Unobtrusive Detection of Respiratory Rate through ...

Unobtrusive Detection of Respiratory Rate through UWB-Sensing for Applications of Ambient Assisted Living Busch, Björn-Helge; Nguendon Kenhagho, Hervé; Welge, Ralph Günter Published in: IADIS International Press

Publication date: 2012 Document Version Early version, also known as pre-print Link to publication

Citation for pulished version (APA): Busch, B-H., Nguendon, H., & Welge, R. (2012). Unobtrusive Detection of Respiratory Rate through UWBSensing for Applications of Ambient Assisted Living. In: IADIS International Press: Applied Computing 2012 (pp. 91-98). IADIS Press.

General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal ? Take down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

Download date: 10. Aug. 2017

UNOBTRUSIVE DETECTION OF RESPIRATORY RATE THROUGH UWB-SENSING FOR APPLICATIONS OF AMBIENT ASSISTED LIVING Dipl. Inf. Bjoern-Helge Busch1 [email protected]

MSc. B(Hons). Eng. Hervé Nguendon Kenhagho1 [email protected] or [email protected],

Prof. Dr. Eng. Ralph Welge2 [email protected] Leuphana University of Lueneburg, Institute VauST Volgershall 1, 21339 Lueneburg, Germany

ABSTRACT This contribution deals with an innovative approach for non-stigmatizing and continuous acquisition of vital signs for applications in the context of Ambient Assisted Living – AAL. Utilizing UWB-sensor components, the aimed vital sign acquisition is based on a mixture of collaborative techniques of preprocessing, feature extraction and information fusion for the treatment of multidimensional reflection data sets from the UWB-sensor. After a brief introduction to the application context including specific user demands, physical fundamentals of measurement and utilized electronics, the applied principles of spatial and temporal data mining are described. Hereafter, the implementation within an AAL-infrastructure is explained and tested in a real world scenario resp. in our living lab mapping a typical residential of a senior citizen. Additionally, preliminary measurement campaigns prove accuracy and robustness of our approach and our further improvements according to certain limitations caused by sensor topology. The discussion of results, the conclusion and outlook close the article. KEYWORDS

Ultra wideband Radar (UWB), Vital Sign Acquisition, Respiratory Rate Detection, Telemedicine, Unobtrusiveness

1. INTRODUCTION In recent years, Ambient Assisted Living – abbreviated by the acronym AAL - has become a distinct area of research and covers technical concepts, products and services for the unobtrusive support of elderly people with special demands at home in their daily life concerning physical or mental issues. Against this background, human-centered assistance systems, which provide domain specific services including situation aware regulatory interventions within the user's environment, gained much more importance. These services address different topics as health, comfort, sustainability, security and personal safety and are executed in relation to the approximated user situation. Thus, situation awareness implies the context sensitive evaluation of distributed sensor- and actuator-networks also as regarding user preferences, user characteristics and the apriori knowledge about the domain. In respect to the target group, services dealing with health problems take an exposed position for the design of the services. Possible functionalities care about early emergency detection, cooperative compliance control, intelligent ambient medication management including reminderfunctionality and continuous diagnosis systems to initiate preventive interventions.

Mobile Application for Android Doctor’s Diagnosis Assistant

Assistance System OSGIMiddleware

DB Rules & CFG

Service level n

DB Adaption

...

Emergency Call

...

Collateral Data

... Service level 1 Service level 0

REST HTTP

Domain

SSL

HR

BP RR

MDA – Smartphone

Webservice

Client-Server based AIS Internet

...

XML/JSON

Emergency Call

DB AEDL

...

Internet

Data Transmission includes Prediagnosis, Trends and reduced Context Information. Emergency Call authorizes for direct link

Weight

DB Preproc.

Doctor

DB Rawdata

Kin People

BA States Position

Posture

Caretaker

BA Events Observable n

Hospital

Figure. 1 Infrastructure of the assistance system

Due to public sponsorship, various research projects were launched, dealing with miscellaneous approaches for the recognition of specific life patterns resp. Activities of Daily Living – ADL. Our approach for activity aware vital sign threshold analysis is part of the research project AAL@Home which is engaged in the design and development of a user related assistance system as a service orientated software infrastructure adopting various information sources as shown in figure 1. The assistance system, consisting of a data mining server (DB) for the storage of manifold information and the assistance system server itself providing various services in form of OSGi-software-bundles, collects and explores any accessible information such as heart rate (HR), blood pressure (BP), information from the building automation (BA) like events or states or e.g. the user's position. In accordance to the claimed unobtrusiveness, non-stigmatizing and continuous measurement of vital signs is an open issue we want to solve through the utilization of UWB-sensing a thereinafter proposed and discussed.

1.1 Challenges of the telemedicine For the automatized preventive diagnosis of deteriorating states of health, it is obviously essential, that the significant parameters are observed continuously with a sufficient accuracy. Common telemedical devices have numerous deficits considering usability, significance and validity of gathered raw data and questions of ethics:  Patients have to use telemedical devices autonomously by their own. In accordance to anamnesis and latest therapy recommendations, this has to be done at determined instants of time. Otherwise, the raw data is not trustworthy or out of sync with the aim of acquisition. The extracted information may be diagnostically less conclusive.  In addition, the usage of telemedical devices is uncomfortable and inconvenient. Further, body attached sensor networks (BSN) restrict the users in their autonomy, agility and mobility.  The usage of BSNs is not only a question of comfort but also a factor of stigmatization. Therefore, in the context of AAL, visible measurement of vital signs must be avoided in home care scenarios.  Imaging techniques based on the evaluation of cameras for the identification of position, posture, activity, agility and emotional shape are inappropriate. The users reject, or often accept them with reservations because this approach procures an impression of observation. In respect to these adducted disadvantages, it is necessary to replace camera-based systems or BSNs by contactless measurement devices which are integrated as a stand-alone component or a data point within a multi-sensor network resp. a complete data acquisition infrastructure. To improve our multi-level approach for a human-centered assistance system, we prefer the utilization of ultra-wideband radar to achieve longterm vital sign measurement campaigns without any notification by the user. The hardware is either embedded within furniture for resting as such as armchairs or mounted on walls as single sensor nodes with wireless communication interfaces (ZigBee) for position tracking or posture recognition.

2. METHODOLOGY Focusing on the early detection of deteriorating states of health, we are constrained to features measureable with non-invasive methods or through surrogating signals describing the real physical phenomena (e.g. bioelectrical signal → mechanical signal). Undergoing a complete medical examination in a hospital affords access to various significant vital parameters. Keeping the signal properties of the UWBradar in mind, thoracic impedance, respiratory rate, heart rate and in especially, breathing patterns are in the focus of further considerations → the model regarding any available source of information is downscaled. In this paper, we concentrate only on the detection of respiratory rate for diagnosis purpose.

2.1 Ultra wideband Sensing – Current Developments UWB-systems with various types of signal modulation are propagated and tested for diverse applications. In order to discover hidden objects or in particular humans through massive walls by the usage of a msequence radar was proposed by [Sachs, 2005]. Another approach for through-wall radar was introduced by [Judson, 2009]. The assignment of localisation tasks in the context of automobile parking system to UWBsolutions was proposed by [Mary, 2008]. Recapitulatory, the elaborate survey about existing through-wall imaging technologies from [Aftanas, 2009] grants a fruitful overview about the state of the art in this area of designated applications. The benefit of UWB-components for BAN-communication in the domain of telemedicine is part of the work of [Hernandez, 2011]. Concerning our focus of work, UWB-radar was also used to detect vital parameters. In order to correct and calibrate MRT-systems, [Thiel, 2009] introduced ultra-wideband sensing for the detection of the patients exact position regarding imaged body tissue, in particular the chest. The examination of different organic materials in accordance to their content of water is part of the work of [Helbig, 2007] which coincides with the specifications in the following section, our UWB-utilized hardware and the experimental setup is briefly described.

2.2 System Setup This section describes our experimental hardware setting and implementation setup. To detect dynamic parameters of health, we utilize a ultra wideband radar system, developed by our project partners from the TU Ilmenau and add algorithms for signal processing and feature extraction.

2.2.1 Performance of the UWB-radar The performance of the UWB-radar is limited by its maximum detection range, radar resolution, search volume and the revisit time of a target under test by measuring the scattered RF-microwave signal. In order to identify, reflected waves resp. echoes from the transmitter to the object need to be analyzed.

Figure 2: Basic Radar system under test with two antennas (Only most useful transceiver signal paths are illustrated)

The simplest way to extract the desired information as such as ranging or material coefficients is an experimental setup as illustrated in Figure 2; one transmitting and one receiving antenna for the electromagnetic waves positioned in a fixed arrangement including a target/tube filled with fluid in

accordance to [Sachs et al, 2000] and [Wiesbeck et al, 2009]. Indeed, a is the column vector of the normalized guided waves transmitter from the antenna feeds to the physical layout antenna, b is the normalized waveguide receiver from the antennas to their feedings (refer to fig.2). In order to have an appropriate power transfer, impedance from the transmission line to the front-end antenna obviously should match. To prevent distortion of the generated pulse, an ideal UWB antenna produce radiation field with constant magnitude and a linear phase shift according to frequency.

2.2.2 Maximum Length Sequences (MLS) We are still using an m-sequence radar system; a sensor node with a specific modulation type which offers a couple of advantages in comparison to other approaches. Therefore, to point the general functional a brief introduction is previously given to the basics of modulation technique. In practice, the impulse response of a system is measured by recording the response cause by an emitted signal resp. the generated pulse. Due to the signal-noise ratio (SNR), the magnitude of the impulse has to be huge. Among different types of spreading sequences such as Gold sequences, Kasami sequences, Walsh Hadamard sequences, the focused spreading sequences is the maximum length pseudo noise (PN) sequence (refer to fig. 3). A MLS is also sometimes called n-sequences or m-sequences. MLSs are spectrally flat, with the exception of a near-zero DC term. These characteristics of flatness and near DC operationally make the MLS suitable for our baseband sensor [Natarajamani, 2010]. By definition, Maximum Length PN Sequences are binary sequences that are able of transferring all possible arrangements of binary sequences in 2m-1 repetitive shift, where m is the Linear Feedback Shift Register (LFSR) use to generate the sequences. In practice, the LFSR feedback connection of such m-sequence is realisable by using the primitive polynomial. In fact, every polynomial which cannot anymore be factorized by another polynomial is called primitive. Thus, a generator polynomial is said to be primitive if it cannot be factored and if it is a factor of X N+1, with N = 2m-1 as the m-sequence length [Niu, 2010]. The values of the sequence are used as the amplitudes of a sequence of transmitted RF pulses, using the values from right-to-left. In the case of a perfect impulse response, as the echo is received and sampled, the incoming samples simply replicate the sequence. A buffer of length 9 is initially filled with zeros. As the pulses come in, they fill the buffer, entering from the left. MLS are inexpensive to implement in hardware or software, and relatively low-order feedback shift registers can generate long sequences; a sequence generated using a shift register of length 9 is 2 9 − 1 samples long (511 samples). Let construct 7 sample length m-sequence generators with 3 registers (m=3), then the primitive polynomial that determines feedback connections is as follows [Nam, 2010]: (1)

N = 2^ m -1 = 2^3 -1 = 7

where N is the time index or number of sample to be used and m the number of registers. A necessary and sufficient condition for the sequence generated by a LFSR to be maximal length is that, its corresponding polynomial be primitive (refer to equation 2).

X^ N +1 = X^7 +1 = (X +1) * (X^3+ X^2 + 1) * (X^3+ X + 1)

(2)

Since the number of registers are m=3, we have to choose a primitive polynomial that is of degree 3. From the above equation, two choices are: for m = 3, ie. X^3 + X^2 + 1 And X^3 + X + 1 . These polynomials are often represented as [3 2 1] and [3 1 0] respectively. The schematic of for MLS of length 7 that generates a polynomial = [3 1 0] is expressed as P = [1 0 1 1] or P=[x3 x2 x1 x0]. Register 1

x0 = 1

Register 2

Output of M-Sequences

Register 3

x1 = 1

Fig.3 : LFSR connections of a PN M-Sequences

x3 = 1

2.2.3 Measurement Approach for Respiratory Rate Recognition The measurement of respiratory rate is done by two main tasks: The isolation of the dynamic content itself within the reflection data and the feature extraction from the isolated signal regarding frequency, amplitude and the variances of these parameters. First of all, the reflection from the background covering the static content as such as walls or furniture including environmental noise is sampled and used for the background removal. After the background removal, dynamic values resp. varying echoes corresponding with breath rate can be detected within; only the differential between static content and current sampling remains in the data package for further treatment via technologies of signal processing. The loudness of echoes is related to the amplitude of the wave, which is also related to the energy of the wave. Loudness is also related to intensity, a measurable quantity, and also to sound pressure level (SPL), also a measurable quantity reflected back from the human chest. These quantities are stated in units of decibels (dB) and related to the target movements, characteristics and distance. One example for these differential images of the environment, produced by a couple of samples, is depicted in the left half of figure 4. The color of the plot is corresponding with the signal amplitude resp. the loudness, the vertical axis is related to the signal propagation time resp. distance and the horizontal axis is related to the number of sample resp. time. In this picture, the dynamic content is the sinusoid-like curve in the upper half of the plot, resulting from a mechanical signal generated by a linear drive. The isolated content is analyzed by a peak-peak-detection in order to find the boundaries for the examination window – vertical and horizontal size of the window and the localisation within the raw data frame. The current algorithm to find signal properties implements a windowed Fast-Fourier-Transform.

Figure. 4 Test measurement of a generated, deterministic mechanical signal

To reduce FFT- leakage effects, we use windowing techniques focusing on the Hamming and Hanningwindow to minimize magnitude of side-lobes. In the right half of figure. 4, signal properties in the frequency domain are illustrated. The huge dc-content result from the windowing process which adds in practice some offset to the signal but is compensated easily.

3. EXPERIMENTS 3.1.1 Preliminary Measurement Campaigns for System Property Identification The first task to solve was the identification of the system properties as such as output power, real antenna characteristics or output spectrum (refer to Figure 5-6), simply done by a short-circuiting of the transmitterreceiver channels. Due to spectrum sensitive examinations like material/body tissue analysis, the identification of non-linearity or frequency gaps is indispensable. In addition, we try to improve our previous hardware setup by the installation of some shielding material to suppress disturbing effects of backscattering.

Figure5: The coupling process of UWB Sensor

Figure 6: The UWB Sensor bandwidth at -10 dB

The results in fig. 6 show clearly that the UWB sensor, equipped with attenuators of 30 dB, has a frequency range from near DC to 2.9 GHz (BW -10dB = 2.9 GHz) with an error of 0.14624%. This error is caused by internal RF-components as such as the ADC, filters, amplifiers and others. A decrease of bandwidth down from 0.5 GHz to 2.5 GHz (2.0 GHz) and amplitude attenuation of 55.4015% are registered when measuring the impulse response of the average UWB radar at 50 cm from each patch antenna. Indeed, the BW -10 dB of this system is 2.0 GHz. The BW-10dB restriction is due to the lowest and upper patch antennas frequency as shown in Figure 6. As mentioned before, another issue was the reduction of the unwanted side-lobes of our patch antenna interferes with moving objects behind the armchair; the aerials are embedded within the armchair for continuous measurement (Refer to Figure 10). Reflections caused by moving persons (or pets) belong to the same dynamic range as the aimed respiratory rate and overlap with the desired signal. To solve this problem, we tested some materials for shielding which have an operating range of 0.5 GHz to 2.5 GHz and meet our conditions for absorption. The equation R = 20log(E_t - E_r) expresses the characteristic of a RF/microwave absorber material under test; R corresponds with the amplitude of the absorber wave performance in dB. Figure 8 is related to the complete system setup; the plot (fig. 8) shows the efficiency of the selected absorber material ( E_t and E_r relate to amplitude of the received wave with and without the absorber). When adding another absorber with the same characteristic and size, we found a new attenuation of 98.7645% with and error of 0.17447% (refer to Figure 9). This means a further absorption of 11.9473%. In fact, this is in accordance to [Nornikman et al, 2009], a very good balance with the shape (more than one wavelength) of the material, and the thickness of the doped carbon is required for absorber performance in magnitude and phase.

Figure7: BW of Sensor before and after the patch antenna

Figure8: Amplitude wave attenuation measured without and with the absorber

Figure9: Measurement without and with two identic RF/microwave absorbers

Figure10: Embedded antenna into armchair with RF/microwave absorber

3.1.2 Preliminary Measurement for Feature Extraction – Finding the System Resolution To test the accuracy of the sensor device and the implemented algorithms to detect movement and motion, we utilized a linear-drive in order to create a deterministic mechanical signal. This linear-drive is configured via a CNC-script with exact values for acceleration ramps, end velocities and motion distances. Analyzing the gathered reflection data, we extracted the motion signal as depicted in Figure 4. The left half of the drawing shows the radargram of gathered reflection data. In the radargram, the reflection caused by the linear-drive's continuous motion (forwards and backwards) complies with a sinusoid with a period of 2.73s. It is a very significant and characteristic signal covering an oscillatory motion over a distance of 47 cm. The right half of the drawing shows the spectrum of the signal only filtered by an ordinary moving average process. The identification of the motion frequency can easily be done; the noise content in the signal is negligible. In order to find the lower sensitivity limit, we configured the linear-drive for smallest possible motion. Consecutive measurements have shown a reachable accuracy of 99.8 % in comparison to adjusted motion characteristics for large distances. In this case, the identifiable accuracy of the measurement device was limited by the signal generator itself. Small signals with a low movement range of 1 mm are detected in a significant and reliable manner by the sensor node with an accuracy of 97.6 %. This fact proves robustness and fitness for our desired vital sign acquisition functionality (refer to fig. 11). The first radargram in the left part shows a significant reflection in the top region. The radargram in the center shows the isolated, windowed signal while the radargram at the bottom documents the power of the reflection signal energy. Extracted from the radargram, the isolated impulse response from the moving, not to say just trembling linear drive, is depicted in the right half of figure 11 with significant frequency component nearby 6 Hz.

Figure 11: Plots documentation a mechanical signal with 1mm movement range

3.1 Real World Measurement After finding the sensitivity of the sensor device, the hardware was integrated within the armchair's backside. The next measurement campaigns concentrate on the identification of real vital signs from humans as such as heart rate or breathing rate. The prototype was tested within our laboratory representing the typical environment of a senior citizen which is called the living lab. There are manifold sources of information subsuming home automation components or telemedicine devices. The embedded sensor node within the armchair is a complementing component inside this information infrastructure (also Figure 1 in section 1).

Figure 12: GUI of the vital sign acquisition tool

For demonstration purpose, the extracted features as such as heart rate or respiratory are transferred to a mobile application called the Medical Diagnosis Assistant - MDA. This application supports the doctor in making the diagnosis and reconciles anamnesis, current disease, individual threshold values for health parameters, therapy recommendation et cetera. Additionally, feature validation is executed in the particular user context to detect abnormal values. In fig. 12, the GUI of our data processing software is depicted. For demonstration effort, some plots care about raw data (radargram) and the different extracted features sorted by their dynamics. The movement of the abdomen is plainly visible in the middle plot in the left half of the GUI. To the right, the extracted signal from the heart muscle is plotted. Below them, their spectra are drawn. Field campaigns, carried out in the living lab with five different persons, concerning their physical appearance and shape, yield promising results regarding accuracy and robustness of the measurement system (refer to the results documented in table 1). Thereby, the results are examined and evaluated in comparison to a reference system from the telemedicine.

4. DISCUSSION OF THE RESULTS The aim of unobtrusive vital sign acquisition via UWB is accomplished and will soon be tested in a clinical trial. Concerning the detection of respiratory rate and the implemented improvement of the hardware setup (shielding and antenna), our approach proves reliability and worth for telemedical implementations. The shielding has a positive influence for respiratory rate recognition. In addition, the accuracy increased after recalibrating the position of the aerials. Test Person A (m), 1: 84 m,74 kg B (m), 1: 72 m,82 kg C (m), 1: 92 m,86 kg D (m), 1: 81 m,78 kg E (f), 1: 67 m,59 kg

RR Detection Rate 1(before shielding and RR Detection Rate 2 (after shielding and antenna adjustment) antenna adjustment) 86 % 99.5% 73% 98.8% 84% 99.4% 87% 98.6% 71% 98.7% Table 1: Measurement Results

Thinking about heart rate detection, which is also part of our work, there are still open issues. Accuracy for the detection of signals in this range drops in comparison to respiratory rate detection. In addition, the

accuracy depends on the constitution of the person and the position of the antenna. Obviously, the coverage of the interesting body region by the sensor resp. antenna causes this abnormality in accuracy. Possible solutions are part of the following section.

5. CONCLUSION AND OUTLOOK The first experiments with the sensor device and the implemented techniques for feature extraction reveal a great potential for unobtrusive vital sign acquisition in the context of geriatric care. In addition, collaborating with intelligent and predictive situation recognition processes, the single sensor node offers the opportunity to adapt to different domains in order to realize an early emergency detection. Respiratory rate measurement proves robustness and a high accuracy in comparison to our reference sensor device. But however, due to the fact that measurement accuracy depends on the patient's shape and stature, it is important to revisit the number of receiving aerials embedded within the armchair. In order to detect much smaller mechanical signs in human’s chest as heart rate, this is a mandatory task. Therefore, the evaluation of an array of four receiving antennas offers the opportunity to select the best fitting one; the one with the highest significance will be chosen for further data analysis. The next steps will be to investigate the best positions of our transmitter and receiver antennas such as the trans-receiver antenna could point the dorsal or ventral para-axial. The use of multiple channels and coupling Patch antenna or m-array antennas for better data collection is another vision. Concerning the problem of varying frequencies and signal modulation types, the identification of specific signal content as heart rate variances, signal rise times or the isolation of characteristic waveform contents like QT-time considering ecg-recordings, a switch from periodic signals (sinusoids) to wavelets for reconstruction is unavoidable and part of the upcoming working package in research. And finally, thinking about real world scenarios, there is still another open issue; the compensation of overlapping signals as such as WLAN.

REFERENCES Book Constantine A. Balanis, 2005. ANTENNA THEORY: Analysis and Design. John Willey & Sons, New Jersey, USA and Canada. Constantine A. Balanis, 2012. ADVANCED ENGINEERING ELECTROMAGNETICS. John Willey & Sons; Arizona State University, USA. Richard G. Lyons, Understanding Digital Signal Processing. Bernard Goodwin, New Jersey, USA Michal Aftanas, 2009. Through Wall Imaging With UWB Radar System". PhD thesis. TECHNICAL UNIVERSITY OF KOSICE Faculty of Electrical Engineering, Informatics Department of Electronics, and Multimedia Communications Journal F. Thiel et al. 2009, Combining magnetic resonance imaging and ultra-wideband radar: a new concept for multimodal biomedical imaging. In: REVIEW OF SCIENTIFIC INSTRUMENTS 80. Wiesbeck W. et al, May 2009. Mikrowellenunterstützte Verleimung von Parkett im Durchlaubetrieb. Parkett magazin, vol. 03 /2009, pp. 54. Wiesbeck W. et al, 2011, Impedance Matching of a Coaxial Antenna for Microwave In-situ, Processing of Polluted Soils. Journal of Microwave Power and Electromagnetic Energy, vol45, No. 2, pp.70-78. Zetik R. et al, 2007. The Architecture of a Baseband, Pseudo-Noise UWB Radar Sensor. In IEEE instrumentation & Measurement magazine, Vol. 07 No. 07, pp.33-45. Conference paper or contributed volume B.H. Busch and R. Welge., 2011, Domain Specic Services for Continuous Diagnoses in the Context of Ambient Assisted Living - AAL. In: Proceedings of the 2011 International Conference on Data Mining (DMIN11) at 2011 World Congress in Computer Science. B.H. Busch et al. \Preventive emergency detection based on the probabilistic evaluation of distributed, embedded sensor networks".In: Ambient Assisted Living: 4. AAL-Kongress 2011 Berlin. Springer

M. Hernandez and R. Kohno. \UWB systems for body area net-works in IEEE 802.15.6". In: Proc. IEEE Int Ultra-Wideband (ICUWB) Conf. 2011, pp. 235{239. doi : 10.1109/ICUWB.2011.6058835. Zheng Li and G. Gielen. \UWB signal acquisition in transmit-only networks". In: Proc. IEEE Int UltraWideband (ICUWB) Conf. 2011, pp. 126{129. doi : 10.1109/ICUWB.2011.6058809. G. I. Mary and V. Prithiviraj. \Improved UWB localization tech-nique for precision automobile parking system". In: Proceedings of TENCON 2008 - 2008 IEEE Region 10 Conf. 2008, J. R. Panda, P. Kakumanu, and R. S. Kshetrimayum. ,A wide-band monopole antenna in combination with a UWB microwave band-pass filter for application in UWB communication system". In: Proceedings of the. Annual IEEE India Conf. (INDICON). 2010 Z. Slimane, A. Abdelmalek, and M. Feham. \OFDM Based UWB Synthetic Aperture Through-wall Imaging Radar". In: Proc. Third Int Broadband Communications, Information Technology & Biomed-ical Applications Conf. 2008 Marko Helbig et al., 2007, Ultrabreitband-Sensorik in der medizinischen Diagnostik. In: 41. Jahrestagung der Deutschen Gesellschaft für Biomedizinische Technik BMT, Aachen, Germany. Sept. 2007. J. Sachs et al., 2005, Through-Wall Radar. In: Proceedings of the IRS J. Sachs., 2004, M-sequence radar. In: Ground Penetrating Radar 2nd edition, IEEE Radar, Sonar, Navigation and Avionics Series 15 J. Sachs et al., Recent Advances and Applications of M-Sequence based Ultra-Wideband Sensors". In: Proceedings of the ICUWB, Singapore. Sept. 2007. SangHyun Chang et al., 2011, UWB human detection radar system: A RF CMOS chip and algorithm integrated sensor. In: Proc. IEEE Int Ultra-Wideband (ICUWB) Conf. 2011, pp. 355 Nornikman H. et al, 2009, Performance of Pyramidal and Wedge Microwave Absorbers. Kuala Lumpur, Malaysia, pp. 649-654 A. Judson Braga and Camillo Gentile. 2009, An Ultra-Wideband Radar System for Through-the-Wall Imaging using a Mobile Robot. In: Proceedings of the ICC'09 - IEEE International Conference on Communications. Sachs J. et al, 2000, Ultra-Wideband Principles for Surface Penetrating Radar. Ilmenau, Germany, pp. 1-11.