DeepOcean: A General Deep Learning Framework for Spatio-Temporal Ocean Sensing Data Prediction

The advancement of Underwater acoustic communication (UAC) holds great promise for the Internet of Underwater Things (IoUT). However, a critical challenge to the development of UAC technology is the difficulty of obtaining sea trial data, such difficulty makes the emerging algorithms inconvenient to evaluate. In this paper, we propose a Shared Underwater Acoustic Communication Layer (SUACL) to collect underwater acoustic signal, the proposed SUACL is founded on the IoUT and users are allowed to remotely access the system to arrange underwater acoustic data transmission and reception via Internet. The SUACL system aims to enable the remote operation of communication units that are deployed in designated sea areas. The merit of the proposed SUACL is that the sea trial data can be obtained without risky and high expense, and the emerging underwater acoustic (UWA) technologies can be evaluated in time. The SUACL is composed of the non-underwater part which includes the application layer and fusion layers, and the above/under water part including the communication layer on which we focus. We provide a detailed description of the communication layer of the proposed SUACL system, such as the pre-amplifier with reconfigurable gain, the analog-to-digital converter with reconfigurable sampling rates, the reconfigurable transducers with different frequency bands, the reconfigurable number of hydrophones. The SUACL's reconfigurability options make it a versatile tool that can meet the diverse experimental needs for underwater communication research. In addition, we also describe the mechanisms of how experimental data is exchanged through the SUACL. Our experimental results demonstrate the feasibility and effectiveness of the proposed SUACL in a shallow-water environment.

Underwater acoustic (UWA) communication is a critical part of information exchange among marine equipment in the Internet of Underwater Things (IoUT). The impact of the marine environment on UWA communication manifests through physical channels, which exhibit multipath propagation, Doppler spread, and time-varying characteristics. The dynamic compressed sensing (DCS) framework, represented by Kalman Filtered CS (KF-CS) algorithm, is introduced to dynamically adapt support sets by real-time tracking of observations. Meanwhile, channel modeling in Delay-Doppler (DD) domain offers a more stable representation for doubly-selective UWA channels, revealing a potential for channel estimation. In this paper, we transform the time-domain channel estimation into the DD domain channel estimation, and formulate the DD domain channel estimation under the DCS framework to exploit the time-varying sparsity of the UWA channel in the DD domain. In addition, to solve the DD model, the classic algorithms need to search for the sparse solution in the extremely large two-dimensional space of the entire UWA channel. Since only a few multipath arrivals within the large time-delay spread of the entire UWA channel have the Doppler value, we propose the multipath arrivals-Doppler function-based DCS channel estimation approach (DCS-MAD) to solve the DD model. The proposed DCS-MAD can remodel the traditional DD model as multipath arrivals and Doppler model, and search for the sparse solution in the two-dimensional space of multipath arrivals, instead of on the entire channel, so as to effectively reduce system complexity and better focus on Doppler value estimation at the UWA channel’s multipath arrivals. Specifically, we first estimate the multipath arrival delay of the UWA channel using the DCS algorithm. Then, the two-dimensional space search is performed specifically at the multipath arrivals and Doppler domains under the DCS framework to exploit the time-varying sparsity. Finally, the results of numerical simulation and sea experiments validate the superiority of the proposed schemes, compared to the benchmark algorithms.

With the advance of undersea communication technologies, the operation of large numbers of small submerged devices and their connection, such as the Internet of Underwater Things (IoUT), is becoming possible. Deployed over long periods of time, energy restrictions limit the application of underwater acoustic (UWA) communication, and thus, the peer-to-peer (P2P) communication packets should be small with minimum overhead dedicated to training symbols. On the other hand, the UWA communication in physical layer requires good equalization performance to manage the inevitably long intersymbol interference. In this article, we propose a semiblind joint channel estimation and decoding (JCED) approach for P2P UWA communication to allow efficient channel equalization despite the short training sequence. We formulate the JCED problem and solve it by passing messages between the raw acoustic observations, the nodes that characterize the connections of the channel's taps, and the information-bearing symbols within a turbo equalization framework. To initialize the prior of the channel's components, we employ the expectation maximization algorithm, assuming a Gaussian mixture model for the channel's taps. Extensive numerical simulation and results from a field experiment show that our approach outperforms state-of-the-art benchmarks in terms of the communication system's capability of performing well with a small number of pilot symbols.

In response to the increasing demand for marine data transmission, underwater acoustic (UWA) communication is widely recognized as a key technology to support various applications of the Internet of Underwater Things (IoUT). However, due to the narrow bandwidth shared by multiple acoustic systems and the constrained energy of the underwater nodes, it is an urgent problem to design an energy-efficient resource allocation scheme which can avoid interference simultaneously. In this article, we establish a cognitive orthogonal frequency-division multiple access (OFDMA) UWA communication system, where dynamic resource allocation relies on the feedback channel state information (CSI) and spectrum usage of interference. In order to detect interference, a spectrum sensing scheme applicable to the underwater environment is proposed and experimentally verified. Subsequently, the resource allocation problem is mathematically modeled with energy efficiency (EE) as the reasonable optimization objective in UWA communication systems. In order to reduce the computational complexity, the original problem is decomposed into two subproblems and a low-complexity algorithm named acrlong BOSC- acrlong TLI is developed for feasible solutions. Simulation and sea experiment results show that the proposed algorithm outperforms the existing underwater resource allocation schemes.

The Underwater Internet of Things (UW-IoT) is a novel class of Internet of Things (IoT) that enables extend the terrestrial internet to the marine environment. However, as the underwater acoustic channel poses extremely challenging difficulties for underwater IoT communication and networking, the performance of the underwater IoT networks lags far behind the terrestrial internet. Due to the expense of conducting experiments in real underwater systems many researches have been endowed with only simulation model and analysis. This paper presents the design and implementation of an underwater IoT network test platform in order to evaluate and analyze the performance of underwater communication and network protocol in physical shallow water environment. The design of this IoT network platform considers the application background of environment parameter acquisition in marine environment stereoscopic monitoring. The software and hardware implementation of the IoT platform are introduced. Sea test result shows that the platform is effective in evaluating the performance of underwater IoT network.

Multicarrier code division multiple access (MC-CDMA) can be applied in underwater acoustic communication networks to combat the narrow bandwidth of the underwater acoustic channel and multipath fading. Power control is one of the key technologies for the MC-CDMA communication networks. In this paper, we propose a multi-objective power control scheme based on improved particle swarm optimization (PSO) algorithm for underwater acoustic MC-CDMA communication networks. The distance of different nodes to base station (BS) is not same, which can cause near-far effect. So we propose a fitness function which is consisted of power consumption and SINR (the signal to interference plus noise ratio) on the receiver. The method effective reduces the power consumption of nodes and decreases the near-far effect. Comparing with the scheme without power control, our results show that the proposed power control algorithm can be used in underwater acoustic channel. Not only the power consumption is lower, but also the bit error rate (BER) performances are better.

T400 conventional underwater acoustic (UWA) wakeup signals detection methods for both underwater cooperative nodes and non-cooperative nodes are FFT based signal detection and matched-filtered based signal detection. While there is a big performance degradation of these conventional frequency-offset sensitive methods due to the relative motion between nodes, complex underwater noise environment, channel time-varying and other interference sources. This paper proposes a dynamic time warping (DTW) algorithm to detect the wake-up signal in UWA communication networks, which is frequency-offset insensitive and performs better than the conventional methods in severe frequency-offset environment by simulations based on the true UWA channel measured from sea trial.

Plant disease can seriously affect the growth and development of plants, resulting in reduced production and even death. Disease control can assist in plant breeding and cultivation, ensuring crop yield and quality. But traditional disease control methods are increasingly difficult to meet modern time, space and cost requirements. Deep learning technology has developed rapidly in recent years; it can improve the efficiency of plant disease diagnosis and detection, and greatly reduce the risk of human intervention and misjudgment. However, challenges such as data acquisition, environmental interference, and model deployment complexity hinder the application of deep learning in plant disease control. Therefore, there is an urgent need for disease feature optimization algorithms to protect current large-scale agricultural production.This research topic provides a platform for sharing the latest research results of deep learning in control of plant disease. The focus on the progress of deep learning technology in this field mainly involves the application of advanced technology in plant disease feature adaptation. This research topic aims to explore the latest research and application of deep learning technology in plant disease identification and control, focusing on: 1) identification techniques for early plant disease, 2) prediction and prevention of the spread of plant disease, 3) new methods for collecting multi-source plant disease image data, 4) quantitative evaluation of plant damage, 5) real-time application deployment based on embedded devices and edge wearable devices, 6) adaptive image processing methods for different application environments.A study by Le Yang et al. explored a new maize weed recognition model, SE-VGG16, and applied the SE-VGG16 model to weed detection of other species to build a weed detection model with the ability to generalize weed control in agricultural production processes. SE-VGG16 model is based on VGG16, adding SE attention mechanism. The model implements automatic focusing and allocates limited information processing resources to important parts. It has the characteristics of high efficiency, stability and high detection accuracy.Pushkar Gole et al. focused on lightweight visual converter networks in a study of automatic plant disease recognition. This study mainly introduces two tasks: one is to replace a multilayer perceptron (MLP) with the initial module and reconstruct the transformer Inception module to reduce the computational complexity of the vision transformer (ViT) model. In addition, the proposed new network is trained and tested on the PlantVillage dataset and the field maize image dataset, which proves its applicability in real scenarios.Further, Xin Xu et al. studied a cotton moth infestation monitoring method. In this study, five different models of cotton aphid infestation identification were constructed, trained and tested, among which YOLOv5 model had the best performance. The real-time identification system established in this study can provide a faster and more convenient technical means for the monitoring of cotton aphid infection.In another article, Nak Jung Choi et al. focused on identifying and tracking the behavior, movement, size, and habits of different species of insects by building a 3D monitoring system. While traditional 2D monitoring systems are static and ignore Zaxis depth dimensions, 3D systems are able to track insect movement more comprehensively, providing valuable depth dimension insights.Another study by Jitong Cai et al., explored a lightweight CNN model, FCA-EfficientNet, to achieve accurate identification of multiple maize disease. The work introduced autofocus modules and FCA modules to filter out background noise, allowing the network to focus on disease areas. It also makes the model more lightweight by removing redundant network structures.In addition, Fei Nan et al. studied the classification of maize disease from another angle. They focused on a deep convolutional neural network with RGB-D images. Among them, a variety of improved data preprocessing methods improved the accuracy and speed of maize diseases image recognition.A study by Zhenghang Zhang et al. explored the potential of applying Generative Adversarial Networks (GANs) to the diagnosis of cotton fusarium wilt. This study designed a model based on small sample learning and proposed an innovative cotton verticillium wilt diagnosis system. The convolutional neural network is used as feature extractor, and the trained GANs model is used for sample enhancement to improve the classification accuracy.In summary, these findings highlight the significant advances that deep learning technology has made in identifying and controlling plant disease in the information age. In the future, these models can be deployed in smart agriculture systems to monitor and identify plant disease in real time, so that appropriate control measures can be taken in a timely manner. As deep learning technology continues to evolve and improve, we can expect these models to play an increasingly important role in plant disease control, providing more effective support for agricultural production.

Filtered multitone (FMT) modulation divides the communication band into several subbands to shorten the span of symbols affected by multipath in underwater acoustic (UWA) communications. However, there is still intersymbol interference (ISI) in each subband of FMT modulation degrading communication performance. Therefore, ISI suppression techniques must be applied to FMT modulation UWA communications. The suppression performance of traditional adaptive equalization methods often exploited in FMT modulation UWA communications is limited when the effect of ISI spans tens of symbols or large constellation sizes are used. Turbo equalization consisting of adaptive equalization and channel decoding can improve equalization performance through information exchanging and iterative processes. To overcome the shortcoming of traditional minimum mean square error (MMSE) equalization and effectively suppress the ISI with relatively low computation complexity, an FMT modulation UWA communication using low-complexity channel-estimation-based (CE-based) MMSE turbo equalization is proposed in this paper. In the proposed method, turbo equalization is first exploited to suppress the ISI in FMT modulation UWA communications, and the equalizer coefficients of turbo equalization are adjusted using the low-complexity CE-based MMSE algorithm. The proposed method is analyzed in theory and verified by simulation analysis and real data collected in the experiment carried out in a pool with multipath propagation. The results demonstrate that the proposed method can achieve better communication performance with a higher bit rate than the FMT modulation UWA communication using traditional MMSE adaptive equalization.

BackgroundRice is one of the major food crops in the world, and the monitoring of its growth condition is of great significance for guaranteeing food security and promoting sustainable agricultural development. Leaf area index (LAI) is a key indicator for assessing the growth condition and yield potential of rice, and the traditional methods for obtaining LAI have problems such as low efficiency and large error. With the development of remote sensing technology, unmanned aerial multispectral remote sensing combined with deep learning technology provides a new way for efficient and accurate estimation of LAI in rice.ResultsIn this study, a multispectral camera mounted on a UAV was utilized to acquire rice canopy image data, and rice LAI was uniformly estimated over multiple periods by the multilayer perceptron (MLP) and convolutional neural network (CNN) models in deep learning. The results showed that the CNN model based on five-band reflectance images (490, 550, 670, 720, and 850 nm) as input after feature screening exhibited high estimation accuracy at different growth stages. Compared with the traditional MLP model with multiple vegetation indices as inputs, the CNN model could better process the original multispectral image data, effectively avoiding the problem of vegetation index saturation, and improving the accuracies by 4.89, 5.76, 10.96, 1.84 and 6.01% in the rice tillering, jointing, booting, and heading periods, respectively, and the overall accuracy was improved by 6.01%. Moreover, the model accuracies (MLP and CNN) before and after variable screening showed noticeable changes. Conducting variable screening contributed to a substantial improvement in the accuracy of rice LAI estimation.ConclusionsUAV multispectral remote sensing combined with CNN technology provides an efficient and accurate method for the unified multi-period estimation of rice LAI. Moreover, the generalization ability and adaptability of the model were further improved by rational variable screening and data enhancement techniques. This study can provide a technical support for precision agriculture and a more accurate solution for rice growth monitoring. More feature extraction and variable screening methods can be further explored in future studies by optimizing the model structure to improve the accuracy and stability of the model.

Soil salinization affects agricultural productivity and ecosystem health in Xinjiang, especially in arid areas. The region’s complex topography and limited agricultural data emphasize the pressing need for effective, large-scale monitoring technologies. Therefore, 1044 soil samples were collected from arid farmland in northern Xinjiang, and the potential effectiveness of soil salinity monitoring was explored by combining environmental variables with Landsat 8 and Sentinel-2. The study applied four types of feature selection algorithms: Random Forest (RF), Competitive Adaptive Reweighted Sampling (CARS), Uninformative Variable Elimination (UVE), and Successive Projections Algorithm (SPA). These variables are then integrated into various machine learning models—such as Ensemble Tree (ETree), Random Forest (RF), Extreme Gradient Boosting (XGBoost), and LightBoost—as well as deep learning models, including Convolutional Neural Networks (CNN), Residual Networks (ResNet), Multilayer Perceptrons (MLP), and Kolmogorov–Arnold Networks (KAN), for modeling. The results suggest that fertilizer use plays a critical role in soil salinization processes. Notably, the interpretable model KAN achieved an accuracy of 0.75 in correctly classifying the degree of soil salinity. This study highlights the potential of integrating multi-source remote sensing data with deep learning technologies, offering a pathway to large-scale soil salinity monitoring, and thereby providing valuable support for soil management.

In order to establish a new personalized breast cancer screening paradigm, it is critically important to accurately predict the short-term risk of a woman having image-detectable cancer after a negative mammographic screening. In this study, we developed and tested a novel short-term risk assessment model based on deep learning method. During the experiment, a number of 270 "prior" negative screening cases was assembled. In the next sequential ("current") screening mammography, 135 cases were positive and 135 cases remained negative. These cases were randomly divided into a training set with 200 cases and a testing set with 70 cases. A deep learning based computer-aided diagnosis (CAD) scheme was then developed for the risk assessment, which consists of two modules: adaptive feature identification module and risk prediction module. The adaptive feature identification module is composed of three pairs of convolution-max-pooling layers, which contains 20, 10, and 5 feature maps respectively. The risk prediction module is implemented by a multiple layer perception (MLP) classifier, which produces a risk score to predict the likelihood of the woman developing short-term mammography-detectable cancer. The result shows that the new CAD-based risk model yielded a positive predictive value of 69.2% and a negative predictive value of 74.2%, with a total prediction accuracy of 71.4%. This study demonstrated that applying a new deep learning technology may have significant potential to develop a new short-term risk predicting scheme with improved performance in detecting early abnormal symptom from the negative mammograms.

The complexity of the underwater environment including long time-delay, large background noise, and excessive transmission loss, leads to two main challenges for the current mid-distance and long-distance underwater acoustic (UWA) communications: limited transmission rate and system reliability. To improve the performance of mid-range and long-range underwater acoustic communication, this paper proposes an M-ary multi-sequence superimposed spread spectrum(M-MSSS) UWA communication scheme based on a virtual time-reversal mirror (VTRM). Multiple sequences with periodic correlation characteristics of the original spread spectrum waveform are superimposed, and M-ary multi-sequence modulation is performed in each data block at the transmitter. A VTRM matched filter is designed for the receiver. The VTRM technology uses the physical characteristics of the typical coherent multipath of the underwater acoustic wave channel to reduce the signal amplitude through its strong spatial focusing ability. The inter-symbol interference is suppressed by reconstructing multipath signals and improved receiver. Simulations show that the scheme not only significantly improves the communication rate, but also ensures the system's better robustness in terms of bit error rate.

Many acoustic channels suffer from interference which is neither narrowband nor impulsive. This relatively long duration partial band interference can be particularly detrimental to system performance. We survey recent work in interference mitigation and orthogonal frequency division multiplexing (OFDM) as background motivation to develop a spatial diversity receiver for use in underwater networks. The network consists of multiple distributed cabled hydrophones that receive data transmitted over a time-varying multipath channel in the presence of partial band interference produced by interfering active sonar signals as well as marine mammal vocalizations. In operational networks, many dropped messages are lost due to partial band interference which corrupts different portions of the received signal depending on the relative position of the interferers, information source and receivers due to the slow speed of propagation.

Due to the frequency-selective nature of the underwater acoustic channel, OFDM has been proposed for use in underwater acoustic communication networks. However, because of the relatively low speed of sound in water in comparison to the speed of electromagnetic waves in air, the underwater acoustic channel is prone to a significant amount of Doppler distortion, which OFDM is particularly sensitive to. MISO architectures have several potential applications in wireless networks, either in increasing signal diversity and thus increasing the effective SNR, or in increasing spectral efficiency of transmissions. However, the degradation effect due to Doppler shifting is exacerbated when a receiver detects signals from multiple transmitters which may see different Doppler distortions. In this thesis, we look at existing proposed OFDM receiver designs for use in underwater acoustic networks, and compare their performance in a MISO network composed of two transmitters and one receiver. These performance comparisons are analyzed first through the use of MATLAB simulations and then through physical experiments in a controlled laboratory environment. Two receiver algorithms in particular are explored in detail. The first is a block-by-block approach in which pilot tones are transmitted in every OFDM block and used to estimate the channel. The second is an adaptive receiver which uses previous channel estimates and a data-directed approach to further estimate the channel response. These two approaches achieve near equal performance, but the block-by-block receiver has a much lower spectral efficiency due to the overhead needed to carry the pilot tones. We thus conclude that in environments which can handle potential reductions in performance, the adaptive receiver is clearly the optimal choice. However, in fast-changing or low SNR environments, the block-by-block approach may be preferred.