An Adaptive Outlier Detection and Processing Approach Towards Time Series Sensor Data

IEEE Access, Journal Year: 2019, Volume and Issue: 7, P. 175192 - 175212

Published: Jan. 1, 2019

The intelligent environment monitoring network, as the foundation of ecosystem research, has rapidly developed with ever-growing Internet Things (IoT). IoT-networked sensors deployed to monitor ecosystems generate copious sensor data characterized by nonstationarity and nonlinearity such that outlier detection remains a source concern. Most models involve hypothesis tests based on setting threshold values. However, signal decomposition describes stationary nonstationary relationships data. Therefore, this paper proposes three-level hybrid model median filter (MF), empirical mode (EMD), classification regression tree (CART), autoregression (AR) exponential weighted moving average (EWMA) methods called MF-EMD-CART-AR-EWMA detect outliers in first-level performance is compared Butterworth filter, FIR wavelet Wiener filter. second-level prediction support vector (SVR), K-nearest neighbor (KNN), CART, complementary ensemble EEMD CART AR (EEMD-CART-AR) CEEMD (CEEMD-CART-AR) methods. Finally, EWMA Cumulative Sum Control Chart (CUSUM) Shewhart control charts. proposed was evaluated real dataset from hydrometeorological observation network Heihe River Basin, yielding experimental results better generalization ability higher accuracy than models, providing extremely effective minor predicted This provides valuable insight promising reference for involving presents new perspective detecting outliers.

Language: Английский

---

Data-Driven Anomaly Detection Approach for Time-Series Streaming Data

Sensors, Journal Year: 2020, Volume and Issue: 20(19), P. 5646 - 5646

Published: Oct. 2, 2020

Recently, wireless sensor networks (WSNs) have been extensively deployed to monitor environments. Sensor nodes are susceptible fault generation due hardware and software failures in harsh Anomaly detection for the time-series streaming data of is a challenging but critical diagnosis task, particularly large-scale WSNs. The data-driven approach becoming essential goal improving reliability stability We propose anomaly this paper, named median filter (MF)-stacked long short-term memory-exponentially weighted moving average (LSTM-EWMA), status data, including operating voltage panel temperature recorded by node field. These can be used diagnose device anomalies. First, (MF) introduced as preprocessor preprocess obvious anomalies input data. Then, stacked memory (LSTM) employed prediction. Finally, exponentially (EWMA) control chart detector recognizing evaluate proposed devices field conditions environmental monitoring. Extensive experiments were conducted on real results demonstrate that compared other approaches, MF-stacked LSTM-EWMA significantly improve rate (DR) false (FR). DR FR values with 95.46% 4.42%, respectively. also achieves better F2 score than achieved methods. provides valuable insights WSNs detecting nodes.

Language: Английский

---

Real-time data integration of an internet-of-things-based smart warehouse: a case study

International Journal of Pervasive Computing and Communications, Journal Year: 2021, Volume and Issue: 18(5), P. 622 - 644

Published: Feb. 23, 2021

Purpose Creating a real-time data integration when developing an internet-of-things (IoT)-based warehouse is still faced with challenges. It involves diverse knowledge of novel technology and skills. This study aims to identify the critical components processes in IoT-based warehousing. Then, design apply framework, adopting IoT concept enable transfer sharing. Design/methodology/approach The used pilot experiment verify system configuration. Radio-frequency identification (RFID) was selected support process this study, as it one most recognized products IoT. Findings experimentations’ results proved that plays significant role structuring combination assorted on from various locations manner. concluded warehousing could be generated into three components: configuration, databasing transmission. Research limitations/implications While framework research carried out counties, study’s findings foundation for future smart warehouse, related topics. provides guidelines practitioners low-cost obtain more accurate timely quick decision-making process. Originality/value at hand groundwork researchers explore proposed theoretical develop further increase inventory management efficiency operations. Besides, offers economical alternate organization implement software reasonably.

Language: Английский

---

An AIoT System for Bat Species Classification

Published: Jan. 27, 2021

Bat species are an integral part of our ecosystem and their monitoring can provide important insights into conservation tracking viruses like Covid-19. Given the difficulty high cost manually bats in natural habitats, this paper proposes Artificially Intelligent Internet Things (AIoT) system that uses audio-based Convolutional Neural Network (CNN) to monitor bat using echolocation calls. The Long Range Wide Area (LoRaWAN) send classified application server real-time. compared performance three different edge devices, Raspberry Pi Model (RPI) 3B+ (RPi), NVIDIA Jetson Nano, Google Coral two deep learning frameworks (TensorFlow Lite TensorRT). Although all devices were able do real-time inference (<; 0.5 seconds/inference for a 3-second audio segment), appears be best choice because it was fastest (0.3917 seconds/audio segment) required least resources (maximum %CPU Utilization = 29.2%). However, if concern then even RPI more than adequate task.

Language: Английский

---

A Novel E‐Sticker Based on Triboelectric Nanogenerators for Wireless Passive Communication

Advanced Functional Materials, Journal Year: 2025, Volume and Issue: unknown

Published: March 12, 2025

Abstract Wireless communication systems based on discharge‐induced displacement current exhibit significant potential for enhancing the convenience, security, and low power consumption of wireless systems. However, their practical applications remain largely constrained by complexity signals in both time frequency domains. Here, a novel compact passive system composed self‐powered e‐sticker (SWES) electronic circuits, enabling long‐distance through real‐time signal processing strategy, thereby applicable smart homes is proposed. The SWES seamlessly integrates triboelectric nanogenerator with an optimized plasma switch to ensure stable transmission under mechanical stimulation, achieving distance as high 13 m, while maintaining lightweight 0.24 g size 3.5 × 2.5 0.0167 cm 3 . Furthermore, multimodal home control that this design dedicated application, monitoring appliance status intelligent control, validating system's versatility demonstrated. proposed poised widespread deployment homes, facilitating various appliances powered municipal electricity holding substantial cities, wearable electronics, human–machine interfaces.

Language: Английский

---

Resource allocation solution for sensor networks using improved chaotic firefly algorithm in IoT environment

Computer Communications, Journal Year: 2020, Volume and Issue: 156, P. 91 - 100

Published: March 30, 2020

Language: Английский

---

Big data and the future of urban ecology: From the concept to results

Science China Earth Sciences, Journal Year: 2020, Volume and Issue: 63(10), P. 1443 - 1456

Published: Aug. 20, 2020

Language: Английский

---

AutoTrust: A privacy-enhanced trust-based intrusion detection approach for internet of smart things

Future Generation Computer Systems, Journal Year: 2022, Volume and Issue: 137, P. 288 - 301

Published: Aug. 4, 2022

Language: Английский

---

Toward a Unified TreeTalker Data Curation Process

Forests, Journal Year: 2022, Volume and Issue: 13(6), P. 855 - 855

Published: May 30, 2022

The Internet of Things (IoT) development is revolutionizing environmental monitoring and research in macroecology. This technology allows for the deployment sizeable diffuse sensing networks capable continuous monitoring. Because this property, data collected from IoT can provide a testbed scientific hypotheses across large spatial temporal scales. Nevertheless, curation necessary step to make heterogeneous datasets exploitable synthesis analyses. process includes retrieval, quality assurance, standardized formatting, storage, documentation. TreeTalkers are an excellent example applied ecology. These smart devices synchronously measuring trees’ physiological parameters. A set be organized mesh permit collection single tree plot or transect scale. such over large-scale needs approach curation. For reason, we developed unified processing workflow according user manual. In paper, first introduce concept TreeTalker process. idea was formalized into R-package, it freely available as open software. Secondly, present different functions “ttalkR”, and, lastly, illustrate application with demonstration dataset. With approach, propose establish new cyberinfrastructure allow activities networks. Our supporting life cycle by improving accessibility thus creating unprecedented opportunities TreeTalker-based macroecological

Language: Английский

---

A cascade ensemble-learning model for the deployment at the edge: case on missing IoT data recovery in environmental monitoring systems

Frontiers in Environmental Science, Journal Year: 2023, Volume and Issue: 11

Published: Oct. 26, 2023

In recent years, more and applied industries have relied on data collection by IoT devices. Various devices generate vast volumes of that require efficient processing. Usually, the intellectual analysis such takes place in centers cloud environments. However, problems transferring large long wait for a response from center further corrective actions system led to search new processing methods. One possible option is Edge computing. Intelligent places their eliminates disadvantages mentioned above, revealing many advantages using an approach practice. computing challenging implement when different collect independent attributes required classification/regression. order overcome this limitation, authors developed cascade ensemble-learning model deployment at Edge. It based principles cascading machine learning methods, where each device collects performs its it contains. The results work are transmitted next device, which analyzes collects, taking into account output previous device. All at-tributes taken way. Because this, proposed provides: 1) possibility effective implementation intelligent analysis, is, even before transmission center; 2) increasing, some cases maintaining, classification/regression accuracy same level can be achieved 3) significantly reducing duration training procedures due smaller number simulation was performed real-world set data. missing recovery task atmospheric air state solved. selected optimal parameters approach. established provides slight increase prediction while procedure. case, main advantage all happens within bounds computing, opens up several benefits

Language: Английский

---