Target Detection using Autoencoders in a Radar Surveillance System

Wagner, Simon; Johannes, Winfried

doi:10.1109/radar41533.2019.171267

Cited by 6 publications

(13 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Use with mmWave radar signals: radar system are prone to noisy data and can be significantly affected by unwanted or fake reflections (e.g., ghost reflections). Due to this, many radar applications use the AE encode/decode functionalities as a middle ground for the reconstruction of the desired observation such as anomaly analysis for human fall detection [209], person detection for surveillance systems [234] and indoor person identification [183].…”

Section: E Main Learning Techniquesmentioning

confidence: 99%

A Review of Millimeter Wave Device-Based Localization and Device-Free Sensing Technologies and Applications

Shastri

Valecha

Bashirov

et al. 2022

IEEE Commun. Surv. Tutorials

View full text Add to dashboard Cite

The commercial availability of low-cost millimeterwave (mmWave) communication and radar devices is starting to improve the adoption of such technologies in consumer markets, paving the way for large-scale and dense deployments in fifthgeneration (5G)-and-beyond as well as 6G networks. At the same time, pervasive mmWave access will enable device localization and device-free sensing with unprecedented accuracy, especially with respect to sub-6 GHz commercial-grade devices.This paper surveys the state of the art in device-based localization and device-free sensing using mmWave communication and radar devices, with a focus on indoor deployments. We overview key concepts about mmWave signal propagation and system design, detailing approaches, algorithms and applications for mmWave localization and sensing. Several dimensions are considered, including the main objectives, techniques, and performance of each work, whether they reached an implementation stage, and which hardware platforms or software tools were used.We analyze theoretical (including signal processing and machine learning), technological, and implementation (hardware and prototyping) aspects, exposing under-performing or missing techniques and items towards enabling a highly effective sensing of human parameters, such as position, movement, activity and vital signs. Among many interesting findings, we observe that device-based localization systems would greatly benefit from commercial-grade hardware that exposes channel state information, as well as from a better integration between standardcompliant mmWave initial access and localization algorithms, especially with multiple access points (APs). Moreover, more advanced algorithms requiring zero-initial knowledge of the environment would greatly help improve the adoption of mmWave simultaneous localization and mapping (SLAM). Machine learning (ML)-based algorithms are gaining momentum, but still require the collection of extensive training datasets, and do not yet generalize to any indoor environment, limiting their applicability. Manuscript received xxxx xx, xxxx . . .

show abstract

Section: E Main Learning Techniquesmentioning

confidence: 99%

A Review of Millimeter Wave Device-Based Localization and Device-Free Sensing Technologies and Applications

Shastri

Valecha

Bashirov

et al. 2022

IEEE Commun. Surv. Tutorials

View full text Add to dashboard Cite

show abstract

“…설계한 칩의 단일 채널 수신기 및 송신기 구조는 기 발표한 77 GHz 레이다 송수신 칩과 유사하다 [3] . [4] . 그림 3은 송수신 모듈과 기저대역 데이터 수집 보 드를 결합한 전체 구조도를 나타낸다.…”

Section: ⅱ 레이다 송수신 모듈unclassified

94 GHz 4 Rx, 1 Tx FMCW Radar System

Kim¹,

Kim²,

Park³

et al. 2022

J. Korean Inst. Electromagn. Eng. Sci.

View full text Add to dashboard Cite

This paper describes a 94 GHz FMCW radar system for unmanned vehicles using 94 GHz 1 Tx and 4 RX chips designed in a 65-nm CMOS process. The measured output power of the transmitter and the conversion gain of the receiver were 9.77 dBm and 26 dB, respectively. The simulated noise figure of the receiver chip was 8 dB. The Tx and Rx antennas were microstrip patch antennas with a gain of 11 dBi on a Duroid substrate. The chirp period of the sawtooth FMCW waveform was 50 μs. The baseband signal was bandpass-filtered with a passband gain of 26 dB and sampled at a rate of 50 MSPS with a 14-bit ADC. The output of the ADC was acquired using an FPGA. The distance and angle information of the target were calculated using the FFT-based algorithm. The processed results closely matched the calculated results based on the specifications of TRx chips and processing gain, which confirmed the validity of the developed radar system.

show abstract

“…AD is already used in various tasks outside of HRRP target recognition within the radar community: Laxhammar [2008] for example used Gaussian Mixture Model as a cluster model to achieve AD on velocities and positions. Going back to HRRPs processing, Auto-encoders (AE) were used in Wagner and Johannes [2019] and Wan et al [2019] respectively to detect targets in background profiles and to reject outliers using a weighted training objective combining HRRP rejection and recognition. In these two cases, AEs are used for AD thanks to their reconstruction error, assumed to be high for anomalous samples when trained on normal ones, as was done in Sakurada and Yairi [2014].…”

Section: Introductionmentioning

confidence: 99%

From unsupervised to semi-supervised anomaly detection methods for HRRP targets

Bauw

Velasco-Forero

Angulo

et al. 2020

2020 IEEE Radar Conference (RadarConf20)

View full text Add to dashboard Cite

Responding to the challenge of detecting unusual radar targets in a well identified environment, innovative anomaly and novelty detection methods keep emerging in the literature. This work aims at presenting a benchmark gathering common and recently introduced unsupervised anomaly detection (AD) methods, the results being generated using high-resolution range profiles. A semi-supervised AD (SAD) is considered to demonstrate the added value of having a few labeled anomalies to improve performances. Experiments were conducted with and without pollution of the training set with anomalous samples in order to be as close as possible to real operational contexts. The common AD methods composing our baseline will be One-Class Support Vector Machines (OC-SVM), Isolation Forest (IF), Local Outlier Factor (LOF) and a Convolutional Autoencoder (CAE). The more innovative AD methods put forward by this work are Deep Support Vector Data Description (Deep SVDD) and Random Projection Depth (RPD), belonging respectively to deep and shallow AD. The semi-supervised adaptation of Deep SVDD constitutes our SAD method. HRRP data was generated by a coastal surveillance radar, our results thus suggest that AD can contribute to enhance maritime and coastal situation awareness.

show abstract

Target Detection using Autoencoders in a Radar Surveillance System

Cited by 6 publications

References 9 publications

A Review of Millimeter Wave Device-Based Localization and Device-Free Sensing Technologies and Applications

A Review of Millimeter Wave Device-Based Localization and Device-Free Sensing Technologies and Applications

94 GHz 4 Rx, 1 Tx FMCW Radar System

From unsupervised to semi-supervised anomaly detection methods for HRRP targets

Contact Info

Product

Resources

About