IEEE 802.15.6-based Prototype System for WBAN: Design and Implementation

1.5 MB·

by  Xiaonan Su

IEEE 802.15.6-based Prototype System for WBAN:
Design and Implementation
Xiaonan Su, Changle Li*, Xiaoming Yuan
State Key Laboratory of Integrated Services Networks, Xidian University, Xi’an, Shaanxi, 710071 China
*[email protected]
Abstract—Various current trends such as ever growing pop- environment primarily designed for communication networks
ulation and accelerated aging effects have effectively promoted [8].NetworkSimulatorVersion3(NS-3)[9]hasbeenapplied
7 the growth of Wireless Body Area Network (WBAN). Being the
to simulating systems in a highly controlled environment of
1 specialized standard of WBAN, the burgeoning IEEE 802.15.6
WBAN.However,therearesomediscrepanciesbetweensimu-
0 is on its way to refinement and perfection. Since experiments
2 are the foundation of research in WBAN that reproduce system lationsonsoftwareandactualresultsinrealisticenvironments
operating patterns, yet existing platforms in WBAN are mainly with time-variant, fading and susceptibility to disturbance
n
simulations that could not highly conform to reality with strict characteristics.Someproposeacooperativesoftware-hardware
a
requirements to some degree. Thus, a reliable and efficient
J approach for WBAN implementation [10] that decreases the
system based on the IEEE 802.15.6, named WBAN prototype
0 time and complexity of implementation. However, it is still a
system,isproposedandimplementedinthispaper.Theprototype
1 system is ascertained to be valid, authentic and efficient via kindofmakeshiftwhichwillbepaleinrealityandcorrectness
analysesofscenariotests.Runningincircumstancesmuchcloser before complete hardware platforms to some degree, since
]
I to reality, the system obtains experimental results that meet it is hardware that applications and protocols finally run on.
N rigid requirements preferably. Furthermore, based on the valid
Hardware platforms can propose models with comprehensive
platform we created, optimizations of two parameters in the
. evaluations and protocols superior in performance and practi-
s IEEE 802.15.6 by actual measurements are derived. We obtain
c an optimum number of transmission retries and relationship of cal in application [11]. Nevertheless, to our best knowledge
[ payload length within 30 with FER, making a trade-off between after serious investigation, few WBAN or IEEE 802.15.6

1 waste of resource and high packet loss rate. module has been constructed on hardware in public.
v
Due to this issue, this paper realizes the WBAN prototype
1
2 I. INTRODUCTION system based on IEEE 802.15.6, and makes research on its
4 actual performance. Specifically, contributions of this paper
Recent years have witnessed the rapid growth of global
2 are presented as follows:
0 population and aggravated aging effects [1], which results
. in urgent demand of comprehensive health care systems, yet
1
medicalresourcescannotalwayssatisfyourneeds.Inaddition, • We not only design and implement the WBAN modules
0
7 a long-term and comprehensive health care is usually so such as PHY module, MAC module, application module
1 costly that normal people cannot afford it. Furthermore, even and security module, but also assemble them via prim-
: in sub-health condition, people are eager for more freedom itives we defined to work coordinately as a prototype
v
i and personalized intelligent lifestyle with medical equipment. system based on IEEE 802.15.6.
X Thus, providing an efficient, convenient and low-cost health • Wecarryoutseveralsetsofscenariotestsunderdifferent
r caresystem,WirelessBodyAreaNetwork(WBAN)isgaining parameterstoverifythevalidityofourprototypesystem.
a
an increasing attention [2] [3]. • We analyze testing results to optimize two parameters
WBAN is a kind of body-centered short-range communi- in IEEE 802.15.6. We have derived the optimal maxi-
cation technology that connects and interacts with various mum transmission retries and payload length within 30,
sensors located inside, on the surface or outside of the human moreover, the relationship between payload and FER is
body,monitoringthehumanbodyconditionsandthesurround- excogitated. These not only provide a vital reference
ing environments [4] [5]. It is an emerging cross-technology to parameters in the standard IEEE 802.15.6, but also
which has a close relationship with Wireless Personal Area supplement to the standard in several specific scenes.
Network (WPAN), Wireless Sensor Network (WSN) [6] and
sensor technology [7]. The rest of the paper is organized as follows. Section II
Existing simulations of WBAN can be clarified into three briefly reviews the most relevant and basic contents in IEEE
categories: software, hardware, and their combination. Pop- 802.15.6. Section III elaborates the scheme design of WBAN
ular simulation software for networks includes Optimized prototypesystemanditsimplementation.SectionIVperforms
Performance Network Engineering Tool (OPNET), Objective a set of scenario tests, makes an evaluation of the prototype
Modular Network Test-bed in C++ (OMNET++) and Net- system, and studies the optimal transmission retries and the
work Simulator (NS). OPNET provides a skeleton of 14- variation of FER with payload. Finally, section V concludes
stageradiopipeline.OMNET++isadiscreteeventsimulation the paper.
II. ANOVERVIEWOFIEEE802.15.6
The The
The IEEE 802.15.6 standard [12] is especially designed for RF FPGA DSP sensor
Node board board
WBANbymodifyingthePHYandMACparameterssimilarto
the IEEE 802.15.4 standard [13] to support short range, ultra- The
The PHY layer security The MAC layer application
low power and reliable wireless communication in vicinity or layer
inside of living tissue [14]. A WBAN consists of a sole hub
and several nodes whose number ranges from 0 to 64. The
The The
nodes are used to collect some parameters of the body, while RF FPGA DSP Bluetooth
Hub board board
thehubexecutesspecialactionsaccordingtothedatareceived
The
or interactions with the user [15]. The PHY layer security The MAC layer appllaiycaetrion
AccordingtothereferencemodelofIEEE802.15.6,thehub
Fig. 2: Hardware structure of a node and a hub
and nodes are divided into two layers: PHY layer and MAC
layer, as depicted in Fig. 1. The service access point (SAP) is
an interface between two layers. MAC layers provide service
management service module, data service module and data
for application layers through MAC SAPs, while PHY layers
transmissionmodule,allofwhicharemainlyachievedonDSP.
provide service for MAC layer through PHY SAPs. When
Polling procedure is run as follows. After being charged, the
transmitting data, service data units exchange between PHY
system enters initialization part. Then, it enters the polling
layer, MAC layer and application layer through respective
section that senses whether data from upper layers arrives. If
SAPs.
it does arrive, different interfaces arrival events will trigger
corresponding disposal mechanisms, otherwise, states and
state transition events in the MAC layer will be polled and
disposed respectively. Interface modules send primitives to
management service module, data service module and data
transmission module.
Fig. 1: Reference model in IEEE 802.15.6 B. Implementation
Since the MAC layer locates in the functional centre of
the system architecture and interacts with every other part
III. SCHEMEDESIGNANDIMPLEMENTATION in the system, we give it more weight. It is developed on
Digital Signal Processor (DSP) and can be divided into three
This section focuses on the ground-breaking exploration
parts: scheme design, software realization and hardware im-
of a WBAN prototype system. System architecture and full
plementation.TheaccomplishedhubisshowninFig.3,which
implementation are to be specified next.
has much similarity with nodes. The platform provides ample
interfaces to access various sensors, such as body temperature
A. System Architecture transducer and heart rate sensor that have already been used.
The whole WBAN prototype system is complicated in both
structure and function, whose content includes realizations of
the application layer, MAC layer, physical (PHY) layer and RF Board(PHY
Layer)
security certification. JTAG
MAC Layer
Wedesignandconstructthehubandnodesaccordingtothe Antenna
communication mechanism of WBAN. A node consists of a
Application
sensor board, DSP, Field Programmable Gate Array (FPGA), Layer
a Radio Frequency (RF) board and an antenna, while a hub
is composed of a Bluetooth board, DSP, FPGA, a RF board
Bluetooth
Board
and an antenna, as illustrated in Fig. 2. Both of them contain
the application layer, MAC layer, physical layer as well as
security certification section.
DSP accomplishes a series of data control and buffering
between MAC and PHY layer through FPGA. In addition,
by cooperation with FPGA, hardware system with complex Fig. 3: Hub
interactions could be achieved. The RF board with antenna is
in charge of communication between the hub and nodes. The Allmodulesorlayersinourprototypesystemcommunicate
Bluetooth board in hub takes charge of transmitting the data with each other using primitives. Therefore, we set about
collected to corresponding terminals via Bluetooth technique. discussing primitives and module implementation.
The implementation of WBAN prototype system is op- 1) Primitives: In order to transmit commands quickly and
erated by polling program that consists of three modules: efficiently between modules and fill in the blank of explicit
primitivesspecificationsinIEEE802.15.6,wedesignawhole its maximum transmission retries is (m−1). The possibility
set of interface primitives. The primitives are classified into of a successful transmission with (m−1) transmission retries
threeparts:managementprimitives,dataserviceprimitivesand can be calculated by the following equations.
datatransferprimitives,correspondingtothethreemodulesin
polling program which are to be discussed next.
P =C∗F (3)
2) Management Module: The management module takes suc suc
charge of the connection establishment for networking, such Fsuc =(1−Pfer)2 (4)
as initialization, connection building and breaking. After re- m−1
(cid:88)
ceiving and processing management primitives, management C = Cmi −1[Pfer∗(1−Pfer)]m−i−1∗Pfier (5)
servicemoduleconstructsvarioustypesofmanagementframes i=0
andsendsthemtothedatatransmissionmodule.Meanwhile,it F represents the possibility of successful transmission of
suc
handles multiple types of management frames and constructs data frames and Ack frames, C means the combinational
different management primitives sent to the application layer. number of successful transmission, and P signifies FER.
fer
3) Data Service Module: Data service module is in charge The relationship diagram is present in Fig. 4. Enlarging and
of processing local nodes data such as data fragmentation, having a close observation of the figure, we can derive at
reconstitution, framing, etc. It constitutes data frames by least two conclusions. First, PER reaches its minimum value
receiving and disposing data service primitives. In addition, when m ranges from 4 to 30. In addition, the possibility of
it receives and processes data frames, constructs the data successful transmission decreases with m. Thus, we can draw
service primitives, and sends them to finish the transmission the conclusion that 4 is the appropriate value of m for a
of business data. minimumPER,namely,3isthepropermaximumtransmission
4) Data Transmission Module: Data transmission module retries. This conclusion will be certificated by our tests in the
is responsible for sending and receiving data frames, which following section.
covers encryption and decryption, CRC check, address judg-
ment, access control, etc. It writes the CRC check in and
1 .0
encryptsthedataframes,thensendsthemtothephysicallayer.
Meanwhile,itreceivesandprocessesdata,anduploadsitafter
0 .8
data transmission.
0 .6
C. Theoretical Analysis
c
FER and PER are selected to examine the performance of usP 0 .4
the WBAN prototype system [16]. The FER is the ratio of
error frames in transmission, and PER is the ratio of error 0 .2
packets in the transmission of sending packets which includes
packetlossrate.HighFERandPERindicateweakadaptability 0 .0 1 .0
0 .8
of system. If so, we should improve the system by adjusting 1 0
0 .6
relevant parameters. 2 0 0 .4
m 3 0 p
The FER can be calculated by formula (1), where Sfrm 4 0 0 .2
and Rfrm represent the numbers of frames sent in nodes 5 0 0 .0
and frames received in the hub respectively. The PER can
Fig. 4: Relationships between successful transmission possi-
be figured out by formula (2), where S and R signify
pkt pkt bility, maximum transmission retries and PER
the numbers of packets sent and received respectively.
2) Payload: Payload,asasignificantelementindatatrans-
P =(S −R )/S (1) mission, is another parameter that we focus on. If it is set
FER frm frm frm
too low, the value of payload will lead to inefficiency and
P =(S −R )/S (2)
PER pkt pkt pkt
time-consumption[18].Whilesettingtoohighresultsinhigher
1) Transmission retries: Transmission retries is quite an packet loss rate, since larger packets keep the channel in busy
important factor in wireless transmission. Setting it too high condition longer [19]. Thus, how to select a appropriate size
leads to much waste time and energy consumption, while too of payload in WBAN becomes a non-ignorable problem.
low may result in communication outage [17]. However, the In our prototype, there are eight bytes apart from payload
IEEE802.15.6standardhasnotspecifiedhowtosetthetrans- in a data frame, and the length of an Ack frame is nine bytes.
missionretries.Therefore,wehavestudiedtherelationshipbe- Since the length of a data frame is comparable to that of an
tween PER and transmission retries. On account of the length Ack frame, the length of Ack frames cannot be neglected.
difference between data frames and acknowledgement (Ack) Suppose that P , L and j represent bit error rate (BER),
ber ack
frames, we cannot neglect error Ack frames in performance lengthofAckframesandtransmissiontimesrespectively,and
investigation.Innaturalrandomizedenvironments,weassume P is a fixed value. Evidently, L obeys the geometric
ber ack
that the PERs of data frames and Ack frames are the same. If distribution and we can work the Eq. (6) to (8) out. The
a data frame is sent successfully over m times transmission, variation of payloadwith P and FER is illustrated in Fig.
ber
5. We can learn from the diagram that FER gets larger with TABLEI:ParametersinTestingScenario
the increase of payload on a given P . Then we exploit
ber Parameters Value
interior point algorithm to search an optimal evaluation of
FrequencyBand 2.4GHz
FER, as shown in the Eq. (9) to (13). Corresponding to the
AverageTimeofOneTest 40min
Fig.5,whenpayloadvariesfrom0to30,thepayloaddeclines
DataTransmissionRate 121.4kbps
with the decrease of FER. This verdict is also certificated in
TestingDistances 1,2,5,10m
later tests.
Transmissionretries 0,1,2,3,4
PayloadLength 5,10,15,20,25,30bytes
L =8∗9∗j∗Pj−1(1−P ), j =1,2,3,4,5 (6)
Ack ber ber
L =8∗(payload+8) (7)
data scenarios are confined to within 10m and two-hop at most,
FER=1−(1−Pber)Ldata+LAck (8) among which seven nodes using single-hop within 5m just as
ourtestingenvironmenthascertainrepresentativesignificance.
The nodes collect data from sensors, and the hub receives
the data sent from the nodes. We can observe variables via
the Code Composer Studio (CCS) software, an integrated
1
developmentenvironmentforDSPembeddedsoftwaredesign.
0.8
We employ distance as one variable, because degradation
0.6 in network performance caused by the coexisting of BANs
R
E
F0.4 using same frequency band can be alleviated by distance. We
confine the distance between node and hub to 5 meters, and
0.2
designatethenumberoftransmissionretriesfrom0to4.Then
0
0.06 wecarryoutaseriesoftestsalteringtransmissiondistanceand
0.04 30 transmission retries respectively.
20
0.02
10
BER 0 0 Payload B. Testing Results and Evaluation
Fig. 5: Variations of FER with payload and BER The analysis of system testing results include three steps.
Firstly, process the experimental data into parameters such as
FER and PER. Secondly, checkout whether the environment
(cid:26) min FER=1−(1−P )Ldata+LAck
ber (9) is configured properly. Finally, examine the recorded results.
s.t. payload(cid:62)0
The variation of FER with transmission distance is illus-
tratedinFig.6.TheFERstayslowwithinthedistanceof4m;
1 FERisabout0.3%inwiredenvironment(connectinghubwith
B(x)= (10)
payload nodes through wires, which is expected to be error-free as a
ε
I(payload;εk)=(1−Pber)Ldata+LAck + paylkoad (11) rweefecreanncken)oawndthabatouthte0.t5ra%nsimniwssiiroenlespserefnovrimroannmceenist.fiTnheerwefitohrien,
ε
I(cid:48)(payload)=−8ln(1−P )(1−P )payload− k 4m and ordinary over 5m. It is obvious and reasonable that
ber ber payload2 FER grows with distance. Since WBAN is a kind of short
(12)
distance wireless networks, the results meet the requirements
ε =−8payload2∗[ln(1−P )∗(1−P )payload] (13) of the WBAN prototype system.
k ber ber
Inordertovalidatetheconclusionthattheoptimumnumber
IV. SCENARIOTESTSANDPERFORMANCEEVALUATION oftransmissionretriesis3,weconductaseriesofexperiments
on our prototype system. We set five different numbers of
In order to verify the validity and effectiveness of our sys-
transmission retries over four distances to observe variations
tem,scenariotestsareexecuted.Inthissection,wefirstdesign
of PER. The PER changes over transmission distances and
the test scheme, then record experimental data. Influences of
transmissionretries,asillustratedinFig.7,inwhichonepoint
parameters are considered, such as the effect of distance on
represents one experiment. We can learn from the diagram
FER and the impact of retransmission on PER. Lastly, the
thatthePERdecreaseswiththemaximumtransmissionretries
performance of the WBAN prototype system is analyzed and
increase.Whenmaximumtransmissionretriesishigherthan3,
some conclusions are drawn.
PERs of different distances are zero, which indicates no error
packet. Therefore, 3 is the optimum maximum transmission
A. Tests Design and Execution
retries for our prototype system, which verifies the theoretical
In the testing scheme, module testing is carried out first, analysis above.
thensystemtestingfollows.Weemploysixnodesandonehub To verify the conclusion that payload declines with the
to consist a typical communication system with star topology, decrease of FER when payload varies from 0 to 30, we
trying our best to match with realistic applications owing to conduct a series of tests in which FER varies with payload
the constraints of hardware and floor space. Since WBAN size and different distances commonly used in WBAN. As
0.040
1 m
0.035 2 m
5 m
5.0 0.030 1 0 m
4.5
4.0 0.025
233..5.05 %) FER0.020
2.0 R ( 0.015
11..05 FE 0.010
0.5
wireless transwmiriesleswisirosenl1wie 0rdsmeislsw5eitrsamesnlw4ecimsresel3weimsrsel2emss1mwired 0.0 Fpaigy.lo8a:d Vsiaz0.re0i0a5 tion0 s of5 FERP1a0ywlo aidt hL1e5n gttrha / nb y2st0emissi2o5 n di30stance and
number of transmission retries is 3, which ensures both FER
Fig. 6: Variations of FER with transmission distance and PER to be comparatively low. Furthermore, we worked
outandvalidatedtheverdictthatFERgetslargerwithpayload
within 30.
8 1 m
2 m
5 m
1 0 m REFERENCES
6
[1] N. Javaid, N. Khan, M. Shakir, M. Khan, S. H. Bouk, and Z. Khan,
R (%)4 “pUrebpirqiunittoaurXsihv:e1a3lt0h3c.a2r0e62in,2w0i1re3l.ess body area networks-a survey,” arXiv
E
P [2] S. Talha, R. Ahmad, and A. K. Kiani, “Priority based energy aware
(pea)routing protocol for wbans,” in Vehicular Technology Conference
2
(VTCFall),2015IEEE82nd. IEEE,2015,pp.1–5.
[3] S.Movassaghi,M.Abolhasan,J.Lipman,D.Smith,andA.Jamalipour,
“Wireless body area networks: A survey,” Communications Surveys &
0
Tutorials,IEEE,vol.16,no.3,pp.1658–1686,2014.
0 1 2 3 4
[4] N.Gu,Y.Jiang,J.Zhang,andH.-T.Zheng,“Animplementationofwban
M a x R e tra n sm issio n T im e s
module based on ns-2,” in 2013 International Conference on Computer
Fig. 7: Variations of PER with transmission distance and SciencesandApplications(CSA). IEEE,2013,pp.114–118.
transmission retries [5] Y.Du,F.Hu,W.Xiong,andX.Liu,“Beam-patternsynthesisforcircular
arrays with sensor selection for wban via convex optimization,” in 2016
IEEE83rdVehicularTechnologyConference(VTCSpring). IEEE,2016,
pp.1–6.
illustratedinFig.8,thelargernumberofretransmissionretries [6] P.Rajeshwari,B.Shanthini,andM.Prince,“Hierarchicalenergyefficient
clusteringalgorithmforwsn,”Middle-EastJournalofScientificResearch,
is, the lower PER will be. FER is comparatively low in four
vol.23,pp.108–117,2015.
different distances when payload length is around 10. FER [7] P.Rawat,K.D.Singh,H.Chaouchi,andJ.M.Bonnin,“Wirelesssensor
increases with payload when other conditions are the same, networks: a survey on recent developments and potential synergies,” The
Journalofsupercomputing,vol.68,no.1,pp.1–48,2014.
namely, the accuracy of information will be less ideal with
[8] F. Bause, P. Buchholz, J. Kriege, and S. Vastag, “A simulation envi-
thepayloadincreaseinthecontentfrom0to30.Thisdiagram ronment for hierarchical process chains based on omnet++,” Simulation,
also coincides with Fig. 5. Owing to all unpredictable adverse vol.86,no.5-6,pp.291–309,2010.
[9] S.Mastorakis,A.Afanasyev,I.Moiseenko,andL.Zhang,“ndnsim2.0:A
factors in actual scenes, system performance in Fig. 8 is
newversionofthendnsimulatorforns-3,”TechnicalReportNDN-0028,
less preferable but more realistic. This is another aspect in NDN,Tech.Rep.,2015.
optimizing the IEEE 802.15.6, and it will probably do much [10] M. Chen, Z. Li, and G. Zhang, “A cooperative software-hardware
approachforwirelessbodyareanetworkimplementation,”in2014IEEE
better if given a fuller development.
4thAnnualInternationalConferenceonCyberTechnologyinAutomation,
Control,andIntelligentSystems(CYBER). IEEE,2014,pp.214–218.
[11] X.Ge,S.Tu,T.Han,Q.Li,andG.Mao,“Energyefficiencyofsmallcell
V. CONCLUSIONS
backhaulnetworksbasedongauss–markovmobilemodels,”IETNetworks,
In order to provide a more realistic and effective WBAN vol.4,no.2,pp.158–167,2015.
[12] T. Chang, T. Watteyne, K. Pister, and Q. Wang, “Adaptive synchro-
platformaswellascontributetosupplementingandoptimizing
nization in multi-hop tsch networks,” Computer Networks, vol. 76, pp.
the IEEE 802.15.6, we designed a WBAN prototype system 165–176,2015.
and constructed the whole system from module to infrastruc- [13] M. Hamalainen, T. Paso, L. Mucchi, M. Girod-Genet, J. Farserotu,
H. Tanaka, W. H. Chin, and L. N. Ismail, “Etsi tc smartban: Overview
ture. Then we created a set of interface service primitives for
ofthewirelessbodyareanetworkstandard,”inMedicalInformationand
interaction between layers. In addition, we made analyses of CommunicationTechnology(ISMICT),20159thInternationalSymposium
its performance on the basis of scenario tests, and drew the on. IEEE,2015,pp.1–5.
[14] P.Khan,N.Ullah,M.N.Alam,andK.S.Kwak,“Performanceanalysis
conclusionthatthesystemwedesignedcanmeetthedemands
of wban mac protocol under different access periods,” International
of WBAN applications basically, and the optimal maximum JournalofDistributedSensorNetworks,vol.501,p.102052,2015.
[15] A.A.Kannan,B.Fidan,andG.Mao,“Robustdistributedsensornetwork
localizationbasedonanalysisofflipambiguities,”inIEEEGLOBECOM
2008-2008 IEEE Global Telecommunications Conference. IEEE, 2008,
pp.1–6.
[16] X.Ta,G.Mao,andB.D.Anderson,“Onthegiantcomponentofwireless
multihopnetworksinthepresenceofshadowing,”IEEETransactionson
VehicularTechnology,vol.58,no.9,pp.5152–5163,2009.
[17] G.Mao,B.D.Anderson,andB.Fidan,“Wsn06-4:Onlinecalibrationof
pathlossexponentinwirelesssensornetworks,”inIEEEGlobecom2006.
IEEE,2006,pp.1–6.
[18] G.MaoandB.D.Anderson,“Graphtheoreticmodelsandtoolsforthe
analysisofdynamicwirelessmultihopnetworks,”in2009IEEEWireless
CommunicationsandNetworkingConference. IEEE,2009,pp.1–6.
[19] R.Cavallari,F.Martelli,R.Rosini,C.Buratti,andR.Verdone,“Asurvey
on wireless body area networks: technologies and design challenges,”
Communications Surveys & Tutorials, IEEE, vol. 16, no. 3, pp. 1635–
1657,2014.