ISMICT 2014 1569883335
Receiver performance evaluation on IEEE 802.15.6 based WBAN for monitoring Parkinson’s disease Mariella Särestöniemi, Ville Niemelä, Matti Hämäläinen, Jari Iinatti Centre for Wireless Communications, University of Oulu, Oulu, Finland. Email to corresponding author:
[email protected]
Niina Keränen, Timo Jämsä Department of Medical Technology, University of Oulu, Oulu, Finland. Medical Research Center Oulu, Oulu University Hospital and University of Oulu, Oulu, Finland. Jarmo Reponen, FinnTelemedicum, University of Oulu, Oulu, Finland and Raahe Hospital, Raahe, Finland.
wireless health monitoring and its applications have achieved a great interest in the recent years, e.g., [1-2].
Abstract—Advantages of remote health monitoring are incontestable: prolonging the patient’s possibility to stay at home by monitoring the health condition remotely, the quality of patient’s life is improved as well as healthcare costs are significantly reduced. Parkinson’s disease is one example, for which some wireless monitoring systems have been presented recently. However, concrete performance evaluations for the IEEE 802.15.6 standard based communication in the realistic use scenarios have not been presented yet in the literature. The aim of this paper is to evaluate performance of ultra wideband energy detector receivers on the physical layer of the IEEE 802.15.6 standard. The monitoring system and node locations are designed for monitoring Parkinson’s disease symptoms. Measurement based radio channel models were used in the evaluations. Two modulation methods, on-off keying and pulse position modulation, are evaluated and compared using two different antennas suitable for body area network communication. Pulse position modulation based system is found to outperform on-off keying based system, which on the other hand, is simpler to implement. Furthermore, it is shown how antenna placement, position, and properties are impacting on the channel characteristics of the radio link, and hence, on the bit error rate performance.
There are several challenges in the design and implementation processes of wireless applications targeted for health monitoring. Data transmission should be fast, energyefficient, highly robust, secure and dependable. In order to avoid harmful effects of the electromagnetic propagation on the human body, low-power communication and use of special antennas is crucial. For successful implementation of medical and non-medical wireless body area network (WBAN) applications, IEEE created and published the standard 802.15 for low-power in-body/on-body node communication [3]. The standard defines one medium access control (MAC) layer which is supporting three physical (PHY) layer solutions: narrowband (NB), ultra wideband (UWB), and human body communication (HBC), targeting for different application requirements. [3-4] Parkinson’s disease (PD) is a relatively common neurodegenerative disorder strongly associated with age. Due to its measurable motor symptoms and fluctuating states that complicate clinical assessment [5], it is an interesting target for WBAN monitoring. The first detailed IEEE 802.15.6 based multisensory application scenario for monitoring Parkinson’s disease was presented in [6]. The same system can also be used for, e.g., fall detection, which is important due to the postural instability associated with the condition. Literature on multi-accelerometer systems prior to [6] have considered the older IEEE802.15.4 standard or off-line analysis and have not discussed the physical communication requirements or beyond BAN communication in detail.
Keywords—health monitoring; performance evaluation; ultra wideband; wireless body area networks;
I.
Juha Partala, Tapio Seppänen* Department of Computer Science and Engineering, University of Oulu Oulu, Finland. *Medical Research Center Oulu, Oulu University Hospital and University of Oulu, Oulu, Finland.
INTRODUCTION
The healthcare sector in all developed countries is facing challenges due to the rapidly aging population. One of the main questions is how to deliver high quality healthcare services to an increasing number of people using limited financial and workforce resources. One solution is to monitor health conditions of certain patient groups remotely using advanced wireless communication techniques and applications. Remote health monitoring could enable patients to stay at home for a longer time, and thus, increasing their quality of life and decreasing healthcare cost. Therefore,
This paper considers WBAN communication system presented in [6] for monitoring Parkinson’s disease and aims to evaluate performance of energy detector (ED) receivers
1
used in the system. The primary focus is to show bit error rate (BER) evaluations with two modulation schemes defined by the IEEE 802.15.6 standard’s default mode. A measurement based radio channel model is used in the evaluations for the determined WBAN links that correspond to the sensor locations to be used in PD monitoring. Radio channel measurements are conducted in two phases: first in an anechoic chamber in order to get a pure link performance without the impact of the environment, and next in a simple room. The focus of this paper is to show bit error rate (BER) performance results with channel measurement data obtained in an anechoic chamber.
Figure 1. System for monitoring Parkinson’s disease
This paper is organized as follows: Section II summarizes the scenario for monitoring Parkinson’s disease. The system model description is presented in Section III and numerical results are shown in Section IV. Finally, summary and conclusions are given in Section V. II.
from nodes N1-N4 to the hub while excluding the link between the fall detector and the hub (N5 -H) as the fall detector is assumed to be attached to the waist worn device (body hub). In the default mode, standard defines impulse radio UWB (IR-UWB) with on-off keying (OOK) modulation as a mandatory radio technology. On-off signaling is considered as an excellent modulation scheme for BAN devices operating on the human body due to the energy efficiency, low risk for temperature rise and low implementation complexity [11]. With OOK based systems, energy detectors are used as receivers. Energy detectors are attractive WBAN receiver candidates because they have low complexity and thus low power consumption. They are also very simple since they do not require, e.g., complex channel estimation. However, estimation of energy threshold is necessary with OOK based systems, and it can sometimes be challenging. As pointed out in [12], even small changes in the accuracy of the threshold can have a big impact on the detection performance.
THE SCENARIO FOR MONITORING PARKINSON’S DISEASE
Figure 1 represents the generic block diagram of the system for monitoring Parkinson’s disease [6]. The monitoring system consists of five triaxial accelerometers and a waist-worn device including the central in-body hub (H). Accelerometers are placed on each limb for monitoring the activity of a patient (nodes N1-N4), and one on the waist (node N5). A node near the center of body mass is especially important for fall detection [7]. Additional nodes might provide information of proximal tremor, but their inclusion was not deemed worth the hit to usability. The intelligent nodes process data and produce five descriptive values per axis using 5-10 s window. The preprocessed data packet of 600 bits is sent to the central hub on the waist worn device every 5 s. The central hub implements a system for analyzing data, which is then further transmitted to the medical server and predefined recipients, such as family members, through local area network (LAN) gateway or directly through mobile networks. The focus of this paper is on the WBAN communication between the sensor nodes and a hub. WBAN is realized according to the requirements determined by the IEEE 802.15.6 standard and realistic channel characteristics. More details about the monitoring scenario can be found from [6]. III.
Energy threshold estimation can be avoided by using pulse position modulation (PPM), in which the receiver simply makes comparison of energy levels captured in two separate predefined time intervals. It is slightly more complex than an OOK based receiver, but it has improved power efficiency and lower bit error probability [12-14]. PPM modulation is the mandatory modulation for IR-UWB in the IEEE802.15.4-2011 wireless personal area networks (WPAN) standard [15] but it is also included in the IEEE 802.15.6-2012 standard due to the waveform coding utilized with OOK (annex C) [3]. More details about the properties of OOK and PPM modulation techniques can be found from [12-14]. Based on the waveform coding for OOK signaling, a receiver can detect the same OOK modulated signal by using PPM and yet, receive the same information with the same bit rate. We chose, for comparison purposes, to use both OOK and PPM demodulations with energy detection receivers.
SYSTEM MODEL DESCRIPTION
A) Background The IEEE 802.15.6 standard [3] determines two modes of operation for UWB communication: default mode for medical and non-medical applications and high quality of service (QoS) mode for high-priority medical applications. UWB technology provides several advantages for short-range communication in the close proximity of the human body: dependable high data rate transmission with low complexity, low power, and low cost solutions, and is robust against interference and jamming. [8-10] In our scenario, the monitoring data on the links between the limb sensors and a hub can be transmitted using the default mode, whereas the high Quality of Service mode is required for the link between the fall detector and the hub [6]. This study focuses on the link performance evaluation for links
The performance evaluations are conducted using Matlab. The simulation model is developed according to the IEEE 802.15.62012 definitions. The simulation scenario consists of transmitters sending bursts of pulses, a realistic multipath propagation channel, and receivers which receive the propagated signal and process data. The receiver is assumed to know the exact transmission instants, i.e., it is perfectly synchronized. The propagation channel is modeled using measurement based radio channel data, which is explained in
2
Section III B in detail. The transmitted signal is also corrupted by the additional white Gaussian noise (AWGN) whose level is varied in order to evaluate performance with different signalto-noise ratio (SNR).
are discussed in [17-18]. More details about the measurement setup and related results can be found from [19]. C. Receiver structure An energy detector receiver was implemented to detect the received signal r(t) with two different demodulations methods, OOK and PPM. In the ED receiver, the propagated signal is passed through an ideal band-pass filter for noise reduction. The decision variable of the first stage of the ED receiver is expressed as
B) Signal model A. Transmitted waveforms For the transmitted OOK (and PPM) modulated signal H xm(t) during the mth symbol interval, the waveform is ∑ ,
,
(1)
(2)
where Tw is the length of the received burst and corresponds to the integration time used by the energy detector and q = mTsym+nTsym/2+hTw. With the OOK demodulation, the integrated energy from (2) is compared to a predefined energy threshold , which is independently defined for each SNR level. The mth received bit is determined using
where is the nth code word component over the mth symbol, Tsym is the symbol duration, h(2Km+n) is the pseudorandom time-hopping sequence, Tw is the length of a burst and wn(t) corresponds to a scrambled burst of Ncpb pulses. The variable K is set to 1, i.e., one information bit is expressed with a two-bit code word per one symbol. For a bit zero, code word bits [1 0] are transmitted per one symbol and for a bit one as [0 1]. The expression ‘transmitted’ is used despite the fact that for OOK the lack of transmission is indicating bit ‘0’. In the simulations, the energy of each transmitted burst is normalized to one both with OOK and PPM..
" "
.
(3)
" "
If the squared and integrated signal sample is greater than the predefined threshold, the received bit is ‘1’, otherwise it is ‘0’. Due to the symbol mapping, i.e., the 2-bit code word for a data bit, there is a possibility for miss-detection by detecting either [0 0] or [1 1] within one symbol interval. If the detection result is one of the previous code words, the decision is made according to the latter bit. In other words, if the 2-bit code word is controversial, the decision is returned to a basic OOK modulation; no transmission indicates data bit ‘0’ and a transmission indicates data bit ‘1’.
B. Channel measurement data In this study, measurement based radio channels’ impulse responses were used for the arm-waist (N1-2-H) and anklewaist (N3-H) links in the performance evaluations. The measurements were carried out in an anechoic chamber using the vector network analyzer (VNA) Rohde & Schwarz ZVA8 with four test ports [16]. The UWB antennas used in the study were dipole and loop antennas, designed for on-off and on-on body communication links. More details about the antennas and their characteristics can be found from [17-18].
a)
1 right wrist - waist link left wrist - waist link ankle - waist link
0.8
The measurements were conducted in a frequency domain within the band 2-8 GHz with 1601 samples resulting in the channel transmission coefficients S21. In order to obtain statistical certainty, 100 consecutive sweeps were recorded for each link. Inverse fast Fourier transform (IFFT) was applied to convert the frequency domain channel responses into time domain impulse responses, which were then further used in the performance evaluations of the WBAN system. The impulse responses used in the BER simulations have been averaged over the 100 realizations for each of the links. Due to the energy concentration, only the beginning of the main peak with the following 31 samples appearing during the time period of 4 ns, are included. Phase of each path in the impulse response is assumed to be uniformly distributed. The averaged channel impulse responses for loop and dipole antennas with all the links are shown in Figs. 2a-b. It is noticed that with dipole antenna the main peaks of the impulse responses are weaker than those with the loop antenna, except for the left wrist waist link (N2 – H). The level of the second peak with respect to the level of the main peak is significantly higher with dipole antennas. The difference between the impulse responses of these antennas is due to the different radiation patterns, which
0.6 0.4 0.2 0
0
0.5
1
1.5
t [ns]
1
b)
right wrist - waist link left wrist - waist link ankle - waist link
0.8 0.6 0.4 0.2 0
0
0.5
1
1.5
t [ns]
Figure 2. Averaged impulse responses of loop (a) and dipole (b) antennas.
3
With the PPM demodulation, the decision on the mth received bit is based on the comparison between the decision variables from (2) and it is expressed as
first and second path, may be destructively combined at the receiver, resulting signal fading is more significant in the link (N2 – H). The reason for this is presumably the more unfavorable position of the left wrist of the test person during the measurements. For instance, the left wrist could have been aligned to the backwards more strongly than the right wrist. If the line-of-sight connection is lost between the antennas, performance decreases significantly. When using pure measurement data in the performance evaluations, it is important to remember these kinds of uncertainties which are always present in the measurement campaigns and which may have strong impact on the results. However, these results present well how the placement and the position of the antennas influences on the channel characteristics and hence on the BER performance. It should also important to keep in mind that in real situation, the channel is moving. During the measurements, the environment was static.
" "
w
w
.
(4)
" "
If the amount of energy in the first received time slot is greater than in the delayed received time slot, the received bit is ‘0’, otherwise it is ‘1’. IV.
RESULTS
In this section, the performance of WBAN system is evaluated in terms of uncoded bit-error-rate versus the energy per bit to noise power spectral density ratio (Eb/N0). The evaluation is shown separately for different links with OOK and PPM modulations, including loop and dipole antennas, which both are suitable for WBAN usage as shown in [17].
Finally, the BER performances obtained with loop and dipole antennas are compared in Fig. 5. For the clarity of the presentation, only the BER curves with PPM modulation are included. With the links (N1 – H) and (N3 – H), the dipole antenna clearly outperforms the loop antenna at higher Eb/N0 range: the difference is up to 6 dB. As discussed earlier, the link (N2 – H) is remarkably poor with the dipole antenna: the BER performance in this link with dipole antenna is now worse than with loop antenna, though the difference is only around 1 dB.
First, the BER performances for loop antenna case are showed in Fig. 3. The results are presented for the right wrist – waist link (N1 – H), the left wrist – waist link (N2 – H), and the ankle – waist link (N3 – H). The radio channel measurements were not carried out for the left ankle – waist link (N4 – H) and thus, that link is left out of the evaluation. As it can be seen from Fig. 3, BER performances for different links are very similar at lower Eb/N0 range for both OOK and PPM. The difference increases as the Eb/N0 ratio increases so that the best BER is obtained for the link (N1 – H) and the worst for the (N3 – H). The BER for the link (N2 – H) is somewhat in the middle. However, the differences between the BERs are minor: around 1 – 2.5 dB.
10
10
When comparing BER performances with OOK and PPM modulations, it is noted that PPM outperforms OOK by maximum 4.5 dB within the simulated Eb/N0 range. These results are coincident with the results presented in the previously published literature, which compares OOK and PPM modulations [12-14], since the PPM modulation is known to have improved power efficiency. The benefit of OOK modulation is the simplicity but as a drawback, with higher error probability. Furthermore, it is reasonable to take into account that OOK based receiver requires energy level estimation.
right wrist - waist link with OOK left wrist -waist with OOK ankle - waist link with OOK right wrist - waist link with PPM left wrist - waist link with PPM ankle - waist link with PPM
-1
-2
BER
10
0
10
10
-3
-4
5
10
15
20
25
30
35
Eb/N0 [dB]
Figure 3. BER performances with loop antenna, OOK and PPM.
Next, the performances are evaluated for dipole antenna. The BER results are shown in Fig. 4. As opposed to the results with loop antenna, the best BER performance is obtained in ankle – waist link (N3 – H). However, the difference for the BER performance in right wrist – waist link (N1 – H) is minor, only 1 dB at highest simulated Eb/N0 range. Instead, the BER performance in the left wrist – waist link (N2 – H) is conspicuously worse: roughly 6 dB at higher Eb/N0 range. In order to understand this phenomenon, the channel impulse responses from Fig. 2 for these links are studied in more details. It is noted that the main peak of the impulse response for the best links (N3 – H) and (N1 – H) is remarkable higher than that of the link (N2 – H). Instead, the level of the second peak respect to the level of the main peak is significantly higher in the link (N2 – H). As the signals propagated via the
10
10
right wrist - waist link with OOK left wrist - waist link with OOK ankle - waist link with OOK right wrist - waist link with PPM left wrist - waist link with PPM ankle - waist link with PPM
-1
-2
BER
10
0
10
10
-3
-4
5
10
15
20
25
30
Eb/N0 [dB]
Figure 4. BER performances with dipole antenna, OOK and PPM.
4
35
10
10
REFERENCES
0
Dipole, right wrist - waist link with PPM Dipole, left wrist - waist link with PPM Dipole, ankle - waist link with PPM Loop, right wrist - waist link with PPM Loop, left wrist - waist link with PPM Loop, ankle - waist link with PPM
-1
[1]
[2] -2
BER
10
[3] 10
-3
[4] 10
-4
5
10
15
20
25
30
[5]
Eb/N0 [dB]
Figure 5. Comparison of BERs obtained with loop and dipole antennas using PPM.
V.
[6]
SUMMARY AND CONCLUSIONS
This paper evaluated the performances of UWB energy detector receivers on the physical layer of the IEEE 802.15.6 standard. The studied wireless body area network was designed to monitor symptoms of Parkinson’s disease. Realistic channel measurement data was used in the evaluations for the determined links, whose positions corresponded to a realistic Parkinson’s disease measurement case. Radio channel data was based on the measurements carried out in an anechoic chamber in order to get pure link performances without the impact of the environment. OOK and PPM modulation methods were tested and compared with energy detector receivers using two different types of UWB antennas, loop and dipole, designed for on-body communication.
[7]
[8]
[9]
[10]
[11]
The results show that the PPM based system outperforms the OOK based system. However, the difference is at most 4.5 dB at higher simulated Eb/N0 range, whereas at lower Eb/N0 range the difference is clearly minor. The benefit of the OOK based system is its simplicity, although it requires energy threshold estimation. The placement, position and properties of the antennas influence the channel characteristics of the link and hence the system’s BER performance. The energy detector based receivers are suitable for the monitoring of the Parkinson’s disease in single-user environments such as at home, as there are no other users causing interference. In multiuser environments, more efficient and interferenceresistant modulation and reception techniques are required. The evaluation of these is left for future work.
[12]
[13] [14]
[15]
[16]
[17]
ACKNOWLEDGMENT This research was supported by Tekes, the Finnish Funding Agency for Technology and Innovation, through the European Regional Development Fund, as a part of ICMA project, as well as Infotech Oulu. Authors would like to thank Mr. Timo Kumpuniemi for providing the channel measurement data and Mr. Tommi Tuovinen for providing the UWB antennas.
[18]
[19]
5
J. Rousselot, J-D. Decotignie, “Wireless Communication Systems for Continuous Multiparameter Health Monitoring”, International Conference on Ultra-Wideband (ICUWB 2009), Canada, Sept. 2009, S. Ivanov, C. Foley, S. Balasubramanian, D. Botvich, “Virtual Groups for Patient WBAN Monitoring in Medical Environments”, IEEE Transaction on Biomedical Engineering, Vol. 59, No. 11, Nov. 2012. IEEE Standard for Local and Metropolitan Area Networks - Part 15.6: Wireless Body Area Networks. IEEE STD 802.15.6-2012, IEEE Computer Society 2012, NY, USA. K. Kwak, S. Ullah, N. Ullah, "An overview of IEEE 802.15. 6 standard," in 3rd International Symposium on Applied Sciences in Biomedical and Communication Technologies, 2010, pp. 1-6. D. Grosset, H. Fernandez and M. Okun, Parkinson's disease: clinician's desk reference. London: Manson Publishing, 2009, ch. 1. N. Keränen, M. Särestöniemi, J. Partala, M. Hämäläinen, J. Reponen, T. Seppänen, J. Iinatti, and T. Jämsä, “IEEE802.15.6 –based multiaccelerometer WBAN system for monitoring Parkinson’s disease,” accepted to 35th Annual International Conference of IEEE Enginerring in Medicine and Biology Society (EMBC), Japan, July 2013. M. Kangas, A. Konttila, P. Lindgren, I. Winblad and T. Jämsä, "Comparison of low-complexity fall detection algorithms for body attached accelerometers," Gait & Posture, vol. 28, pp. 285-291, 2008 W. Rhee, N. Xu, B. Zhou, Z. Wang, "Low power, non invasive UWB systems for WBAN and biomedical applications," 2010 International Conference on Information and Communication Technology Convergence (ICTC), , 2010, pp. 35-40. V. Niemelä, M. Hämäläinen, J. Iinatti, "IEEE 802.15. 4a UWB receivers in medical applications," International Journal of Ultra Wideband Communications and Systems, vol. 2, pp. 73-82, 2011. H. Viittala, B. Nahar, M. Hämäläinen, J. Iinatti, "Medical applications adapting ultra wideband: a system study," International Journal of Ultra Wideband Communications and Systems, vol. 1, pp. 237-247, 2010. M. Hernandez and R. Kohno, “UWB on-off waveform coded modulation for body area networks,” 5th International syposium on Medical Information and Communication Technology (ISMICT), 2011, Switzerland V. Niemelä, J. Haapola, M. Hämäläinen, J. Iinatti, ”Integration interval and threshold evaluation for an energy detector receiver with PPM and OOK modulations,” Workshop on ultra wideband for body area networking (UWBAN) in conjunction with 7th international conference on body area networks (2012), Oslo, Norway, Sept. 2012. I. Guvenc, H. Arslan, “On the modulation options for UWB systems”, Military Communication Conference, Oct 2003, Boston, USA. O. Abedi, J. Nielsen, “UWB Data Rate and Channel Capacity in Modulation Schemes”, Canadian Conference on Electrical and Computer Engineering, May 2006, Canada. IEEE Standard for Local and Metropolitan Area Networks – Part 15.4: Low-Rate Wireless Personal Area Networks (LR-WPANs). IEEE STD 802.15.4-2011, IEEE Computer Society 2011, NY, USA. Rohde & Schwarz (accessed in November 2012), [Online]. Available: http.//www2.rohde-schwarz.com/en/products/ test_and_measurement/network_analysis/ZVA8_%284-Port%29.html. T. Tuovinen, K. Yekeh Yazdandoost, J. Iinatti, ”Comparison of the performance of two different UWB antennas for the use in WBAN onbody communications”, 6th Europ. Conf. Antennas and Propagation (EuCAP), March 2012, Czech. T. Tuovinen, M. Berg, K. Yekeh Yazdandoost, E. Salonen, J. Iinatti, ”Impedance behaviour of planar UWB antennas in the vicinity of a dispersive tissue model” Loughborough Antennas and Propagation Conference (LAPC), Nov 2012, UK. T. Kumpuniemi, T. Tuovinen, M. Hämäläinen, K. Yekeh Yazdandoost, R. Vuohtoniemi, J. Iinatti, ”Measurement-based on-body path-loss modelling for UWB WBAN communications”, International Symposium on Medical Information and Communication Technology (ISMICT), Japan, March 2013.
