On-board Processing Research Articles

Correction and calibration of atmospheric impact observations in GOES GLM data.

The Earth's atmosphere is impacted daily by both meteoroids and artificial objects. Calibrated observations of the emitted light at sufficiently high sampling rates can enable or improve the estimation of impactor attributes such as size, cohesion, trajectory, and composition, but are difficult to obtain owing to the unpredictability, brevity, and high dynamic (brightness) range of impacts. Ground-based camera systems have successfully monitored small regions of the atmosphere at video frame rates and with limited radiometric capabilities, but most impacts occur over the 70% of the Earth's surface covered by water and are therefore missed by these networks. The Geostationary Lightning Mapper (GLM) instruments aboard Geostationary Operational Environmental Satellites 16 and 17 provide near-hemispherical coverage at 500 frames per second. These data have been shown to contain the signatures of many independently confirmed impacts, often from both viewing angles simultaneously, and constitute an observational resource that is currently unparalleled in the public domain. NASA's Asteroid Threat Assessment Project has implemented an automated impact detection pipeline that processes data from GLM daily. Given a detected impact, the GLM data contain a wealth of information for use in quantitative follow-up analyses. However, impact events differ from lightning in ways that violate key assumptions built into GLM's design. The result is that GLM's onboard processing introduces errors into pixel observations of impact events and the calibrated energies near the periphery of the detector may be substantially overestimated. We present methods for mitigating these and other issues to produce a data product more suitable for impact analyses than the existing GLM lightning product.

Measurement principle and technology of miniaturized strapdown inertial wave sensor

This paper presents the design and implementation of a low-cost directional inertial wave sensor, DWS19-2, which is suitable for installation on ocean air-sea coupling buoys, offshore environmental monitoring buoys and surface drifting buoys. DWS19-2 integrates an STM32F446 embedded controller with a 9-axis MEMS inertial module QZ901 to measure buoy pitch, roll, heading and accelerations. These parameters are further used to calculate wave parameters and estimate directional and non-directional wave spectra. These wave parameters and spectral information are then reported to the buoy data logger for real-time transmission to shore-based data centers. Details of the electrical design and onboard wave processing algorithm (coordinate projection, numerical integration in the frequency domain, time-based wave parameters, frequency-based wave parameters, directional wave spectrum), in addition to turntable tests, filed marine comparison with DWR-MKIII wave buoy and related ocean observation applications enabled by this device, are described in this paper.

Cloud Detection Autonomous System Based on Machine Learning and COTS Components On-Board Small Satellites

One of the main applications of small satellites is Earth observation. CubeSats and different kinds of nanosatellites usually form constellations that obtain images mainly using an optical payload. There is a massive amount of data generated by these satellites and a limited capacity of download due to volume and mass constraints that make it difficult to use high-speed communication systems and high-power systems. For this reason, it is important to develop satellites with the autonomy to process data on board. In this way, the limited communication channel can be used efficiently to download relevant images containing the required information. In this paper, a system for the satellite on-board processing of RGB images is proposed, which automatically detects the cloud coverage level to prioritize the images and effectively uses the download time and the mission operation center. The system implements a Convolutional Neural Network (CNN) on a Commercial off-the-Shelf (COTS) microcontroller that receives the image and returns the cloud level (priority). After training, the system was tested on a dataset of 100 images with an accuracy of 0.9 and it was also evaluated with CubeSat images to evaluate the performance of a different image sensor. This implementation contributes to the development of autonomous satellites with processing on board.

Age of Processing-Based Data Offloading for Autonomous Vehicles in MultiRATs Open RAN

Today, vehicles use smart sensors to collect data from the road environment. This data is often processed onboard of the vehicles, using expensive hardware. Such onboard processing increases the vehicle’s cost, quickly drains its battery, and exhausts its computing resources. Therefore, offloading tasks onto the cloud is required. Still, data offloading is challenging due to low latency requirements for safe and reliable vehicle driving decisions. Moreover, age of processing was not considered in prior research dealing with low-latency offloading for autonomous vehicles. This paper proposes an age of processing-based offloading approach for autonomous vehicles using unsupervised machine learning, Multi-Radio Access Technologies (multi-RATs), and Edge Computing in Open Radio Access Network (O-RAN). We design a collaboration space of edge clouds to process data in proximity to autonomous vehicles. To reduce the variation in offloading delay, we propose a new communication planning approach that enables the vehicle to optimally preselect the available RATs such as Wi-Fi, LTE, or 5G to offload tasks to edge clouds when its local resources are insufficient. We formulate an optimization problem for age-based offloading that minimizes elapsed time from generating tasks and receiving computation output. To handle this non-convex problem, we develop a surrogate problem. Then, we use the Lagrangian method to transform the surrogate problem to unconstrained optimization problem and apply the dual decomposition method. The simulation results show that our approach significantly minimizes the age of processing in data offloading with 90.34% improvement over similar method.

NanoMap: A GPU-Accelerated OpenVDB-Based Mapping and Simulation Package for Robotic Agents

Encoding sensor data into a map is a problem that must be undertaken by any robotic agent operating in unknown or uncertain environments, and real-time updates are crucial to safe planning and control. Most modern robotic sensors produce some form of depth data or point cloud information that is only useful to the agent after being processed into the appropriate data structure, oftentimes an occupancy map. However, as the quality of sensor technology improves, so does the magnitude of the input data, which can creates a problem when trying to construct occupancy maps in real-time. Populating such an occupancy map using these dense point clouds can quickly become an expensive process, and many robotic agents have limited onboard computational bandwidth and memory. This results in delayed map updates and reduced operational performance in dynamic environments where real-time information is crucial for safe operation. However, while many modern robotic agents are still relatively limited by the power of onboard central processing units (CPUs), many platforms are gaining access to onboard graphics processing units (GPUs), and these resources remain underutilised with respect to the problem of occupancy mapping. We propose a novel probabilistic mapping solution that leverages a combination of OpenVDB, NanoVDB, and Nvidia’s Compute Unified Device Architecture (CUDA) to encode dense point clouds into OpenVDB data structures, leveraging the parallel compute strength of GPUs to provide significant speed advantages and further free up resources for tasks that cannot as easily be performed in parallel. An evaluation of our solution is provided, with performance benchmarks provided for both a laptop and a low power single board computer with onboard GPU. Similar performance improvements should be accessible on any system with access to a CUDA-compatible GPU. Additionally, our library provides the means to simulate one or more sensors on an agent operating within a randomly generated 3D-grid environment and create a live map for the purposes of evaluating planning and control techniques and for training agents via deep reinforcement learning. We also provide interface packages for the Robotic Operating System (ROS1) and the Robotic Operating System 2 (ROS2), and a ROS2 visualisation (RVIZ2) plugin for the underlying OpenVDB data structure.

Deep Learning with Spatial Attention-Based CONV-LSTM for SOC Estimation of Lithium-Ion Batteries

Accurate estimation of the state of charge (SOC) is an indispensable part of a vehicle management system. The accurate estimation of SOC can ensure the safe and reliable operation of the vehicle management system. With the development of intelligent transportation systems (ITS), vehicles can not only obtain the dynamic changes inside the battery through sensors, but also obtain the traffic information around the vehicle through vehicle–road collaboration. In addition, the development of onboard graphic processing units (GPUs) and Internet of Vehicles (IOV) technology make the computing power of vehicles no longer limited by hardware, which makes neural networks applied to the intelligent control of vehicles. Aiming at the problem that the traditional network cannot effectively obtain the complex spatial information of sample attributes, we developed an attention-based CONV-LSTM module for SOC prediction based on a convolutional neural network (CNN) and a long short-term memory (LSTM) network. Different from the traditional LSTM network, the algorithm not only considers the temporal correlation of the data stream, but also captures the spatial correlation information of the input data by convolution. It then uses the different weights, automatically assigned by the attention mechanism, to correctly distinguish the importance of different input data streams. In order to verify the validity of the model, this paper selects the degradation data set of the aeroengine as the verification data set. Experiments show that the proposed model has achieved good results. Finally, the proposed model is applied to the actual vehicle running data, and the effectiveness of the proposed model is verified by comparing it with the Multi-Layer Perceptron (MLP), LSTM, and CNN-LSTM models.

Increasingly detailed insights in animal behaviours using continuous on-board processing of accelerometer data

Background:Studies of animal behaviour, ecology and physiology are continuously benefitting from progressing biologging techniques, including the collection of accelerometer data to infer animal behaviours and energy expenditure. In one of the most recent technological advances in this space, on-board processing of raw accelerometer data into animal behaviours proves highly energy-, weight- and cost-efficient allowing for continuous behavioural data collection in addition to regular positional data in a wide range of animal tracking studies.Methods:We implemented this latest development in collecting continuous behaviour records from 6 Pacific Black Ducks Anas superciliosa to evaluate some of this novel technique’s potential advantages over tracking studies lacking behavioural data or recording accelerometer data intermittently only. We (i) compared the discrepancy of time-activity budgets between continuous records and behaviours sampled with different intervals, (ii) compared total daily distance flown using hourly GPS fixes with and without additional behavioural data and (iii) explored how behaviour records can provide additional insights for animal home range studies.Results:Using a total of 690 days of behaviour records across six individual ducks distinguishing eight different behaviours, we illustrated the improvement that is obtained in time-activity budget accuracy if continuous rather than interval-sampled accelerometer data is used. Notably, for rare behaviours such as flying and running, error ratios > 1 were common when sampling intervals exceeded 10 min. Using 72 days of hourly GPS fixes in combination with continuous behaviour records over the same period in one individual duck, we showed behaviour-based daily distance estimation is significantly higher (up to 540%) than the distance calculated from hourly sampled GPS fixes. Also, with the same 72 days of data for one individual duck, we showed how this individual used specific sites within its entire home range to satisfy specific needs (e.g. roosting and foraging).Conclusion:We showed that by using trackers allowing for continuous recording of animal behaviour, substantial improvements in the estimation of time-activity budgets and daily traveling distances can be made. With integrating behaviour into home-range estimation we also highlight that this novel tracking technique may not only improve estimations but also open new avenues in animal behaviour research, importantly improving our knowledge of an animal’s state while it is roaming the landscape.

A thermal emissions-based real-time monitoring system for in situ detection of fatigue cracks

The advent of packaged infrared (IR) bolometers has led to thermography-based techniques becoming popular for non-destructive evaluation of aerospace structures. In this work, a real-time monitoring system for in situ crack detection has been presented which uses an original equipment manufacturer microbolometer. The system costs one-tenth the price of a packaged bolometer and has the potential to transform the use of IR imaging for condition and structural health monitoring in the aerospace industry and elsewhere. A computer, consisting of a single circuit board with dimensions comparable to a credit card, has been integrated into the system for real-time, on-board data processing. Crack detection has been performed based on the principles of thermoelastic stress analysis (TSA). Proof-of-concept laboratory tests were performed on open-hole aluminium specimens to compare the performance of the proposed system against a state-of-the-art cooled IR photovoltaic effect detector. It was demonstrated that cracks as small as 1 mm in length can be detected with loading frequencies as low as 0.3 Hz. This represents a significant advance in the viability of TSA-based crack detection in large-scale structural tests where loading frequencies are usually lower than 1 Hz.

Kalman filtering for SAR satellite yaw steering

The algorithm based on Kalman filtering was proposed to calculate yaw steering angles of spaceborne Synthetic Aperture Radar (SAR) systems. We establish the process equation and the observation equation by proper design. Performance analysis and computer simulations demonstrate that Kalman filtering may take advantages in both numerical precision and computational efficiency. It therefore can be applied to SAR satellites in a wide range of microwave bands (C, X, Ku) along noncircular orbits. In addition, an efficient algorithm based on simple interpolation may be employed in practice. Since the Kalman filter is linear in nature, the fast computation makes possible the implementation of accurate onboard processing.

Constrained Optimization of FPGA Design for Spaceborne InSAR Processing

With the development of spaceborne processing technologies, the demand for on-board processing has risen sharply. Against this background, spaceborne Interferometric Synthetic Aperture Radar (InSAR) processing has become an important research area. In many cases, high processing capacity is required during on-board InSAR processing, yet Field-Programmable Gate Array (FPGA) resources on the satellites are limited. To improve the performance of spaceborne remote sensing processing, this paper designs a high-performing FPGA system for the coarse registration and interferogram generation process of InSAR. Moreover, to address this dual-constraint problem of resource and processing capacity, the paper proposes an FPGA design method based on the gradient descent theory, which can identify the optimum trade-off scheme between two such constraints. Finally, the proposed system design and method are implemented in FPGA. Experiments showed that the FPGA system outperformed the NVIDIA (Santa Clara, CA, USA) GTX Titan Black Graphics Processing Unit (GPU), and the optimum trade-off scheme only increases the entire time by 1.1% but reduces the FPGA BRAM usage by 8.7%. The experimental results proved the effectiveness and validity of the proposed system and method.

A Low-Cost Multi-Purpose IoT Sensor for Biologging and Soundscape Activities.

The rapid expansion in miniaturization, usability, energy efficiency, and affordability of Internet of Things (IoT) sensors, integrated with innovations in smart capability, is greatly increasing opportunities in ground-level monitoring of ecosystems at a specific scale using sensor grids. Surrounding sound is a powerful data source for investigating urban and non-urban ecosystem health, and researchers commonly use robust but expensive passive sensors as monitoring equipment to capture it. This paper comprehensively describes the hardware behind our low-cost, small multipurpose prototype, capable of monitoring different environments (e.g., remote locations) with onboard processing power. The device consists of a printed circuit board, microprocessor, local memory, environmental sensor, microphones, optical sensors and LoRa (Long Range) communication systems. The device was successfully used in different use cases, from monitoring mosquitoes enhanced with optical sensors to ocean activities using a hydrophone.

Real-time acoustic monitoring with telemetry to mitigate potential effects of seismic survey sounds on marine mammals: a case study offshore Sakhalin Island

Exxon Neftegas Ltd. (ENL) carried out three 4D seismic surveys during the summer of 2015. Seismic operations in two of these fields (Odoptu and Chayvo) ensonified the nearshore feeding area of Korean-Okhotsk (western) gray whales (Eschrichtius robustus), potentially disturbing feeding activities. Following model-based optimization of the source design to minimize its lateral acoustic footprint, pre-season modeling was used to compute the acoustic exposure along each survey line. Real-time acoustic data facilitated implementation of mitigation measures aimed to minimize disturbance of whales. Acoustic data originated from underwater recorders deployed on the seafloor. Two complementary approaches were used to transmit recorded sound data to a computer housed at the Central Post (CP), where decisions regarding mitigation shut downs were made. In the first approach, a limited bandwidth (2–2000 Hz) sampling of the data was transmitted via cable to a surface buoy, which relayed these data to a shore station up to 15 km away via digital VHF telemetry. At the shore station, acoustic impulses from the seismic surveys were processed to compute impulse characteristics in the form of estimates of sound exposure level and peak sound pressure level, as well as one-minute-average 1/3-octave power spectral density coefficients, which were then transmitted to the CP via the internet. In the second, the pulse characteristics were computed through algorithms running on an onboard processor in each recorder’s surface buoy and sent directly to the CP computer via an Iridium satellite uplink. Both methods of data transfer proved viable, but Iridium transmission achieved the goal without the need for any shore based relay stations and is therefore more operationally efficient than VHF transmission. At the CP, analysts used the real-time acoustic data to calibrate and adjust the output of pre-season acoustical model runs. The acoustic footprint for the active seismic source, advancing synchronously with the motion of the seismic vessel and changing as the sound propagation environment changed, was computed from the calibrated and adjusted model output and integrated through the software Pythagoras with locations of gray whales provided by shore-based observers. This enabled analysts to require air gun array shutdowns before whales were exposed to mean square sound pressure levels greater than the behavioral response threshold of 163 dB re 1 μPa2. The method described here provides a realistic means of mitigating the possible effects of air guns at a behavioral response level, whereas most seismic surveys rely on pre-established mitigation radii to manage the risk of injury to a whale.

Graph-Based Heuristic Solution for Placing Distributed Video Processing Applications on Moving Vehicle Clusters

Vehicular fog computing (VFC) is envisioned as an extension of cloud and mobile edge computing to utilize the rich sensing and processing resources available in vehicles. We focus on slow-moving cars that spend a significant time in urban traffic congestion as a potential pool of onboard sensors, video cameras, and processing capacity. For leveraging the dynamic network and processing resources, we utilize a stochastic mobility model to select nodes with similar mobility patterns. We then design two distributed applications that are scaled in real-time and placed as multiple instances on selected vehicular fog nodes. We handle the unstable vehicular environment by a), Using real vehicle density data to build a realistic mobility model that helps in selecting nodes for service deployment b), Using community-detection algorithms for selecting a robust vehicular cluster using the predicted mobility behavior of vehicles. The stability of the chosen cluster is validated using a graph centrality measure, and c), Graph-based placement heuristics is developed to find the optimal placement of service graphs based on a multi-objective constrained optimization problem with the objective of efficient resource utilization. The heuristic solves an important problem of processing data generated from distributed devices by balancing the trade-off between increasing the number of service instances to have enough redundancy of processing instances to increase resilience in the service in case of node or link failure, versus reducing their number to minimize resource usage. We compare our heuristic to a mixed integer program (MIP) solution and a first-fit heuristic. Our approach performs better than these comparable schemes in terms of resource utilization and/or has a lesser service latency when compared to an edge computing-based service placement scheme.

Evaluating Link Lifetime Prediction to Support Computational Offloading Decision in VANETs.

In urban mobility, Vehicular Ad Hoc Networks (VANETs) provide a variety of intelligent applications. By enhancing automobile traffic management, these technologies enable advancements in safety and help decrease the frequency of accidents. The transportation system can now follow the development and growth of cities without sacrificing the quality and organisation of its services thanks to safety apps that include collision alerts, real-time traffic information, and safe driving applications, among others. Applications can occasionally demand a lot of computing power, making their processing impractical for cars with limited onboard processing capacity. Offloading of computation is encouraged by such a restriction. However, because vehicle mobility operations are dynamic, communication times (also known as link lifetimes) between nodes are frequently short. VANET applications and processes are impacted by such communication delays (e.g., the offloading decision when using the Computational Offloading technique). Making an accurate prediction of the link lifespan between vehicles is therefore challenging. The effectiveness of the communication time estimation is currently constrained by the link lifespan prediction methods used in the computational offloading process. This work investigates five machine learning (ML) algorithms to predict the link lifetime between nodes in VANETs in different scenarios. We propose the procedures required to carry out the link lifetime prediction method using existing ML techniques. The tactic creates datasets with the features the models need to learn and be trained. The SVR and XGBoost algorithms that were selected as part of the assessment process were trained. To make the prediction using the trained models, we modified the lifespan prediction function from an offloading approach. To determine the viability of applying link lifespan predictions from the models trained in the road and urban scenarios, we conducted a performance study. The findings indicate that compared to the conventional prediction strategy described in the literature, the suggested link lifetime prediction via regression approaches decreases prediction error rates. An offloading method from the literature is extended by the selected SVR. The task loss and recovery rates might be significantly reduced using the SVR. XGBoost outperformed its ML competitors in task recovery or drop rate by 70% to 80% in an assessed hypothesis compared to an offloading choice technique in the literature. With greater offloading rates from an application on the VANET, this effort is intended to give better efficiency in estimating this data using machine learning in various vehicular settings.

An Edge-Based Architecture for Offloading Model Predictive Control for UAVs

Thanks to the development of 5G networks, edge computing has gained popularity in several areas of technology in which the needs for high computational power and low time delays are essential. These requirements are indispensable in the field of robotics, especially when we are thinking in terms of real-time autonomous missions in mobile robots. Edge computing will provide the necessary resources in terms of computation and storage, while 5G technologies will provide minimal latency. High computational capacity is crucial in autonomous missions, especially for cases in which we are using computationally demanding high-level algorithms. In the case of Unmanned Aerial Vehicles (UAVs), the onboard processors usually have limited computational capabilities; therefore, it is necessary to offload some of these tasks to the cloud or edge, depending on the time criticality of the application. Especially in the case of UAVs, the requirement to have large payloads to cover the computational needs conflicts with other payload requirements, reducing the overall flying time and hindering autonomous operations from a regulatory perspective. In this article, we propose an edge-based architecture for autonomous UAV missions in which we offload the high-level control task of the UAV’s trajectory to the edge in order to take advantage of the available resources and push the Model Predictive Controller (MPC) to its limits. Additionally, we use Kubernetes to orchestrate our application, which runs on the edge and presents multiple experimental results that prove the efficacy of the proposed novel scheme.

A Design Method for Pre-Distortion Compensation of SAR Chirp Signal based on Envelop Sampling and Interpolation Filter

The synthetic aperture radar(SAR) is an equipment that can acquire images in all weathers day and night based on radar signals. The on-board processor of satellite SAR generates transmission signal by digital signal processing, converts it into an analog signal and transmits to antenna. Until the transmission signal generated by on-board processor is output, the signal passes the transmission cables and analog devices. At this time, these hardware distort the signal and makes SAR performance worse. To improve the performance, pre-distortion technique is used. But, general pre-distortion using taylor series is not sufficient to compensate for the distortion. This paper suggests transmit signal design method with improved pre-distortion. This paper uses envelop sampling method and interpolation filter for frequency domain compensation. The proposed method accurately compensates the hardware distortion and reduces resource usage of FPGA. To analyze proposed method’s performance, IRF characteristics are compared when the proposed method applies to signal with errors.

Point Cloud Compression: Impact on Object Detection in Outdoor Contexts.

Increasing demand for more reliable and safe autonomous driving means that data involved in the various aspects of perception, such as object detection, will become more granular as the number and resolution of sensors progress. Using these data for on-the-fly object detection causes problems related to the computational complexity of onboard processing in autonomous vehicles, leading to a desire to offload computation to roadside infrastructure using vehicle-to-infrastructure communication links. The need to transmit sensor data also arises in the context of vehicle fleets exchanging sensor data, over vehicle-to-vehicle communication links. Some types of sensor data modalities, such as Light Detection and Ranging (LiDAR) point clouds, are so voluminous that their transmission is impractical without data compression. With most emerging autonomous driving implementations being anchored on point cloud data, we propose to evaluate the impact of point cloud compression on object detection. To that end, two different object detection architectures are evaluated using point clouds from the KITTI object dataset: raw point clouds and point clouds compressed with a state-of-the-art encoder and three different compression levels. The analysis is extended to the impact of compression on depth maps generated from images projected from the point clouds, with two conversion methods tested. Results show that low-to-medium levels of compression do not have a major impact on object detection performance, especially for larger objects. Results also show that the impact of point cloud compression is lower when detecting objects using depth maps, placing this particular method of point cloud data representation on a competitive footing compared to raw point cloud data.

Urban Intersection Management Strategies for Autonomous/Connected/Conventional Vehicle Fleet Mixtures

Connected Vehicles and Autonomous Vehicles (CAVs) provide various sources of vehicular related information to intersection infrastructure by integrating on-board sensors processing, wireless communication and other Vehicle-to-Infrastructure (V2I) technologies. Thus connected vehicle technologies can potentially remedy data collection limitations of existing urban intersection managements, enhancing the performances of intersection controls such as reducing vehicle delay, reducing vehicle number of stops and improving energy efficiency. This paper reviews optimization-based signal controls for different penetrations of connected vehicles and conventional vehicles environments, autonomous intersection management specific to completely 100% AVs road states, as well as signal-trajectory joint control for different adoptions of conventional vehicles, CVs and AVs mixture environments. Real time data processing, signal timing optimizations, vehicle trajectory motion planning and evaluation frameworks are summarized to highlight the advantages and limitations of respective intersection control paradigms. It is important to recognize that realistic scenarios in comparative assessments for proposed methods need to be achieved in future works. The effectiveness of different approaches is challenging to be compared without complete evaluation frameworks, and sensitivity analysis and hypothesis tests involving variety penetration rates and flow demands should be performed in order to test the stability of methods in different scenarios.

Expandable On-Board Real-Time Edge Computing Architecture for Luojia3 Intelligent Remote Sensing Satellite

Since the data generation rate of high-resolution satellites is increasing rapidly, to relieve the stress of data downloading and processing systems while enhancing the time efficiency of information acquisition, it is important to deploy on-board edge computing on satellites. However, the volume, weight, and computability of on-board systems are strictly limited by the harsh space environment. Therefore, it is very difficult to match the computability and the requirements of diversified intelligent applications. Currently, this problem has become the first challenge of the practical deployment of on-board edge computing. To match the actual requirements of the Luojia3 satellite of Wuhan University, this manuscript proposes a three-level edge computing architecture based on a System-on-Chip (SoC) for low power consumption and expandable on-board processing. First, a transfer level is designed to focus on hardware communications and Input/Output (I/O) works while maintaining a buffer to store image data for upper levels temporarily. Second, a processing framework that contains a series of libraries and Application Programming Interfaces (APIs) is designed for the algorithms to easily build parallel processing applications. Finally, an expandable level contains multiple intelligent remote sensing applications that perform data processing efficiently using base functions, such as instant geographic locating and data picking, stream computing balance model, and heterogeneous parallel processing strategy that are provided by the architecture. It is validated by the performance improvement experiment that following this architecture, using these base functions can help the Region of Interest (ROI) system geometric correction fusion algorithm to be 257.6 times faster than the traditional method that processes scene by scene. In the stream computing balance experiment, relying on this architecture, the time-consuming algorithm ROI stabilization production can maintain stream computing balance under the condition of insufficient computability. We predict that based on this architecture, with the continuous development of device computability, the future requirements of on-board computing could be better matched.

Unsupervised Dynamic Sensor Selection for IoT-Based Predictive Maintenance of a Fleet of Public Transport Buses

In recent years, big data produced by the Internet of Things has enabled new kinds of useful applications. One such application is monitoring a fleet of vehicles in real time to predict their remaining useful life. The consensus self-organized models (COSMO) approach is an example of a predictive maintenance system. The present work proposes a novel Internet of Things based architecture for predictive maintenance that consists of three primary nodes: the vehicle node, the server leader node, and the root node, which enable on-board vehicle data processing, heavy-duty data processing, and fleet administration, respectively. A minimally viable prototype of the proposed architecture was implemented and deployed to a local bus garage in Gatineau, Canada.The present work proposes improved consensus self-organized models (ICOSMO), a fleet-wide unsupervised dynamic sensor selection algorithm. To analyze the performance of ICOSMO, a fleet simulation was implemented. The J1939 data gathered from a hybrid bus was used to generate synthetic data in the simulations. Simulation results that compared the performance of the COSMO and ICOSMO approaches revealed that in general ICOSMO improves the average area under the curve of COSMO by approximately 1.5% when using the Cosine distance and 0.6% when using Hellinger distance.