing images. Experimental results on three CD datasets (CDD, SYSU-CD and LEVIR-CD) indicate that DAFL outperforms all baseline approaches. Our proposed method achieves the maximum improvement, with an F1-score of 0.33%,0.8% and 0.94% for sufficient sample size, and 2.15%, 2.61% and 3.89% for small-sample size, respectively. This advancement is crucial for the application of CD, which provides an alternative method for solving sample imbalance in the condition of varying sample size, especially small sample condition that is common in the real scenario.

The future integrated sensing and communication (ISAC) networks is expected to equip with sufficient computation resources. However, current research focuses on single-domain resource allocation in ISAC and computing force networks, leaving the joint optimization of sensing, communication, and computation resource allocation unexplored. In this paper, we propose a novel approach to this problem by deep incorporating computation resources, combined with a federated learning framework, while considering sensing precision and power consumption. Firstly, a multi-objective optimization is designed, involving Cramer-Rao Bound, sum rate of ISAC networks, and power consumption of computing force networks. Subsequently, the multi-objective optimization is transformed into a multi-task learning model. We aim to obtain joint optimization of sensing, communication, and computation resource allocation via deep learning techniques. Towards the multi-task learning model, the multiple-gradient descent algorithm is utilized to obtain the multi-objective optimization. Furthermore, a practical low-complexity the multiple-gradient descent algorithm is developed to reduce the computational cost. Finally, the effectiveness of the proposed deep learning algorithms is verified by simulations results.

Unmanned aerial vehicle (UAV) and sidelink technology are becoming more and more important in emergency communication. To optimize overall energy efficiency, joint subchannel, transmit power, and multi-UAV trajectory optimization algorithms are examined in a multi-cell heterogeneous network of UAV and sidelink with quality of service (QoS) sensitivity restrictions. To allocate subchannel appropriately in each period, a grouping and matching approach is first developed that can handle the subchannel assignment of multi-cell. Then, successive convex approximation method is used to approximate the non-deterministic polynomial hard problem of power allocation. Taylor expansion approximation method is finally introduced to deal with multi-UAV trajectory optimization. In addition, the complexity analysis is provided and numerical results confirm the optimization methods’ reasonableness.

One of the main challenges of popularizing the integration of sensing and communication (ISAC) network is mutual interference between the two functions. A viable solution is the time division scheme where communication and sensing are separated in time domain. This paper considers a multi-cluster ISAC network model, where the time-domain radio resources are allocated to sensing and communication. At the same time, the time resources can be reused in space-domain. Particularly, the terminals in different work phases can access radio resources of different or the same clusters simultaneously, depending on interference. In this way, the interference is isolated while the resources utilization is improved. Two different resource allocation approaches are proposed according to how interference is considered. The aim is to maximize the sensing detection probability under the constraint of network throughput. The performance improvement in terms of target detection probability brought by the proposed schemes is shown by numerical results compared with benchmark methods.

Earth observation technology has improved the detection of land cover changes. However, current pixel-based change detection methods cannot adequately describe the evolutionary process and spatiotemporal association of geographic entities. Therefore, we developed a method for analyzing the processes and patterns of land cover evolution based on spatiotemporal graphs. First, a spatiotemporal graph was generated from a time series of land cover maps according to the spatial and temporal relationships between land cover objects, as defined by spatial adjacency and temporal transition, respectively. Subsequently, structural characteristics, such as the spatial roles, adjacency type, temporal transitions and evolution trajectories, were derived from the spatiotemporal graph to describe and analyze the evolution of land cover. Finally, this method was applied to analyze land cover evolution in Fujian Province, China, from 2001 to 2019. The proposed method not only completely preserves the spatial adjacency and temporal transition details among land cover objects in a spatiotemporally unified graph framework but also extracts evolution-related spatiotemporal structural characteristics. This study provides a reliable scientific basis for analyzing the consistency of long-term land cover dynamics and has practical value for other geographic applications.

Many real-time application scenarios are developed in 6G communications. Driven by the low-latency data processing requirements, multi-tier computing has become an important technology to improve user experience and reduce network overhead. In this paper, we consider a multi-tier computation offloading network structure for 6G applications, in which the cloud computing center and the nearby vehicle edge server (VES) are able to partially calculate the tasks offloaded from the user equipment (UE), and the remaining task is processed locally in the UE. By jointly optimizing user scheduling, cloud offloading ratio, VES offloading ratio, and VES mobility, the objective function is to minimize the delay of the system transmission and computation under the constraints of discrete variables and energy consumption. To solve the problem, a primal-dual deep deterministic policy gradient (PD-DDPG) algorithm based on multi-tier computation offloading is proposed. Simultaneously, compared with baseline algorithms, PD-DDPG algorithm has an obvious advantage in both the speed of convergence and the system delay.

The digital twin (DT) and terahertz (THz) wireless communication technologies have promoted the innovative development and application of 6G networks. Combining DT can obtain efficient, collaborative, and intelligent management for THz wireless networks. However, the conflicts between large amounts of twin data and limited network resources make it difficult to improve the performance of DT networks. In this paper, a DT architecture for THz wireless networks is proposed, which maps a physical network in the THz band into a virtual DT network and represents the DT network as a graph structure. Furthermore, the THz channel model is provided, and the resource management problem with weighted mean rate as the optimization objective is proposed, which is transformed into a graph optimization problem. Based on this, a distributed message propagation algorithm is proposed, which uses the graph neural network to provide a solution. Simulation results show that the proposed scheme improves the weighted mean rate of the DT network for the THz band and outperforms the benchmark methods. It is also proved that the proposed distributed message propagation algorithm is scalable and can maintain good performance under different conditions.

Traffic control is regarded as a key issue to alleviate congestion in internet of vehicles (IoV) machine-type communications (MTC). Recently, many traffic control schemes have been studied, such as access class barring (ACB) scheme and back-off (BO) scheme. However, the dynamics of traffic and the heterogeneous requirements of different IoV applications are not considered in most existing studies, which is significant for the random access resource allocation. In this paper, we consider a hybrid scheme, combining the priority dynamic ACB (PDACB) scheme and BO scheme. The IoV devices are classified depending on different delay characteristics, where the delay-sensitive devices are classified as high priority. The target is to maximum the successful transmission of packets with the success rate constraint by adjusting the various ACB factors. Proximal policy optimization (PPO) algorithm as a unique deep reinforcement learning (DRL) method is utilized in this paper, which can obtain continuous action space and solve for the optimal ACB factors without estimating backlog of nodes. A quick convergence is achieved by designing sensible state space, action space and reward. The access capability of the PDACB traffic control scheme is verified by simulations.

Nowadays, the multi-access interference problem in the ISAC systems can not be ignored. The study on interference management in ISAC has been envisioned as one of key technologies to support ubiquitous sensing functions. Different from the current work, a communications-sensing-intelligence converged network architecture is proposed to coordinate interference in this paper. Each base station equips with the individual deep neural networks to allocate power and beamforming. On this basis, the interference management is transformed into a functional optimization with stochastic constraints. An unsupervised learning algorithm is proposed to allocate power for interference management. Furthermore, a transfer learning method is presented to obtain the interference management in terms of transmit beamforming. Finally, the distributed management is obtained from the local channel state information in the multi-cell scenario. Simulation results verify the effectiveness of the proposed unsupervised learning interference management method in the ISAC systems.

With the development of internet of things, latency-sensitive applications such as telemedicine are constantly emerging. Unfortunately, due to the limited computation capacity of wireless user devices, the real-time demands can not be met. Multi-access edge computing (MEC), which enables the deployment of edge access points (E-APs) to support computation-intensive applications, has become an effective way to meet the real-time demands. However, the number of WUDs that E-APs can serve are limited. To increase system capacity, the unmanned aerial vehicle (UAV) assisted computation offloading architecture in the terahertz (THz) band is proposed. In this paper, the problem of UAV placement optimization, resource allocation, and computation offloading is investigated considering the quality of service and resource constraints. The joint optimization problem is non-convex and hard to be solved in time by using traditional algorithms, such as successive convex approximation. Therefore, deep reinforcement learning (DRL) based approach is a promising way to solve the formulated non-convex problem of minimizing latency. Double deep Q-learning (DDQN) and deep deterministic policy gradient (DDPG) algorithms are provided to search for near-optimal solutions in highly dynamic environments. The effectiveness of the proposed algorithms is proved by simulation results in different scenarios.

With the rapid development of mobile services and applications, the transmitting of massive data makes low-cost communication a challenge. Edge-based wireless communication technology is developed to be a promising approach to satisfy the communication requirements. Edge caching technology is one of effective methods to reduce the overhead of communication system and the pressure of backhauls. In this paper, the joint optimization problem of caching placement and power allocation in virtualized low earth orbit (LEO) satellite-terrestrial networks is proposed, which is based on cooperative caching, by considering cache size limits and power constraints. The optimization problem is solved using an algorithm inspired by the courtship movements and random flights of mayflies. Simulation results show the effectiveness of the proposed scheme in improving system performance and reducing power consumption.

Due to the rapid growth of data transmissions in internet of vehicles (IoV), finding schemes that can effectively alleviate access congestion has become an important issue. Recently, many traffic control schemes have been studied. Nevertheless, the dynamics of traffic and the heterogeneous requirements of different IoV applications are not considered in most existing studies, which is significant for the random access resource allocation. In this paper, we consider a hybrid traffic control scheme and use proximal policy optimization (PPO) method to tackle it. Firstly, IoV devices are divided into various classes based on delay characteristics. The target of maximizing the successful transmission of packets with the success rate constraint is established. Then, the optimization objective is transformed into a markov decision process (MDP) model. Finally, the access class barring (ACB) factors are obtained based on the PPO method to maximize the number of successful access devices. The performance of the proposal algorithm in respect of successful events and delay compared to existing schemes is verified by simulations.

Random access schemes in satellite Internet-of-Things (IoT) networks are being considered a key technology of new-type machine-to-machine (M2M) communications. However, the complicated situations and long-distance transmission can make the current random access schemes not suitable for the satellite IoT networks. The random access problem in the satellite IoT networks is studied in this article. A novel random access scheme for machine-type-communication devices (MTCDs) is proposed, to maximize the efficiency of random access for contention-based and contention-free random access. Under the set of random access opportunities (RAOs) and limited delay, the random access control model is designed via maximizing efficiency of random access. The model-free deep reinforcement learning (DRL) algorithm is proposed to tackle the problem based on the random access model. Subsequently, the deep Dyna-$Q$learning algorithm is introduced to deal with the proposed random access control model. In this proposed scheme, the random access model-free DRL algorithm is developed using simulated experience. The proposed algorithms’ performances are discussed, and simulation results show the desirable performance of the proposed DRL methods on different system parameters.

The use of cell-free massive multiple-input–multiple-output (MIMO) is regarded as a novel technique in the Industrial Internet of Things (IIoT) networks, and many studies have been reported on its cross-layer optimization, including random access and power allocation. Nevertheless, the cooperation of deep reinforcement learning (DRL) and cell-free massive lacks of deep study. In this article, a primal–dual deep deterministic policy gradient (DDPG) algorithm is designed to obtain cross-layer radio resource management, including power allocation in the physical layer and random access in the medium access layer. Different from the current studies, the random access and power allocation is formulated in cell-free massive MIMO IIoT networks, utilized by the stochastic ergodic optimization. In contrast to the stochastic policy gradient algorithm, a primal–dual DDPG algorithm is designed for the cross-layer optimization. Moreover, a multiagent primal–dual DDPG algorithm is proposed to different scenarios in the cell-free massive MIMO IIoT networks. Simulations are presented to verify the effectiveness of the primal–dual DDPG algorithm for random access and power allocation in the cell-free massive MIMO IIoT networks.

In this paper, an IRS-aided integrated sensing and communications (ISAC) system operating in the terahertz (THz) band is proposed to maximize the system capacity. Transmit beamforming and phase-shift design are transformed into a universal optimization problem with ergodic constraints. Then the joint optimization of transmit beamforming and phase-shift design is achieved by gradient-based, primal-dual proximal policy optimization (PPO) in the multi-user multiple-input single-output (MISO) scenario. Specifically, the actor part generates continuous transmit beamforming and the critic part takes charge of discrete phase shift design. Based on the MISO scenario, we investigate a distributed PPO (DPPO) framework with the concept of multi-threading learning in the multi-user multiple-input multiple-output (MIMO) scenario. Simulation results demonstrate the effectiveness of the primal-dual PPO algorithm and its multi-threading version in terms of transmit beamforming and phase-shift design.

The remote sensing time-series change detection algorithm based on the pixel or single landscape patch ignores the change analysis of spatial structure information. Inspired by graph convolutional network (GCN) modeling, a set of landscapes (nodes) and their relationships (edges) are proposed. In this study, a parallel strategy in spatial GCN and progressive strategy in temporal GCN, called the hybrid GCN model network as a holistic framework, was proposed to accurately capture both spatial and temporal variations in landscape patterns based on yearly Landsat time series. A super-patch (i.e., fixed patch class surrounded by a one-hop neighbor patch) was selected as the input of the GCN network. First, a spatial GCN model with three parallel graph convolutional layers was adopted to classify landscape pattern types. Three dominant categories of landscape patterns over the past three decades have been identified. Second, four landscape metrics in super-patches were proposed for the quantitative characterization of changes in landscape patterns. Finally, a temporal GCN model with two progressive graph convolutional layers was used to detect six types of patch changes, which were applied to continuously detect the landscape pattern evolution processes. Regardless of the spatial GCN and temporal GCN, they provided satisfactory performance using the training and validation sets with overall accuracy > 92% and Kappa coefficient > 0.90, and loss values converging to 0.049 and 0.128, respectively, based on NLLLoss function until 500 epochs. It is believed that the hybrid GCN model has great potential for mining possible implicit spatiotemporal relationships and future evolution of landscape patterns.

Optical remote sensing time series are optimal for understanding and monitoring biochemical changes of key phenological parameters, which is essential for the assessment of vegetation health. However, due to cloud contamination, optical images often lack several days’ to months’ worth of information. Therefore, reconstructing optical time series based on the synthetic aperture radar (SAR) is necessary, which has the advantage of the production of continuous images under all weather conditions. In this study, an improved sequence-to-sequence (Seq2Seq) model was proposed, which integrated teacher forcing (TF) with the attention mechanism to handle input and output time series with unequal lengths. The proposed model could be used to reconstruct optical full-band time series using SAR time series data and reduce sample requirements based on the modification of the loss calculation method. To explore the spatiotemporal scalability of the proposed model, the test samples were divided into three categories: same time but different spatial domain, different time but same spatial domain, and both different temporal and spatial domains. The major findings can be summarized as follows: 1) 72.7% of the generated Landsat 8 sequence values had an absolute error of less than 0.05; 2) The mean absolute error of all bands was less than 0.0812; 3) The mean squared errors of all samples were lower than 0.015 regardless of the type of the test sample; and 4) TF was introduced to the model to improve the accuracy of the generated Landsat 8 sequence, yielding an increase of more than 5.97%. We concluded that the proposed model had good robustness both temporally and spatially. It performed well in the reconstruction of optical image time series, provided a basis for the use of time series pairs with unequal lengths, and can be applied to cloud removal and vegetation index reconstruction.

Detecting the evolution of large-area landscape patterns using long-term remote-sensing images is helpful in supporting research on the relationship between landscape patterns and ecological processes, as well as the development of ecological process simulations and spatiotemporal interaction models. However, detection methods have generally been developed as separate applications, each with a separate type of landscape pattern change; remote-sensing images are acquired at epochal timesteps. Consequently, in practical applications, many omission changes for some types of pattern changes and inaccurate evolution time are presented in the detected map. In this article, state-and-evolution detection models (SEDMs) are promoted to obtain complete information about the evolution of landscape patterns based on yearly land cover data. In the proposed framework, we first define the major categories of landscape pattern changes to comprehensively reveal the characteristics of landscape pattern changes associated with real change cases. Next, a morphological rule-based pattern recognition approach is proposed for quantitative discrimination among these categories. This approach is then applied in annual land cover data to continuously detect landscape pattern evolution processes and evolution time. Finally, the detected evolution time in different evolution processes is applied to measure the timestep between two disparate types. The performances of the SEDMs are presented by Landsat-derived land cover evolution in Shanxi, China. The detected results are indirectly verified by the land cover conversion matrix and connect index, indicating strong robustness and generalization ability of the SEDMs.

Mapping forest disturbances using dense time series can timely identify disturbances at the subannual scale. However, these change detection methods using dense time series may be infeasible when not enough temporal observations are available. In this article, an online change detection algorithm that identifies forest disturbances at a subannual scale using spatial context from the sparse Landsat time series was proposed. First, the spatial normalized index that removed forest seasonality was prepared for establishing a simplified model instead of the harmonic model, thereby reducing the requirements for a high temporal frequency of clear observations for model initialization. Second, by using the spatial errors model to establish the simplified model, the normally distributed residual time series that removed the spatial autocorrelation were obtained. Third, the spatial statistic$t$time series transformed from residual time series within a$3\times3$spatial window were subsequently subjected to the exponentially weighted moving average$t$chart (EWMA-t), which is a statistical process control chart for a short cycle corresponding to sparse Landsat time series. Fourth, disturbed pixels were labeled if the chart values persistently deviated from the control limits of the chart. The proposed algorithm was applied to a subtropical forest with low Landsat data availability and yielded an overall accuracy of 86% in the spatial domain and temporal accuracy of 93.7%, achieving accurate and timely identification of forest disturbances. The proposed method called the EWMA-t change detection (EWMATCD) algorithm provides an alternative for disturbance detection at the subannual scale in regions with low data availability.

Terahertz communications is regarded as a promising technology due to its higher bandwidth and narrower beamwidths, which can improve capacity and coverage for indoor wireless users. In this paper, the intelligent reflecting surface (IRS) technique and non-orthogonal multiple access (NOMA) are utilized to compensate drawbacks of indoor transmission mismatch in the terahertz band. Then wireless resource allocation optimization in indoor terahertz IRS-aided systems is transformed into a universal optimization problem with ergodic constraints. With the aid of parametrization features of deep neural networks (DNNs), proximal policy optimization (PPO) is adopted to train the policy and corresponding actions to allocate power and bandwidths. The actor part generates continuous power allocation, and the critic part takes charge of discrete bandwidths allocation. In the design of a deep reinforcement learning (DRL) framework, primal dual ascent is proposed to realize model-free training. Simulation results demonstrate the effectiveness of the primal dual PPO learning algorithm in different settings.

The combination of resource allocation and fog computing based radio access network (Fog-RAN) have great potential for future wireless networks. However, the cross-tier interference in the spectrum-sharing deployment of Fog BSs could affect the network performance seriously and most of the solutions focus on the spectral efficiency optimization. In this paper, the user association, caching strategy, and power allocation are investigated in Fog-RAN with consideration of energy efficiency and cross-tier interference mitigation. The user association, caching, and power allocation are formulated as a non-convex optimization problem and then transformed into a convex problem, which is solved by Alternating Direction Method of Multipliers (ADMM). Then ADMM-based resource allocation algorithms are proposed to improve the energy efficiency of Fog-RAN. Simulation results demonstrate the proposed algorithms's convergence and effectiveness by comparing with existing method.

The purpose of this study focused on integrating spectral indices with spatio-temporal characteristics to identify multi-stressor in crops. The experimental areas are located in Dongting Lake (DL), Hunan Province, China. Multitemporal Sentinel-2 (S2) images in 2016, 2017 were collected. Red-edge chlorophyll index (CIred-edge), rededge position (REP), normalized difference red-edge 2 (NDRE2) were calculated. The coefficients of spatiotemporal variation (CSTV) from spectral indices allowed us to discriminate crops exposed to pollution from heavy metal as well as environmental stressors. The results indicated that three indices were good indicators for identifying different environmental stressor in agriculture ecosystem. Crops under heavy metal stress remained stable with lower CSTV values, while crop `hot spots' (with greater CSTV values) were affected by abrupt stressors (i.e., pest and disease, drought) at some growth stage. It concluded that spectral indices and spatio-temporal characteristics show promise for monitoring crops with various stressors.

Accurate assessment of Fraction of Absorbed Photosynthetically Active Radiation (FAPAR) in large scale is significant for crop productivity estimation and climate change analysis. The object of study is to simulate FAPAR in the rice growth period for exploring photosynthetic capacity of rice in large-scale. The daily FAPAR is calculated based on a coupled model consisting of the leaf-canopy radiative transfer model (PROSAIL) and the World Food Study Model (WOFOST). Due to the limitation of the PROSAIL and WOFOST model, we introduced the remote sensing data assimilation method, which assimilated the Normalized Difference Vegetation Index (NDVI) into the coupled model, to improve the prediction accuracy and carry out the large-scale application. The results show high correlation between the simulated FAPAR and the measured data, with the determinate coefficient$(R^{2})$of 0.75 in the study area. The spatial distribution of FAPAR is uniform in flat area, which indicates that the rice in the whole study area has well growth condition and photosynthetic capacity. This study suggest that the coupled model (PROSAIL + WOFOST) assimilated with remote sensing data could accurately simulate daily FAPAR during the crop growth period.

The methods of the image noise reduction based on optimal channel‐processing to enhance image quality are studied. According to different noise with the different ratio in colour components, different noise reduction technologies are used for noise reduction. Then the optimal algorithm is automatically selected as the ultimate way of noise reduction in each channel. For the purpose of optimal noise reduction effect, this study presents a method of combining the quadratic optimisation with the variable window processing. The quadratic optimisation provides a good environment for noise reduction by decreasing complexity of mixed noise and the variable window processing calibrates the image smoothing result. Compared with mean filtering, median filtering and adaptive filtering, the image quality processed by the proposed algorithm is generally improved by >2 dB.

Heavy metal stress will damage the normal growth and change the bio-parameters like chlorophyll, nitrogen and water content of rice. Simple, speedy, non-invasive test of heavy metal stress to crops means a lot to the agriculture and even the suspend of human being. Many traditional spectral index computed using hyperspectral data can inverse the above bio-parameters. While it is found that some indies do not work well when the heavy metal stress exists, however, some indies still has ability to estimate the above three bio-parameters. In this paper, 28 vegetation Indies were checked to make certain of their ability to reverse the bio-parameters of rice when there were heavy metal stress. A method to classify the stress level based on both the physical mechanism analysis and the statistic model are proposed. The 3-axes spectral indices spaces, which are constructed of 3 spectral indices sensitive to rice's chlorophyll concentration, nitrogen concentration and water concentration respectively, are used to visualize the linkage between heavy metal stress and spectrum of rice canopy.

Based on the introduction to the characters and constructing flow of space semantic model, the feature space and context of house information in high resolution remote sensing image are analyzed, and the house semantic network model of Quick Bird image is also constructed. Furthermore, after region segmentation and the edge extraction to the image taking advantage of window threshold value method and Hough transformation, house information is extracted automatically from Quick Bird image through the whole space semantic model.

The deterioration of ecology environment has proved to induce various vector-borne diseases directly or indirectly. The diseases and epidemic situations erupted recently have a robust relationship with the worsening of environment conditions. Vector-borne diseases are plaguing much of the world, though some remarkable achievements have been gained in the battle against vector-borne diseases. The threats to the public health from vector-borne diseases are at least as important now as at any time in history, especially in developing countries. Most of the vector-borne diseases are associated with specific environmental factors. Control of vector-borne diseases requires the knowledge of the ecology of habitats and breeding sites of vectors, and understanding the life cycles of vectors and pathogens. The past few years have seen a rapid development in spatial information technologies represented by 3S (GIS, RS, GPS), which are suitable for identifying and monitoring environmental targets associated with vector-borne diseases. Combined with other mathematical analysis toolboxes, spatial information technologies provide a new powerful solution to analyzing and predicting the spatial-temporal patterns of vector- borne diseases, which will be helpful for risk assessment and disease prevention. This article gives a general review of the major environmental factors, including climate factor, land use/land cover, hydrologic factor and terrain factor, and analyses briefly the mechanism of each factor acting on vector-borne diseases, then summarizes the major achievements of related work on identifying and monitoring the environmental factors with remote sensing and geographic information system.