Performance Analysis of IEEE 802.15.6-Based Coexisting ... · Carrier Sense Multiple Access with...

1536-1276 (c) 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TWC.2018.2848223, IEEETransactions on Wireless Communications

1

Performance Analysis of IEEE 802.15.6-BasedCoexisting Mobile WBANs with Prioritized

Traffic and Dynamic InterferenceXiaoming Yuan, Student Member, IEEE, Changle Li, Senior Member, IEEE, Qiang Ye, Member, IEEE,

Kuan Zhang, Member, IEEE, Nan Cheng, Member, IEEE, Ning Zhang, Member, IEEE, Xuemin(Sherman) Shen, Fellow, IEEE

Abstract—Intelligent Wireless Body Area Networks (W-BANs) have entered into an incredible explosive populariza-tion stage. WBAN technologies facilitate real-time and reliablehealth monitoring in e-healthcare and creative applications inother fields. However, due to the limited space and medicalresources, deeply deployed WBANs are suffering severe in-terference problems. The interference affects the reliabilityand timeliness of data transmissions, and the impacts ofinterference become more serious in mobile WBANs becauseof the uncertainty of human movement. In the paper, weanalyze the dynamic interference taking human mobility intoconsideration. The dynamic interference is investigated indifferent situations for WBANs coexistence. To guarantee theperformance of different traffic types, a health critical indexis proposed to ensure the transmission privilege of emergencydata for intra- and inter-WBANs. Furthermore, the perfor-mance of the target WBAN, i.e., normalized throughput andaverage access dealy, under different interference intensity areevaluated using a developed three-dimensional Markov chainmodel. Extensive numerical results show that the interferencegenerated by mobile neighbour WBANs results in 70%throughput decrease for general medical data and doubles thepacket delay experienced by the target WBAN for emergencydata compared with single WBAN. The evaluation resultsgreatly benefit the network design and management as wellas the interference mitigation protocols design.

Index Terms—Wireless body area network (WBAN), inter-ference, CSMA/CA, IEEE 802.15.6, Markov chain model.

I. INTRODUCTION

A Wireless Body Area Network (WBAN) is a human-centered, highly reliable short-range wireless communica-tion network [1]. With flexibility, scalability and low-costcharacteristics, WBANs have been widely used to providereal-time and continuous care for health monitoring in e-healthcare [2]. The application of WBAN alleviates the

Corresponding author: Changle Li. X. Yuan and C. Li are with theState Key Laboratory of Integrated Services Networks, Xidian Univer-sity, Xi’an, Shaanxi 710071, China (e-mail: [email protected],[email protected]).

Q. Ye, N. Cheng and X. Shen are with the Department of Electricaland Computer Engineering, University of Waterloo, Waterloo, ON N2L3G1, Canada (e-mail: q6ye, n5cheng, [email protected]).

K. Zhang is with the Department of Electrical and Computer Engi-neering, University of Nebraska-Lincoln, Omaha, NE 68182, USA(email:[email protected]).

N. Zhang is with the Department of Computing Sciences, TexasA&M University Corpus Christi, Corpus Christi, TX 78412, USA(email:[email protected]).

conflicts between limited healthcare resources and increas-ing needs of aging population. WBAN technology hasbeen envisioned as one of the primary technologies for theubiquitous Internet-of-Things (IoT) to enhance the qualityof people’s life [3]–[5].

WBANs are widely deployed in densely populated sce-narios, for example, wards and waiting rooms in hospitals,resident-centered nursing homes, and even public transport-s. The communication zones of adjacent WBANs are closeor even overlapped. Coexisting densely deployed WBANsinevitably suffer severe intra- and inter- WBAN interfer-ence. Intra-WBAN interference occurs when nodes in aWBAN transmit data at the same time. Inter-WBAN inter-ference is caused by simultaneous data transmission amongtwo or more adjacent WBANs. Inter-WBAN interferencedecreases the reliability and timeliness of physiologicaldata transmission, increasing the network management andeconomic cost paid on healthcare. Particularly, interferencemay lead to catastrophic incomplete or overdue data inemergency medical diagnosis and threaten the life safetyof people. Therefore, interference problem of coexistingmulti-WBANs needs in-depth study.

Many researchers have paid attention to the interferenceof coexisting WBANs. The inter-user interference amongadjacent WBANs is investigated in [6]. An interferencedistribution model [7] is proposed by using a geomet-rical probability approach. The inter-WBAN interferenceis related to the activities of users carrying them andis different from other types of wireless networks [8].The traditional interference mitigation approaches are notsuitable for WBANs. The inter-WBAN coexistence andinterference mitigation schemes are surveyed and discussedin [9]. A Bayesian game based power control scheme[10] is introduced to mitigate the impact of inter-WBANinterference. A beacon interval shifting scheme [11] and abeacon scheduling scheme [12] are developed to executecarrier sensing before beacon transmission to reduce inter-WBAN interference.

However, in the study of interference analysis and in-terference mitigation, there are few work considering thedynamic interference of mobile neighbor WBANs. Mobilityof WBANs affects the service-oriented as well as the



2

application-oriented network performance [13]. Interfer-ence intensity varies with human movements. Interferencelasts for a short period when neighbor WBAN just passesby or lasts long when there is no relative movement be-tween inter-WBANs. Network performance, i.e., through-put and delay, fluctuate with interference variation and be-come severe when adjacent WBANs are deeply overlappedwith each other [14]. Therefore, analyzing the dynamicinterference in coexisting multi-WBANs is required, whichsatisfies the practical scenario well and can greatly benefitnetwork design and management.

On the other hand, different types of traffic in the W-BAN have different Quality-of-Service (QoS) requirements.For example, life-critical emergency electroencephalogram(EEG) and electrocardiogram (ECG) data require highreliability and low transmission delay while regular temper-ature data are delay-tolerant. It is necessary to differentiatethe traffic priorities to satisfy the diversified performancerequirements. In addition, some monitoring physiologicaldata in medical services exceed the normal range andtrigger other abnormal changes occasionally. The variationmessage and the abnormal data have higher priority thanusual, and they should be transmitted to the hub or datacenter timely for further processing. Both of the generaldata and the abnormal data are required to be consideredin multi-WBANs coexistence.

In this paper, we analyze the interference of coexist-ing WBANs by considering human mobility and trafficdifferentiation. The network topology changes when hu-man moves. Different nodes with different traffic prior-ities in the network have diverse performance require-ments. In this case, the asynchronous IEEE 802.15.6-basedCarrier Sense Multiple Access with Collision Avoidance(CSMA/CA) protocol [15] is employed rather than thescheduled Time Division Multiple Address (TDMA). TheCSMA/CA scheme is adaptive to the network topology andnetwork load changes, and it does not require complex timesynchronization compared to TDMA method. We developthe CSMA/CA Medium Access Control (MAC) mechanismin this paper to improve the flexibility and scalability of aWBAN. We investigate the MAC performance with inter-ference dynamics from neighboring WBANs and guaranteethe required QoS for different types of traffic. Our maincontributions is summarized as follows.

• Firstly, we analyze dynamic inter-WBAN interferencewith the movement of neighbor WBANs in rectilinearmotion and curvilinear motion cases. We also investi-gate the interference dynamics in different situationsfor a multi-WBANs coexistence scenario (e.g. differ-ent distances between inter-WBANs, varying numberof nodes of each WBAN, different number of coex-isting WBANs, etc.). Furthermore, we consider theinterference dynamics in developing the performanceanalytical model, which facilitates the practicabilityand credibility of the model in WBANs.

• Secondly, we propose a health critical index to guar-

antee the privilege of time sensitive emergency andburst medical data traffic. The definition of the healthcritical index takes into account service types, trafficdifferentiation, and data severity. The burst abnormaldata enjoy higher health critical index value to ensurethe channel access privilege than other normal data.In addition, an algorithm is designed to manage theservice order of all the nodes in multi-WBANs coex-istence scenario.

• Finally, we develop a three-dimensional Markov chainperformance analytical model to investigate the impactof interference on throughput and packet access delayof prioritized traffic in the target WBAN. Extensivenumerical results demonstrate the accuracy and ef-fectiveness of our proposed performance analyticalframework.

The remainder of paper is organized as follows. SectionII summarizes the related work. The network model andtraffic model is presented in Section III. In Section IV,the changing interference and the proposed severity indexare described. Performance on normalized throughput andaverage packet access delay are analyzed in Section V.Section VI demonstrates analytical results and discussions.Finally, concluding remarks and future work outlines aregiven in Section VII.

II. RELATED WORK

Multiple WBANs coexist in one place in order to accessto healthcare services for different patients. The interfer-ence from neighbor WBANs is inevitable in dense area,degrading network performance and increasing networkmanagement cost [16]. Interference in multiple WBANsinclude the mutual or inter-user interference incurred bysimultaneous transmissions in the vicinity of multiple W-BANs, and cross-technology interference caused by theutilization of different transmission technologies operatingin the same spectrum range [17]. More and more re-searchers have paid their attention on interference analysisand mitigation in coexisting WBANs.

The coexistence and interference management issueswere summarized [18], exploring a wide variety of commu-nication standards and methods deployed in WBAN. Sun etal. [19] presented a stochastic geometry analysis model toanalyze the inter-user interference in IEEE 802.15.6 basedWBANs and derived the optimal interference detectionrange to tradeoff between outage probability and spatialthroughput. Jameel et al. [20] analyzed the impact of co-channel interference on WBANs under generalized fading.The progress of interference generation between two deeplyoverlapped WBANs [21] was introduced. Three heuristicsolutions [22] were presented to solve the problems ofmutual interference and cross-technology interference. Thenodes of intra-WBAN employed the IEEE 802.15.4 proto-col [23] with not distinguishing the traffic priorities. Zhanget al. [24] described the necessity of classifying differenttypes of health data for a WBAN.



3

Moreover, different analytical models and proposed al-gorithms are used in the MAC performance analysis for W-BANs. In the presence of inter-user interference, a stochas-tic geometry model [25] were proposed to analyze theeffects of IEEE 802.15.6 MAC. The outage probability andspatial throughput of WBANs were derived on the spatialdistribution of the interfering WBANs. An analytical modelfor the IEEE 802.15.4 MAC protocol [26] was proposedto explore the time-varying feature of the MAC statisticsunder periodic traffic. Based on Bianchi’s work [27], manyauthors contributed to the Markov analytical models forWBANs [28]–[31]. A criticality index was introduced [28]to identify the critical physiological parameter, using aMarkov chain-based analytical approach to maximize thereliability of the critical node. Nevertheless, the differentUser Priorities (UPs) of various traffic in a WBAN shouldbe considered. A four dimensional Markov chain [30]was used to evaluate the power consumption, throughputand average delay concerned the time spent by a nodeawaiting the acknowledgement frame. The IEEE 802.15.6CSMA/CA mechanism with different UPs was analyzedemploying a three-dimensional Markov chain model [31].The proposed scheduling techniques [32] were evaluatedwithin various traffic rates and time slot lengths to improveWBAN reliability and energy efficiency. However, theseworks are more comprehensive if they considered the trafficcritical conditions out of safety range and the influence ofinterference from adjacent WBANs.

Several approaches were proposed to mitigate the intra-and inter- WBAN interference [33]–[39]. A superframeoverlapping scheduling method [33] based on beacon shift-ing and superframe interleaving interference mitigationtechniques was developed for coexisting WBANs. A two-layer interference mitigation MAC protocol (2L-MAC)was proposed based on IEEE 802.15.6 for WBANs [34].However, the scheme evaluated the long-term interferencein non-adaptive and scheduled one node for transmissionregardless of the interference level. Cheng et al. [36]employed TDMA method to reduce mutual interferenceof nearby WBANs. A horse racing scheduling scheme[37] was raised to decrease single-WBAN interference. Butthe resource allocation scheme might degrade once thetopology of a WBAN changes with human movements.Therefore, the random adaptive CSMA/CA method is moresuitable in the analysis of interference dynamics withhuman movements and diversity traffic types. The authors[39] proposed an adaptive CSMA/CA MAC protocol byadjusting the length of each polling period/MAC framebased on its perceived interference level to mitigate theinter-WBAN interference.

III. SYSTEM MODEL

In this section, we present the network model as wellas the traffic model. Some important notations used in thepaper are summarized in Table I.

A. Network Model

WBAN 1 WBAN 2

WBAN 4

WBAN 3

Interference

Interference

InterferenceHub’s interference area

On-body communication

HubNode Data center

Free space wireless channel

¡

1hN

2hN

3hN4hN

1

1s

2

1s

3

1s

4

1s

5

1s

22s

5

2s

3

2s

1

2s5

3s

23s

43s

44s

5

4s

Fig. 1: Network model.

The WBAN usually consists of a centralized entityhub and a number of sensor nodes to monitor the vitalsigns of human body. The Hub is usually performed by asmartphone or a Personal Digital Assistant (PDA), whichhas greater computing capability with power-rechargeable.Sensor nodes collect EEG, ECG, motion signals, and gen-eral physiological data. There is a set of coexisting WBANs{Nm|m = 1, 2, . . . ,M} in the system, as shown as in Fig.1, where M is the number of WBANs and Nm stands forWBAN m. The hub is denoted by {Nmh|m = 1, 2, . . . ,M}and several numbered physiological sensor nodes are de-noted by {sym|y = 1, 2, . . . , Y }.

There are four communications types in WBANs, namelyin-body, on-body, off-body and body-to-body [40]. Usually,the collected data traffic is transmitted from nodes to hubthrough on-body channel and from hub to the data centerthrough free space wireless channel. The body-to-bodytype represents communications from one person’s bodyto another person’s body. In general, the distance betweenWBANs is large enough such that there is no inter-WBANinterference, just as the non-intrusive distribution of N3

and N4. With the movement of human body, the distancebetween WBANs is constantly changing. As the inter-WBANs distance decreases, the communication zones areoverlapped partially. The interference occurs between theadjacent WBANs if there are senor nodes in the communi-cation area of neighbor WBANs, as the distribution of N1,N2 and N3 in Fig. 1. The data transmissions of sensor nodes41 in the overlapped area of N1 and N2 intrude all the datatransactions of N2, since s41 is in the interference range ofhub N2h. s53 in N3 also interferes data transactions in N2,increasing the data collision probability. If the inter-WBANdistance becomes smaller, the overlapped area as well asthe number of interference nodes increases. The worst caseis that two or more communication zones are completely



4

TABLE I: Basic notations

Parameter Description Parameter DescriptionΥ Interference range of hub Hy

m Health critical index value of node y in Nm

Nm WBAN m Pidle Probability of the channel being idledm The distance between Nm and target WBAN Rs Symbol rateUPk User (or traffic) priority for a node Pb Probability of packet collisionl Failure times before the successful transmission Pu Probability of backoff counter being unlockedτk Medium access probability of UPk node Tc−slot CSMA slot lengthY The number of nodes in a WBAN Tpk Packet transmission timenk The number of nodes of UPk Ntotal Total bits flowing in the physical layernI The number of interfering nodes BTmax The maximum backoff stage

CWmin[UPk]Minimum contention window (CW) value of UPk Wk,j CW value of UPk at jth backoff stage

CWmax[UPk] Maximum CW value of UPk TkCW Mean backoff time

PUPk−inlockProbability that the backoff counter is locked due

to insufficient remaining time for a transaction PkidleProbability of channel being idle during

the backoff period of UPk node

overlapping. The interference seriously aggravates the net-work performance.

B. Traffic Model

TABLE II: UP mapping and CW bounds

Userpriorities Traffic designation CWmin CWmax

7 Emergency or medicalimplanted event report 1 4

6 High priority medical dataor network control 2 8

5 Medical data or network control 4 84 Voice data or management 4 163 Video data 8 162 Excellent effort data 8 321 Best effort data 16 320 Background data 16 64

Sensor nodes in each WBAN collect and transmit real-time sensing data to the hub. Different sensing data typeshave different reliability and delay requirements, especiallythe life-critical emergency data in medical applicationswhich require high reliability and minimum transmissiondelay. The sensing health data are classified into differentdata or user priorities to satisfy the diversified performancerequirements. In IEEE 802.15.6, eight User Priorities (UPs)are predefined in Table II. The higher user priority hasprivilege to access the channel. The user priorities ofmedical data traffic are from 5 to 7, guaranteeing the severertraffic in high priority to be transmitted timely. We considereach node in the WBAN only collects one kind of userpriority traffic and the WBAN is saturated that every nodein the network always has packets to contend to accessthe channel. Data with higher user priority enjoy smallContention Window (CW) size to guarantee the low accessdelay.

TABLE III: WBAN Priority of Service.

BAN priorities WBAN services3 Highest priority medical services2 General health services1 Mixed medical and non-medical services0 Non-medical services

The definition of user priority is based on the collecteddata. The collected data are used to provide various ap-plications and services (medical or non-medical services).The definition of BAN priority is from the perspectiveof applications. In multi-WBANs coexistence scenario, itis necessary to classify the BAN priority with differentservices among inter-WBANs. Different applications maycollect the same kind of data. For example, both interactivemotion sensing games and medical applications for dis-ability assistance are required to collect the motion data.According to the traffic designation in Table II, the motiondata have the same kind of user priorities and contentionwindow. The collision probability increases when the sameUP data are transmitted at the same time in neighborinterference area. But the motion data require lower delayand higher reliability for medical service than that forentertainment. Therefore, we differentiate the BAN prioritybesides the user priority, as shown in Table III. The medicalapplication takes precedence over other non-medical appli-cations. The BAN priority guarantees the privilege of nodesaccessing the channel for health monitoring, especially inthe condition of existing few interference nodes in multipleWBANs coexistence scenario.

IV. INTERFERENCE OF COEXISTENCE MOBILE WBANS

The mutual interference of multi-WBAN coexistenceis caused by the concurrent data transactions of sensornodes from neighbor WBANs. In other words, the body-to-body communications generate interference. Human, asthe wearer of WBAN, is not static. The slow or quickmovement leads to channel dynamics. The body-to-bodyWBAN interference channel modeling with movement canbe found in [41], [42]. In this section, we first calculatethe changing interfering nodes and discuss the dynamicinterference of multiple coexisting WBANs in differentcases. Then we propose a health critical index to guaranteethe transmission privilege of emergency data of a particularnode or WBAN in multi-WBAN coexistence scenario.

A. Dynamic Interference of Coexisting Mobile WBANs

The human mobility is reflected by the velocity anddirection angle of WBAN movement. The interference



5

¡ A CB D

2d > ¡=2d ¡

2d < ¡

mhN

mhN1h

N

X

Y

Fig. 2: The relative position of N1 and Nm.

channel gain is dominated mostly by human movements[43]. For very slow movement, the channel coherence timeis relative long, and the interference channel gain could beconsidered unchanged. The channel fading and path lossof the movement is negligible. In this paper, we pay ourattention to the influence of changing interference intensityrelated to human mobility on network performance, insteadof how human movement manners affect time varyinginterference channel.

The number of sensor nodes attached on a humanbody varies depending on the individual requirements. Weassume all the WBANs are in same configuration. Thetopology of a WBAN can be modeled as a hub centeredcircle. The hub is located near the center of a humanbody with the communication range Υ. We consider thatthe interference range of sensor nodes is the same ascommunication range Υ. The projections of all the nodesare relative uniformly random distribution within the circle.The modeled topology of a WBAN is two dimensionalfor the tractable analysis instead of the three-dimensionaldistribution in reality.

Assuming N1 as the target WBAN, we firstly pay ourattention to the impact of one existing neighbour WBANNm. Every hub knows its coordinate in the network. Thecoordinates of N1h and Nmh are (x1, y1) and (xm, ym),respectively. The inter-WBAN distance d between N1 andNm is defined as the distance from N1h to Nmh. Thedistance d can be calculated as

d =

√(xm − x1)

2+ (ym − y1)

2 (1)

WBAN movements lead to changing relative distanceof inter-WBANs. If an arbitrary WBAN Nm moves inthe direction of target WBAN N1, there are three relativepositions of the two WBANs, as shown in Fig. 2. If d > 2Υ,there is no inference between them. Nm keeps moving fromposition D to position C, then it reaches the edge that caninterfere N1. We mark this critical point d = 2Υ as themoment t = 0. Supposing that Nm continues moving toN1 till position B where d is less than 2Υ resulting inan overlap of the communication zones. Sensor nodes ofNm in the overlapped area become interference sources,increasing the collision probability of data transactions ofthe target WBAN and degrading the network performanceon throughput and packet delay at the same time. The

interference duration ends at the moment when the inter-WBAN distance d equals 2Υ the next time.

The moving direction of Nm at next moment is uncertainsince people have a variety of options. In this paper,the movement of Nm could be classified into two cases:rectilinear motion and redirection curvilinear motion. Thevelocity of Nmh is denoted by vm. The direction anglebetween the leaving and the approaching direction of Nmh

is α. The rectilinear motion of Nm means the directionalangle α is zero or π, as shown in Fig. 3(a). Nmh movesfrom position B to position E. Section 1 is the originaloverlapping area while section 2 is the overlapping areaat the next moment. The curvilinear motion refers thatthe value of α ranges in (0, π), as shown in Fig. 3(b).Nmh moves from position B at moment t1 to position F atmoment t2. The value of α could be acquired in the lawof cosines from Eq. (2).

cosα =d2t1 + d2m∆t − d2t2

2dt1dm∆t(2)

where dt1 and dt2 are the inter-WBAN distance at momentt1 and t2, respectively. dm∆t is the straight-line distanceof Nmh in the time duration ∆t (∆t = t2 − t1). WhenNmh is in continuous movements, we can get dm∆t =√(xmt2 − xmt1)

2+ (ymt2 − ymt1)

2= vm(t2 − t1). The

directional angle α and the value of vm affect the interfer-ence duration.

The interference intensity is directly decided by thenumber of interference nodes nI , which is related to theinter-WBAN distance d. The intersection area of the targetWBAN N1 and Nm is twice of the difference betweensector area (SXAY ) and triangle area (S△XAY ), as shownin Fig. 2. Therefore, the nI can be obtained from Eq. (3).

nI =2Y

πΥ2×

(arccos

d

2ΥΥ2 − d

2

√Υ2 − d2

4

)(3)

where Y is the number of sensor nodes in a WBAN. Themaximum nI can be reached at d = 0 where the twoWBANs are totally overlapped, the value of which is Y .The interference nodes continuously affect the performanceof N1 until Nm leaves.

For the coexistence of more than two WBANs, thedistance between any WBAN and the target WBAN af-fects the number of interference nodes. It makes no greatdifference to the target WBAN whether the other WBANshave overlap or not. In other words, we only focus on thenumber of nodes in the interference range of target WBANand ignore the interference among any other WBANs. Onecase that any two of the three WBANs have an overlap isdescribed in Fig. 4. The number of interference nodes to thetarget WBAN N1 equals the sum of interfering nodes fromNp and Nq in the interference range and can be obtainedusing Eq. (4).

nI = 2Yπ

(arccos

dp

2Υ + arccosdq

2Υ

)−

YπΥ2

(dp

√Υ2 − d2

p

4 + dq

√Υ2 − d2

q

4

) (4)



6

¡

AE B

2d < ¡

mhN1h

N

mv

2

1

mhN

(a) Rectilinear motion

1 hN

¡

A

F

B1

2

¡

mv

1hN

mhN

2d < ¡

mhN

N

2< ¡

v

¡

a

1t

2t

(b) Redirection curvilinear motion

Fig. 3: The relative motion of Nm.

¡

A B

2qd < ¡

phN1hN

qhN

2pd < ¡

Fig. 4: The relative position ofthree WBANs.

where dp =√(xp − x1)

2+ (yp − y1)

2 and dq =√(xq − x1)

2+ (yq − y1)

2, respectively.Given that there are M WBANs in the WBAN co-

existence scenario, the potential interference sources setto the target WBAN at moment t can be represented as{NK |0 ≤ K ≤ M ; ∀K, dk < 2Υ}. Calculating the wholeintersection area of the target WBAN N1 and interferenceWBAN NK(0 ≤ K ≤ M), we can get the total number ofinterference nodes by Eq. (5).

nI =2Y

πΥ2

∑m∈K

(arccos

dm2Υ

Υ2 − dm2

√Υ2 − d2m

4

)(5)

where dm is the distance from an arbitrary WBAN m inthe interference source sets to the target WBAN. dm andnI are affected by the directional angle and the velocityvm of WBAN m.

B. Health Critical Index

The definition of critical index indicate the ratio ofdeviation in sensed physiological data and the normalvalue of that physiological parameter is proposed in [44].The critical index is used to assist a data-rate tuningmechanism for resource sharing [44]. The mechanism isIEEE 802.15.4-based, without differentiating the trafficpriority and the access privilege of the nodes as well asthe interference problem. A fuzzy inferencing-based healthcriticality assessment approach followed by an efficientdecision making model based on the concepts of Markovdecision process [45] is proposed to optimize the energyconsumption in WBAN. It takes body temperature andhuman age as the influential criteria. However, the work in[45] focuses on the individual instead of the multi-WBANscoexistence. To guarantee the transmission delay of timesensitive emergency and burst data, the heath critical index(H) is introduced to determine the order of nodes enjoyingservice and transmission in multi-WBAN coexistence withinterference.

The heath critical index H in our paper considers thetraffic user priority, WBAN priority (BP ) and the severitycondition. The H helps select the critical data among

multiple nodes in the WBANs coexistence scenario. Thenthe node with emergency data performs the CSMA/CA con-tention progress to access the channel. The traffic priorityUPk and WBAN priority BPm can be found in Table IIand Table III. We use the severity index (SI) denotes thedegree of data deviating from the safety range. The detailsof this severity index are available in the original paper[46]. The value of SI satisfies 0 ≤ SI ≤ 1. The values ofH can be obtained through Eq. (6).

H = ρ1 × UPk

7 + ρ2 × SI + ρ3 × BPm

3 , (6)ρ1 + ρ2 + ρ3 = 1

where ρ1 is the weight factor of the user priority, ρ2 isthe weight factor of the data critical condition while ρ3is corresponding to the BAN priority. The values of theweight factors could be calculated according to the AnalyticHierarchy Process (AHP) method [47], [48].

The AHP method provides an effective solution to quan-tify qualitative problems through pairwise comparisons.The comparisons are made using a scale of numbers thatindicates, how much more, one element dominates anotherin regard to the criterion or property. The fundamental scaleof absolute numbers is from 1 to 9. The minimum value 1means the two attributes contribute equally to the objectivewhile the maximum value 9 means one attributes is extremeimportant over another.

TABLE IV: Pairwise comparison matrix.

UPs SI BP Weight factor ρUPs 1 1/5 1/5 0.15SI 5 1 1/3 0.51BP 5 3 1 0.34

Three attributes are considered in the paper, i.e. WBANpriority, user priority, and severity condition. The WBANpriority is strongly important over data deviation. Dataseverity is more important than user priority. In other words,nodes for medical applications have privilege than non-medical applications. Nodes in abnormal condition havehigher heath critical index value than that of normal, whichgets preferential order to access to services. For example,



7

the monitoring data of a UP5 node in a serious fluctuationare more important than a UP6 node in normal condition.The pairwise comparison matrix of the three attributes withrespect to the H is shown in Table IV. The attributes listedon the left are matched one by one to each attribute listedon top according to their importance with respect to thehealth critical index. The weight factors are obtained byAsymptotic Normalization Coefficient (ANC). The derivedweight factor vector (ρ1 ρ2 ρ3)

T based on the judgementsin the matrix is (0.15 0.51 0.34)T . With regard to nodes indisparate WBANs of interfering sources sets, the proposedH guarantees their service quality more thoroughly.

Algorithm 1 Algorithm for determining the service orderof nodesInput: Physiological data from sensor nodesOutput: Service priority of different data types

1: There are M WBANs in the system while each WBANhas Y sensor nodes.

2: for m=1:M do3: compute Hm according to Eq. 64: sort H= [Hm] ↓5: if Hstart

m ≥ Hstartm+1 then

6: serve WBAN m7: for y=1:Y do8: compute Hy

m according to Eq. 69: sort H= [Hy

m] ↓10: if Hy

m > Hy+1m then

11: serve node y in WBAN m12: else13: serve node y + 1 in WBAN m14: end if15: end for16: end if17: end for

The service order algorithm is described in Algorithm 1for different nodes in multi-WBANs coexistence scenario.Hstart

m is the health severity index value of the pendingdata packet located at the head of data queue in hub Nmh.Hy

m is the health severity index value of node y in WBANm. According to Eq. 6, the values of health critical indexof different WBANs and different nodes could be acquired.Then we sort the WBANs or nodes by the value of H fromlarge to small. Anyone with the higher H value is conferredthe higher privilege to access the channel. The introducedH helps to ensure that nodes in emergency condition couldbe processed in lower delay and better services for healthmonitoring.

V. PERFORMANCE ANALYSIS

Multiple concurrent WBANs share the same operatingchannel. Sensors in different WBANs may collide over thechannels. We investigate the slow movement scenario. Thechannel coherence time is relative long. The interferencechannel gain could be considered unchanged. The channel

fading and path loss of the movement is neglected. In thissection, we analyze the normalized throughput and averagepacket delay by employing a three-dimensional Markovmodel.

A. Superframe Structure

In the light of IEEE 802.15.6, two Exclusive AccessPhases (EAPs), two Random Access Phases (RAPs), twoManaged Access Phases (MAPs), and a Contention AccessPhase (CAP) are placed in order in the superframe. Theseven access phases phases are optional. In this paper,only the contention phases EAP and RAP are selected.All the nodes in the network are synchronized to supportthe contention-based mechanism while the hub establishesthe time reference operating in the beacon mode withsuperframes. Each superframe is bounded by a maximumnumber of 255 equal length allocation slots. The layout ofsuperframe is shown in Fig. 5.

B

Superframe n

EAP1 RAP1

SIF

S

slot

slot

slot

slot

slot

Frametransaction

Inactive

SIF

S

Fig. 5: The superframe structure.

The Beacon (B) frame is transmitted by the hub at the be-ginning of each superframe to notify network managementinformation such as BAN identification, synchronizationand coordination of medium access. If there is highestpriority data, the hub and nodes could acquire a contendedallocation Short Interframe Space (SIFS) after the begin-ning of EAP (EAP1 or EAP2) without performing theCSMA/CA. In RAP, all UPk nodes perform CSMA/CAto contend for available slots to transmit management ordata frames. If there are no more pending transmissions indata queues, all nodes including hub turn into inactive stateover some time intervals for energy saving.

B. Markov Chain Model for A UPk Node With Interference

Nodes contend to access the channel in CSMA/CAmechanism for intra-WBAN. In light of the random char-acteristics of contention progress of the nodes, the accessattempts of node having pending data to be transmitted canbe analyzed with a three-dimensional discrete time Markovchain. The Markov chain model for the contention progressis shown as Fig. 6. All the notations used in the analysisare presented in Table I.

The state of a node at moment t in the con-tention progress can be described by three variables{k,BT (t), CW (t)}, where k is the user priority of anode, BT is the backoff times, and CW is the contentionwindow value in Backoff Counter (BC). Prioritized accessfor different types of traffic with different UPs shall beattained through the predefined CW values. The relation



8

k,0,1

1-Pu

Pu k,0,Wk,0-1

1-Pu

Pu k,0,Wk,0

1-Pu

PuPu

k,1,1Pu k,1,Wk,1-1

1-Pu

Pu k,1,Wk,1

1-Pu

PuPu

k,j,1Pu k,j,Wk,j-1

1-Pu

Pu k,j,Wk,j

1-Pu

PuPu

k,s,1Pu k,s,Wk,s-1

1-Pu

Pu k,s,Wk,s

1-Pu

PuPu

k,BTmax,0 k,BTmax,1Pu k,BTmax,Wk,s-1

1-Pu

Pu k,BTmax,Wk,s

1-Pu

PuPu

k,0,0

k,1,0

k,j,0

k,s,0

,0b kP W

,b k sP W

max,BTb kP W

Idle

max,b k BTP W

1 bP-

,b k jP W

1-Pu

,1b kP W

1-Pu

1-Pu

1-Pu

1 bP-

1 bP-

1 bP-

1 bP-

Fig. 6: Markov model for a node with UPk.

between contention window bounds CWmax and CWmin

and the corresponding data traffic is shown in Table II.Nodes which have data to be transmitted will switch their

states between back-off and data transmission, or nodes willturn into inactive state to save energy. At the initial state,a node with user priority k denoted by UPk senses thechannel having been idle for at least a short interframespacing (pSIFS) time interval. Then it selects a randominteger from the interval [1, CWUPk

] as the BC values,where CWUPk

∈ (CWmin[UPk], CWmax[UPk]) . The valuesof CWmin[UPk] and CWmax[UPk] can be obtained fromTable II. Furthermore, the BC is decreased by one for eachidle CSMA/CA slot, and nodes will switch to transmissionstate once the BC reaches zero meanwhile the channel isidle. Otherwise, the BC will be locked if the channel isbusy or the remaining time interval is not long enough forcompleting a frame transaction. The BT increases by onein case of an unsuccessful medium access due to collisions.Nodes conduct the backoff progress until the frame istransmitted successfully or initiate a new data transmissionwhen the BT value reaches the maximum backoff retrylimits BTmax.

We suppose that there are M WBANs in a dense deploy-ment scenario. There are one hub and nk nodes of UPk,

where7∑

k=0

nk = Y, k = 0, ..., 7, operates in the beacon

mode with superframe boundaries in a random selectedtarget WBAN. Assuming that the access probability ofnode UPk in a random CSMA/CA slot is τk. The accessprobability of interference nodes is considered sharing thesame probability with a UPk node. The probability thatchannel is idle during the contention intervals of the target

WBAN is

Pidle =7∏

k=0

(1− τk)nk (7)

The backoff counter is locked when the channel is busyand the remainder time is not long enough completing aframe transaction during its backoff stage. The probabilitiesthat the backoff counter is locked due to the insufficientremaining time for a frame transaction from the currentCSMA contended allocation slot to the end of the EAPand RAP period are

PUPk−inlock =TR − Tdata − TkCW

TR, RAP (8)

PUP7−inlock =TE + TR − Tdata − T7CW

TE + TR, EAP and RAP

(9)where TE and TR is the number of CSMA slots allocated

to UPk in EAP and RAP access phases, respectively. Tdata

is the successful data transmission time. The backoff timesis marked as l and the mean backoff value is approximatelycalculated as

CWmin[UPk]+CWmax[UPk]

4 [31]. TkCW is thebackoff time and can be calculated as

TkCW =

BTmax∑l=0

Tc−slot

(CWmin[UPk] + CWmax[UPk]

)4

(10)Tc−slot = TCCA + TMP (11)

where Tc−slot is the CSMA slot length with a fixedduration. TCCA is the Clear Channel Assessment (CCA)time and TMP is the time cost that MAC transfers a frameto the Physical (PHY) layer.

For the highest user priority data, the contention attemptsare exclusively allowed in EAP. A hub or a node may regardthe combined EAP (EAP1 or EAP2) and RAP (RAP1or RAP2) as a single EAP1 or EAP2 to allow continualinvocation of CSMA/CA and improve channel utilization.Therefore, the probability of the channel is idle for UP7

in a CSMA slot during EAP and RAP is different [49].The idle channel probability P7idle of UP7 node in EAPand RAP can be obtained from Eq. (12). Then the randomselected UP7 node lock its backoff counter owing to thebusy channel can be obtained from P7b = 1− P7idle.

P7idle =TE(1− τ7)

n7I+n7−1

TE + TR+

TRPidle(1− τ7)n7I−1

TE + TR(12)

where n7I is the number of interference nodes with UP7

from inter-WBANs. The value of n7I can be obtainedfrom Eq. (5). n7 is the number of UP7 nodes from intra-WBAN. Pidle is the probability that channel is idle duringthe contention intervals of the target WBAN.

In RAP, a node with UPk locks its backoff counter whenit finds the channel is busy in its backoff counter count-down. In other words, at least one of the remaining nodesof intra-WBAN or interference nodes of inter-WBAN sendspackets or ACK packets, leading to the collision. There arenkI interference sources from inter-WBAN affecting the



9

data transactions. Supposing that the node finds the channelis idle in probability Pkidle during its backoff period, thePkidle is described as

Pkidle =

(1− τk)nkI+nk

∏i=k

(1− τi)ni

(1− τk)

= (1− τk)nkI−1

7∏i=0

(1− τk)nk

(13)

The packet collision probability Pb for each UPk nodeat any backoff stage equals the probability of busy channeland could be expressed as

Pb = 1− Pkidle = 1−(1− τk)nkI−1

7∏i=0

(1− τk)nk (14)

According to Eq. (7)-Eq. (14), the probability Pu thatthe backoff counter is unlock can be obtained in Eq. (15).Eq. 15 (a) is particularly for UP7 nodes to get the unlockedprobability in EAP and RAP. The probability of the otherUPk(k = 0, ..., 6) data unlock its back counter in RAP isindicated in Eq. 15 (b).

Pu = (1− Pb)(1− PUPk−inlock)

=

{P7idle(

Tdata+TCW

TE+TR), (a)

Pkidle(1− τk)nI−1

(Tdata+TkCW

TR

), (b)

(15)

If the collision occurs, the node resets the CW value ofbackoff counter according the regulations aforementioned,selects a random integer from the interval [1, CWUPk

] andturns into the next backoff stage. Thereafter, the changingCWUPk

value Wk,j during the jth backoff period for anode of UPk is predicted as following,Wk,0 = CWmin[UPk], j = 0Wk,j = Wk,j−1, forj is an odd number, 1 ≤ j ≤

BTmax

Wk,j = min{2Wk,j−1, CWmax[UPk]}, forj is an evennumber, 1 < j ≤ BTmax

The first time CW value reaches CWmax[UPk] at itssth backoff stage and remains unchanged in the follow-ing backoff retry procedure. The stationary distributionprobability of the Markov chain is Mk,j,i, where k isthe user priority (k = 0, ..., 7), j is the value of backoff stages (j = 0, 1, ..., BTmax), and i is the BC values(i = 0, 1...,Wk,j). The state transition probabilities aregiven by

P {k, j + 1,Wk,j |k, j, 0} =Pb

Wk,j, ∀j ∈ [0, BTmax − 1]

(16)P {k, j + 1, 0 |k, j, 0} = Pb (17)

P {k, j,Wk,j − 1 |k, j,Wk,j } = Pu (18)P {k, j,Wk,j |k, j,Wk,j } = 1− Pu (19)

Nodes will discard the packet if the packet fails in access-ing the channel. The failure data transmission probabilityPf of UPk node is expressed as

Pf = PBTmax+1b (20)

In steady state, we can get the following relations ac-cording to the Markov chain regularities. Eq. (25) showsthe sum of all the state probabilities is equal to one.

Mk,j,0 = P jbMk,0,0, ∀j ∈ [0, BTmax] (21)

Mk,BTmax,0 =PBTmax

b

1− PbMk,0,0 (22)

M0,0,0 =Pb

1− PbMk,0,0 (23)

Mk,j,i =Wk,j − i+ 1

Wk,jPuMk,j,0, i ∈ [1,Wk,j ](24)

1 = M0,0,0 +

BTmax∑j=0

Wk,j∑i=0

Mk,j,i (25)

All the probabilities have a strong correlation with ac-cess probability τ . Once the value of backoff counter iszero, the node initiates an attempt to access the channel.Therefore, the access probability of node UPk in a randomCSMA slot τk is the sum all the steady states probabilities

τk=BTmax∑j=0

Mk,j,0.

C. Normalized Throughput and Average Access DelayWe define the normalized throughput as the ratio of

time cost of average successful transmitted payloads tothe total intervals between two consecutive transmissions.A successful packet transmission time is Tpk and can beobtained from

Tpk =Npacket

Rs

=Npreamble +Nheader × Sheader +

Ntotal×SPSDU

log2M

Rs(26)

where Rs is the sysmbol rate. Npreamble and Nheader

stand for the length of Physical Layer Convergence Pro-tocol (PLCP) preamble and header, respectively. Sheader

denotes the spreading factor for the PLCP header. SPSDU

denotes the spreading factor for Physical Service Data Unit(PSDU). M is the cardinality of the constellation of a givenmodulation scheme. Ntotal is the total bits which flow inthe PHY layer and can be derived in Eq. 27.

Ntotal = NPSDU + log2M

⌈NPSDU + (n− k)NCW

log2M

⌉+

(n− k)NCW − [NPSDU + (n− k)NCW ](27)

where NCW denotes the number of Bose, Ray-Chaudhuri,Hocquenghem code (BCH) codewords in a frame. In IEEE802.15.6, NCW =

⌈NPSDU

k

⌉, where n and k are selected

by BCH encoder to achieve the desired code rate k/n.NPSDU is the length of the PSDU. This component isformed by concatenating the MAC header with the MACpayload and frame check sequence (FCS). Given the lengthof MAC header NMHeader, MAC payload NMPayload andFCS NFCS , the NPSDU can be acquired from Eq. (28).

NPSDU = 8× (NMHeader +NMPayload +NFCS) (28)



10

TMPayload=NMPayload × SPSDU

log2M ×Rs(29)

The normalized throughput with the interference is cal-culated as

Sk=τkPkidleTMPayload

PidleTc−slot + τkPkidleTpk + τk (1− Pkidle)Tfpk(30)

where Tc−slot is the contended CSMA/CA slot length.Pkidle is the probability that the channel is idle during thebackoff period of node UPk. Pkidle is influenced by thenumber of interfering nodes, the value of which can beobtained from Eq. (12) and Eq. (13). Ts in Eq. (31) is themean time of a successful transmission.

Ts = Tpk + TpSIFS + TI−ACK (31)

While the occupied slot length for a failure packetdelivery Tfpk is described as

Tfpk = Tpk + TpSIFS + Ttimeout (32)

An absent I-ACK in an interval Ttimeout will be treatedas a failure transmission. The Tfail is the total time costbefore a packet has been successfully transmitted.

Tfail =

BTmax∑l=0

lPbl(Tpk + TpSIFS + Ttimeout) + TkCW

(33)where l is the failure times before the packet is successfullytransmitted. The average packet delay is defined as the timeinterval from the packet that is ready to be transmitted tothe moment that the I-ACK frame of this packet is received.In Eq. (34), the average delay Dk is given as

Dk=Ts + Tfail (34)

By substituting Eqs. (31) and (33) into Eq. (34), theaverage access delay can be calculated. Both the normalizedthroughput and the average access delay consider the trafficdifferentiation and interference dynamics.

VI. PERFORMANCE EVALUATION

In this section, we evaluate the normalized throughputand the average access delay of coexisting WBANs basedon IEEE 802.15.6 CSMA/CA mechanism. Numerical re-sults of the performance evaluation based on the proposedinterference model are implemented and verified by thesimulation tool MATLAB. One-hop star topology structurein a WBAN is considered in the monitoring scenario.The channel fading and path loss of the movement couldbe neglected in slow movements. The channel conditionis ideal during the data transmission. Each hub operatesin the beacon mode with superframes and each nodeproduces one UP traffic in the network with all the da-ta fluctuation in normal range. The health critical indexvalue H is in corresponding to their UP values of eachnode. The length of an allocation slot Tslot is equal topAllocationSlotMin+ L× pAllocationSlotResolution[15]. The maximum MAC payload size is 250 bytes in theanalysis. The other relevant simulation parameters are listedin Table V.

TABLE V: System parameters

Parameters Values Parameters ValuesSuperframe length 1 s Beacon 15 bytesCSMA slot length 145 µs FCS 2 bytes

pAllocationSlotMin 500 µs L 7pAllocationResolution 500 µs M 4

Max BAN Size 64 SPSDU 1MAC header 7 bytes Sheader 4

PLCP preamble 90 bits pCCA time 105 µsPLCP header 31 bits pMIFS 20 µsSymbol rate 600 ksps pSIFS 75 µsBuffer size 20000 bytes Coderate(k/n) 51/63

Frequency band 2400-2438.5 MHz Timeout 30 µs

A. Normalized Throughput

The normalized throughput of the nodes with user pri-orities varying from 0 to 7 are shown in Fig. 7- Fig. 9. Atfirst, we evaluate the normalized throughput of nodes withno interference. In other words, there is no other coexistingWBANs in the communication range of the target WBANN1. The relationship between normalized throughput andthe varied payloads is shown in Fig. 7. There are 8sensor nodes except the hub in the network. It is obviousthat the normalized throughput of UPk node increases asthe payload increases and the UP7 node increases fasterthan any other nodes. The reason is that the nodes withhighest user priority traffic can transmit data in the EAPperiod and RAP period. Moreover, UP7 node enjoys asmallest contention window value and has a higher accessprobability as well as lower collision probability than anyother nodes in the CSMA/CA period. We can see that thenormalized throughput of UP5 and UP4 nodes, UP3 andUP2 nodes, UP1 and UP0 nodes almost have the sametendency while the payload increases. This is because someof them share the same minimum contention window valueand have small difference in access probability.

Fig. 8 describes the relationship between normalizedthroughout and the number of nodes with a payload of 100bytes and there is no interfering nodes. As the number ofnodes in the network increases to the maximum WBANsize, i.e. 64 sensor nodes in a WBAN, the normalizedthroughput suffers a sharply decrease. Especially when thenumber of nodes is more than 40 in the network, thenormalized throughput of UPk nodes (except UP7 nodes)is almost zero. All this is owing to the high collisionprobability of increasing number of nodes. Although insuch a condition, the highest priority UP7 node can stilltransmit data in the particular EAP period, contending toaccess the channel with few UP7 traffic data. The QoSof nodes with the higher user priority can be guaranteed.The number of nodes really has a big influence on thenormalized throughput. It is suggested that the number ofnodes in one WBAN is better no more than 24 from Fig.8.

We investigate how a random mobile WBAN Nm af-fects the performance of target WBAN N1 on normalizedthroughput in Fig. 9. There are 8 nodes with payload of



11

0 50 100 150 200 250Payload(bytes)

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

Nor

mal

ized

Thr

ough

put

Normalized throughput of nodes with UPk

UP7UP6UP5UP4

UP3UP2UP1UP0

Fig. 7: Normalized throughput versuspayloads.

0 10 20 30 40 50 60 70Number of nodes

0

0.05

0.1

0.15

0.2

0.25

0.3

Nor

mal

ized

Thr

ough

put


UP7UP6UP5UP4UP3UP2UP1UP0

Fig. 8: Normalized throughput versusnumber of nodes.

0 0.2 0.4 0.6 0.8 1 1.2

Velocity (m/s)

0.12

0.14

0.16

0.18

0.2

0.22

0.24

0.26

0.28

Nor

mal

ized

Thr

ough

put

Normalized throughput of nodes with UPkUP7UP6UP5UP4UP3UP2UP1UP0

Fig. 9: Normalized throughput versusvelocity.

100 bytes in target WBAN N1. The communication rangeof N1 is 1.5 m. The movement of Nm does not affectthe performance of N1 when the inter-WBAN distanceis large than 3 m. The velocity of Nm regards zero toN1. When the inter-WBAN distance is equal to or lessthan 3 m, the Nm becomes interference source. We focuson the slow motion of patients or old people in healthmonitoring. Nm keeps in slow linear motion towards toN1 during the interference duration. Within the specifictime, the velocity of Nm generates varying interferencedynamics. The normalized throughput of UPk nodes inN1 decrease with the increasing velocity of Nm. Thenormalized throughput variation of high UP node is largerthan lower UP node. The normalized throughput of UP7

decreases almost 32% when Nm moves to N1 at the speed1.2m/s. The higher user priority nodes have higher accessprobability and suffer higher collision probability with theincreasing interference.

The impacts of velocity combined the effects of pay-loads and number of nodes acting together on normalizedthroughput of UPk nodes are shown in Fig. 10 and Fig.11, respectively. The comparison of normalized throughputof different user priority traffic varieties is obvious. Thenormalized throughput of UPk nodes increases with thepayloads increasing, and decreases with the increasingvelocity. UP7 traffic always has highest throughput whileUP0 has the lowest throughput. The normalized throughputof several typical user priority are shown in Fig. 10 (b) andFig. 11 (b). The normalized throughput is affected greaterby the number of nodes in the network than the velocityparameter in Fig. 11. The variation of velocity means thevarying number of interference nodes in the network, whichseriously affects higher UPs. Once the number of nodes inthe WBAN is larger than 40, the packets can hardly betransmitted successfully.

B. Average Access Delay

The variations of average packet access delay against thepayloads for different UPs with no interference are shownin Fig. 12. We choose the typical UPk(k = 3, ..., 7) nodes

to illustrate the results. The results of all the UPk nodescan be found in Fig. 15 and Fig. 16. The delay increasesas the length of payloads increases. In this case, only 8nodes are considered in a WBAN. Every node producesone kind of user priority data. The higher UP of the nodeis, the lower access delay is. In the maximum payload 250bytes, the delay of UP7 node is merely ninth of the UP0

node, and is only 62% of the UP6 node. The reason isthat nodes with higher UPs have higher packet transmissionprobability compared to nodes with lower UPs.

The delay of nodes increases with the increasing numberof nodes in the network, as shown in Fig. 13. All thenodes are in a fixed payload (100 bytes). Nodes have toback off more times in a successful data transmission withthe increasing number of nodes in the network. The accessprobabilities of all UPk nodes become smaller with theincreasing number of nodes, leading to the increasing delay.As the number of nodes increases from 8 to 32, the accessdelay of UP7 have doubled, and the delay of UP6 increases70%, while the growth of UP5 and UP4 is 60% and 38%,respectively. If there are too many nodes in a WBAN, allnodes suffer a severe contention progress and very highaverage access delay.

The average access delay of the target WBAN N1

increases with the increasing velocity of neighbor inter-ference WBAN Nm. The higher velocity of Nm within thespecific time makes more nodes in Nm become interferencenodes to N1. Fig. 14 shows impacts of interference nodeson the average access delay of UPk nodes. The UP7 nodeswith no interference cost less than 70% time of Nm inwalking motion. The average access delay of UPk nodesincreases with the increasing interference nodes due to thechanging velocity. The reason is that the nodes have tosuffer a relative long back-off procedure.

The combined action of velocity and different length ofpayloads on average access delay are shown in Fig. 15. Allthe nodes are in normal states that there is no dramatic shifton the monitoring data. The delay increases as the length ofpayloads increases and higher UP nodes have lower delay.The UP7 always has the lowest delay while UP0 alwayshas the highest delay. The reason is that nodes with higher



12

Velocity(m/s) Payload(bytes)

0

0.1

0.2

1.5 300

0.3

Nor

mal

ized

thro

ughp

ut


0.4

0.5

1 2000.5 100

0 0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

UP6

UP5,UP4

UP7

UP3

UP2

UP1,UP0

(a) Influence of velocity and payloads of UPk (b) Subgraphs of UPkFig. 10: Normalized throughput versus velocity and payloads.

(a) Influence of velocity and number of nodes (b) Subgraphs of UPkFig. 11: Normalized throughput versus and velocity and number of nodes.

0 50 100 150 200 250

Payload(bytes)

0

5

10

15

20

25

Ave

rage

acc

ess

dela

y(m

s)

Average access delay of nodes with UPkUP7UP6UP5UP4UP3

Fig. 12: Average access delay versuspayloads.

0 10 20 30 40 50 60 70Number of nodes

5

10

15

20

25

30

Ave

rage

acc

ess

dela

y(m

s)

Average access delay of nodes with UPkUP7UP6UP5UP4UP3

Fig. 13: Average access delay versusnumber of nodes.

0 0.2 0.4 0.6 0.8 1 1.2

Velocity(m/s)

5

10

15

20

25

Ave

rage

acc

ess

dela

y(m

s)

Average access delay of nodes with UPk

UP7UP6UP5UP4UP3

Fig. 14: Average access delay versusvelocity.



13

Payload(bytes)Velocity(m)

0300

5

1.5

10


15

Ave

rage

acc

ess

dela

y(m

s)

200

20

1

25

1000.500

5

10

15

20UP3

UP4

UP5

UP6

UP7

(a) Influence of velocity and payloads of UPk (b) Subgraphs of UPkFig. 15: Average access delay versus velocity and payloads.


Velocity(m/s) Number of nodes

5

10

1.5

15

20

Ave

rage

acc

ess

dela

y(m

s)

25

1

30

60400.5

200 0

10

15

20

25

UP5

UP3

UP4

UP7

UP6

(a) Influence of velocity and number of nodes (b) Subgraphs of UPkFig. 16: Average access delay versus velocity and number of nodes.

UPs have higher packet transmission probability comparedto nodes with lower UPs. From Fig. 15 (b), we can seethat the delay is much influenced by the payload when itis larger than 100 bytes. The varying velocity of neighborWBAN makes little effect on the average access delay ofUPk nodes in the target WBAN in spite of the changinglength of payloads.

The average access delay of the IEEE 802.15.6 basedCSMA/CA protocol for different UP nodes by consideringthe interference of coexisting mobile neighbour WBAN andthe different number of nodes in intra-WBAN are shown inFig. 16 (a) and (b). The network performance on delay isseverely deteriorated when the number of nodes increases,since the packet collision happens not only among nodeswith different UPs but also among nodes with same UPs.The higher UPs are more sensitive to the collision dueto the high access probability. Though the UP7 nodes aremore volatile than any other UP nodes when the numberof nodes and the velocity of neighbor WBAN changes, theaverage access delay of UP7 nodes is the lowest in thesame condition. The delay of UP3 is almost two to threetimes as the delay of UP7 nodes in different number of

nodes at the same velocity. To guarantee the monitoringdata be transmitted in time, the number of nodes in oneWBAN had better not exceed 30.

Human movements lead to the increasing number ofinterfering WBANs and interfering nodes, which increasesthe collision probability of nodes accessing the channel.The impacts of dynamic mobility with different numberof coexisting WBANs on the normalized throughput andaverage access delay are reflected in Fig. 17 and Fig. 18.The performance of UP7 node in target WBAN is evaluatedwith different WBAN priorities. The BP3, BP2 and BP1make the target WBAN in highest, moderate and low healthcritical index among the coexisting WBANs, respectively.The normalized throughput decreases 70% while the aver-age access delay is doubled compared with single WBAN.The higher WBAN priority guarantees the node in highernormalized throughput and lower average access delay withthe increasing interference. The evaluation results in Fig.17 and Fig. 18 facilitate the network configuration andoptimization in WBAN coexistence scenario.



14

0 2 4 6 80.00

0.05

0.10

0.15

0.20

0.25

0.30

No

rmal

ized

Th

rou

gh

pu

t

Number of coexisting WBANs

BP1

BP2

BP3

Fig. 17: Normalized throughput versus number ofcoexisting WBANs.

0 2 4 6 80

5

10

15

20

25

30

Av

era

ge a

ccess

dela

y (

ms)

Number of coexisting WBANs

BP1

BP2

BP3

Fig. 18: Average access delay versus number of coexistingWBANs.

VII. CONCLUSION

In this paper, we have analyzed the dynamic interferencegenerated by varying number of interfering nodes fromneighbour WBANs owing to human mobility in multi-WBAN coexistence scenario. We have developed a threedimensional Markov chain model to investigate the interfer-ence dynamics on network performance of IEEE 802.15.6-based CSMA/CA protocol. Moreover, a health criticalindex considering the data traffic differentiate, data severityand WBANs priority has been designed to guarantee theemergency data taking precedence over general traffic toaccess the channel with lower delay and high reliability.The normalized throughput and average access delay un-der different interference conditions have been derived intractable expressions. The appropriate number of nodes ina WBAN and distance to the nearby WBANs could beacquired according to the analytical results, which is usefulto network configuration and interference mitigation. Inthe future, we will consider the dynamic channel statusin different movement manners and analyze the impacts onnetwork performance. The interference mitigation schemesare also scheduled in WBANs coexistence.

ACKNOWLEDGMENT

This work was supported by the National Natural ScienceFoundation of China under Grant No. 61571350, Key Re-search and Development Program of Shaanxi (Contract No.2017KW-004, 2017ZDXM-GY-022), and the 111 Project(B08038).

REFERENCES

[1] R. Chavez-Santiago, K. Sayrafian-Pour, A. Khaleghi, K. Takizawa,J. Wang, I. Balasingham, and H.-B. Li, “Propagation models forIEEE 802.15. 6 standardization of implant communication in bodyarea networks,” IEEE Commun. Mag., vol. 51, no. 8, pp. 80–87,2013.

[2] K. Zhang, K. Yang, X. Liang, Z. Su, X. Shen, and H. H. Luo,“Security and privacy for mobile healthcare networks: from a qualityof protection perspective,” IEEE Wireless Commun., vol. 22, no. 4,pp. 104–112, 2015.

[3] A. Al-Fuqaha, M. Guizani, M. Mohammadi, M. Aledhari, andM. Ayyash, “Internet of things: A survey on enabling technologies,protocols, and applications,” IEEE Commun. Surveys Tuts., vol. 17,no. 4, pp. 2347–2376, 2015.

[4] Q. Ye and W. Zhuang, “Distributed and adaptive medium access con-trol for internet-of-things-enabled mobile networks,” IEEE InternetThings J., vol. 4, no. 2, pp. 446–460, 2017.

[5] Q. Ye, W. Zhuang, L. Li, and P. Vigneron, “Traffic-load-adaptivemedium access control for fully connected mobile ad hoc networks,”IEEE Trans. Veh. Technol., vol. 65, no. 11, pp. 9358–9371, 2016.

[6] X. Wu, Y. I. Nechayev, C. C. Constantinou, and P. S. Hall, “Interuserinterference in adjacent wireless body area networks,” IEEE Trans.Antennas Propag., vol. 63, no. 10, pp. 4496–4504, 2015.

[7] X. Wang and L. Cai, “Interference analysis of co-existing wirelessbody area networks,” in Proc. IEEE GLOBECOM. Houston, TX,USA, Dec. 2011, pp. 1–5.

[8] Z. Zhang, H. Wang, C. Wang, and H. Fang, “Interference mitigationfor cyber-physical wireless body area network system using socialnetworks,” IEEE T. Emerg. Top. Com., vol. 1, no. 1, pp. 121–132,2013.

[9] T. T. T. Le and S. Moh, “Interference mitigation schemes for wirelessbody area sensor networks: A comparative survey,” Sensors, vol. 15,no. 6, pp. 13 805–13 838, 2015.

[10] L. Zou, B. Liu, C. Chen, and C. W. Chen, “Bayesian game basedpower control scheme for inter-WBAN interference mitigation,” inProc. IEEE GLOBECOM. San Diego, CA, USA, Dec. 2015, pp.240–245.

[11] S. Kim, S. Kim, J.-W. Kim, and D.-S. Eom, “A beacon intervalshifting scheme for interference mitigation in body area networks,”Sensors, vol. 12, no. 8, pp. 10 930–10 946, 2012.

[12] S. Kim, S. Kim, J. W. Kim, and D. S. Eom, “Flexible beaconscheduling scheme for interference mitigation in body sensor net-works,” in Proc. IEEE SECON. Seoul, Korea (South), Jun.2012,pp. 157–164.

[13] A. Meharouech, J. Elias, and A. Mehaoua, “Future body-to-bodynetworks for ubiquitous healthcare: a survey, taxonomy and chal-lenges,” in International Symposium on Ubi-HealthTech. Beijing,China, May 2015, pp. 1–6.

[14] X. Yuan, C. Li, K. Zhang, Q. Ye, N. Cheng, N. Zhang, and X. S.Shen, “Dynamic interference analysis of coexisting mobile WBANsfor health monitoring,” in Proc. IEEE ICC. Kansas City, MO, USA,May 2018, pp. 1–6.

[15] I. S. Association, “IEEE standard for local and metropolitan areanetworks-part 15.6: wireless body area networks,” IEEE Std forInformation Technol., IEEE, vol. 802, no. 6, pp. 1–271, 2012.

[16] A. Samanta and S. Misra, “Energy-efficient and distributed networkmanagement cost minimization in opportunistic wireless body areanetworks,” IEEE Trans. Mobile Comput., vol. PP, no. 99, pp. 1–14,2017.

[17] Y. Hou, M. Li, and S. Yu, “Making wireless body area networksrobust under cross-technology interference,” IEEE Trans. WirelessCommun., vol. 16, no. 1, pp. 429–440, 2017.



15

[18] S. Marwa, A.-D. Ahmed, R. Imed, and N. Youssef, “Wireless bodyarea network (WBAN): A survey on reliability, fault tolerance, andtechnologies coexistence,” ACM Comput. Surv., vol. 50, no. 1, p. 3,2017.

[19] W. Sun, Y. Ge, and W.-C. Wong, “A stochastic geometry analysisof inter-user interference in IEEE 802.15. 6 body sensor networks,”in Proc. IEEE WCNC. New Orleans, LA, USA, Mar. 2015, pp.1912–1917.

[20] F. Jameel, A. Butt, and U. Munir, “Analysis of interference inbody area networks over generalized fading,” in Proc. IEEE ICET.Islamabad, Pakistan, Oct. 2016, pp. 1–6.

[21] M. N. Anjum and H. Wang, “Optimal resource allocation for deeplyoverlapped self-coexisting WBANs,” in Proc. IEEE GLOBECOM.Washington, DC, USA, Dec. 2016, pp. 1–7.

[22] J. Elias, S. Paris, and M. Krunz, “Cross-technology interferencemitigation in body area networks: An optimization approach,” IEEETrans. Veh. Technol., vol. 64, no. 9, pp. 4144–4157, 2015.

[23] I. . W. Group, “IEEE standard for local and metropolitan areanetworks-part 15.4: Low-rate wireless personal area networks (LR-WPANs),” IEEE Std, vol. 802, pp. 4–2011, 2011.

[24] K. Zhang, X. Liang, M. Baura, R. Lu, and X. S. Shen, “PHDA: Apriority based health data aggregation with privacy preservation forcloud assisted WBANs,” Information Sciences, vol. 284, pp. 130–141, 2014.

[25] W. Sun, Y. Ge, Z. Zhang, and W.-C. Wong, “An analysis frameworkfor interuser interference in IEEE 802.15. 6 body sensor networks: Astochastic geometry approach,” IEEE Trans. Veh. Technol., vol. 65,no. 10, pp. 8567–8577, 2016.

[26] X. Cao, J. Chen, Y. Cheng, X. S. Shen, and Y. Sun, “An analyticalmac model for IEEE 802.15. 4 enabled wireless networks withperiodic traffic,” IEEE Trans. Wireless Commun., vol. 14, no. 10,pp. 5261–5273, 2015.

[27] G. Bianchi, “Performance analysis of the IEEE 802.11 distributedcoordination function,” IEEE J. Sel. Areas Commun., vol. 18, no. 3,pp. 535–547, 2000.

[28] S. Moulik, S. Misra, and D. Das, “AT-MAC: Adaptive MAC-framepayload tuning for reliable communication in wireless body areanetworks,” IEEE Trans. Mobile Comput., vol. 16, no. 6, pp. 1516–1529, 2017.

[29] P. Park, P. Di Marco, C. Fischione, and K. H. Johansson, “Modelingand optimization of the IEEE 802.15. 4 protocol for reliable andtimely communications,” IEEE Trans. Parallel Distrib. Syst., vol. 24,no. 3, pp. 550–564, 2013.

[30] S. Sarkar, S. Misra, B. Bandyopadhyay, C. Chakraborty, and M. S.Obaidat, “Performance analysis of IEEE 802.15. 6 MAC protocolunder non-ideal channel conditions and saturated traffic regime,”IEEE Trans. Comput., vol. 64, no. 10, pp. 2912–2925, 2015.

[31] S. Rashwand, J. Misic, and V. B. Misic, “Analysis of CSMA/CAmechanism of IEEE 802.15. 6 under non-saturation regime,” IEEETrans. Parallel Distrib. Syst., vol. 27, no. 5, pp. 1279–1288, 2016.

[32] M. Salayma, A. Al-Dubai, I. Romdhani, and Y. Nasser, “Newdynamic, reliable and energy efficient scheduling for wireless bodyarea networks (WBAN),” in Proc. IEEE ICC. Paris, France, May2017, pp. 1–6.

[33] D. R. Chen, “A QoS bandwidth allocation method for coexistence ofwireless body area networks,” in Proc. PDP 2017. St. Petersburg,Russia, Mar. 2017, pp. 251–254.

[34] G.-T. Chen, W.-T. Chen, and S.-H. Shen, “2L-MAC: A MACprotocol with two-layer interference mitigation in wireless body areanetworks for medical applications,” in Proc. IEEE ICC. Sydney,Australia, Jun. 2014, pp. 3523–3528.

[35] G. R. Tsouri, S. R. Zambito, and J. Venkataraman, “On the benefitsof creeping wave antennas in reducing interference transactions onbiomedical etween neighboring wireless body area networks,” IEEETrans. Biomed. Circuits Syst., vol. 11, no. 1, pp. 153–160, 2017.

[36] S. H. Cheng and C. Y. Huang, “Coloring-based inter-wban schedul-ing for mobile wireless body area networks,” IEEE Trans. ParallelDistrib. Syst., vol. 24, no. 2, pp. 250–259, 2013.

[37] L. Wang, C. Goursaud, N. Nikaein, L. Cottatellucci, and J.-M. Gorce,“Cooperative scheduling for coexisting body area networks,” IEEETrans. Wireless Commun., vol. 12, no. 1, pp. 123–133, 2013.

[38] Y. Tselishchev, A. Boulis, and L. Libman, “Variable schedulingto mitigate channel losses in energy-efficient body area networks,”Sensors, vol. 12, no. 11, pp. 14 692–14 710, 2012.

[39] W. Huang and T. Q. Quek, “Adaptive CSMA/CA MAC protocol toreduce inter-WBAN interference for wireless body area networks,”in Proc. IEEE BSN. Cambridge, MA, USA, Jun. 2015, pp. 1–6.

[40] D. B. Smith, D. Miniutti, T. A. Lamahewa, and L. W. Hanlen,“Propagation models for body-area networks: A survey and newoutlook,” IEEE Antennas Propag. Mag., vol. 55, no. 5, pp. 97–117,2013.

[41] D. B. Smith and L. W. Hanlen, “Channel modeling for wireless bodyarea networks,” in Ultra-Low-Power Short-Range Radios. Springer,2015, pp. 25–55.

[42] S. Van Roy, F. Quitin, L. Liu, C. Oestges, F. Horlin, J.-M. Dricot,and P. De Doncker, “Dynamic channel modeling for multi-sensorbody area networks,” IEEE Trans. Antennas Propag., vol. 61, no. 4,pp. 2200–2208, 2013.

[43] L. Hanlen, D. Miniutti, D. Smith, D. Rodda, and B. Gilbert, “Co-channel interference in body area networks with indoor measure-ments at 2.4 GHz: Distance-to-interferer is a poor estimate ofreceived interference power,” Int. J. Wireless Inf. Networks, vol. 17,no. 3-4, pp. 113–125, 2010.

[44] S. Misra, S. Moulik, and H. C. Chao, “A cooperative bargainingsolution for priority-based data-rate tuning in a wireless body areanetwork,” IEEE Trans. Wireless Commun., vol. 14, no. 5, pp. 2769–2777, 2015.

[45] S. Moulik, S. Misra, C. Chakraborty, and M. S. Obaidat, “Prioritizedpayload tuning mechanism for wireless body area network-basedhealthcare systems,” in Proc. IEEE GLOBECOM. San Diego, CA,USA, 2015, pp. 2393–2398.

[46] X. Yuan, C. Li, L. Yang, W. Yue, B. Zhang, and S. Ullah, “Atoken-based dynamic scheduled mac protocol for health monitoring,”EURASIP J. Wirel. Comm., vol. 2016, no. 1, pp. 125–143, 2016.

[47] T. L. Saaty, “Decision making with analytic hierarchy process,”International Journal of Services Sciences, vol. 1, no. 1, pp. 83–98, 2008.

[48] S. Roy, I. Mallik, and S. Moulik, “PAG-MAC: Prioritized allocationof GTSs in IEEE 802.15.4 MAC protocol-a dynamic approach basedon analytic hierarchy process,” in Proc. IEEE INDICON. IITRoorkee, India, Dec. 2017, pp. 1–5.

[49] S. Rashwand and J. Misic, “Effects of access phases lengths onperformance of IEEE 802.15. 6 CSMA/CA,” Comput. Netw., vol. 56,no. 12, pp. 2832–2846, 2012.

Xiaoming Yuan (S’17) received her B.E. degreein Electronics and Information Engineering fromHenan Polytechnic University, China, in 2012.She received the Ph.D. degree in Communica-tion and Information System from Xidian Uni-versity, China, in 2018. From September 2016to August 2017, she was a Visiting Scholarwith the Broadband Communications Research(BBCR) Group, Department of Electrical andComputer Engineering, University of Waterloo,Waterloo, ON, Canada. Her research interests

include cloud/edge computing, medium access control and performanceanalysis for wireless body area networks and the Internet of Things.

Chang Le (M’09-SM’16) received the Ph.D.degree in Communication and Information Sys-tem from Xidian University, China, in 2005. Heconducted his postdoctoral research in Canadaand the National Institute of information andCommunications Technology, Japan, respective-ly. He had been a Visiting Scholar with theUniversity of Technology Sydney and is current-ly a Professor with the State Key Laboratoryof Integrated Services Networks, Xidian Uni-versity. His research interests include intelligent

transportation systems, vehicular networks, mobile ad hoc networks, andwireless sensor networks.



16

Qiang Ye (S’16-M’17) received his B.S. degreein network engineering and his M.S. degree incommunication and information systems fromthe Nanjing University of Posts and Telecom-munications, China, in 2009 and 2012, respec-tively, and his Ph.D. degree in electrical andcomputer engineering from the University ofWaterloo, Ontario, Canada, in 2016. He has beena postdoctoral fellow with the Department ofElectrical and Computer Engineering, Universityof Waterloo, since 2016. His current research

interests include software-defined networking and network function virtu-alization, network slicing for 5G networks, virtual network function chainembedding and end-to-end performance analysis, medium access controland performance optimization for mobile ad hoc networks, and the Internetof Things.

Kuan Zhang (S’13-M’17) has been an assistantprofessor at the Department of Electrical andComputer Engineering, University of Nebraska-Lincoln, USA, since September 2017. He re-ceived the B.Sc. degree in Communication En-gineering and the M.Sc. degree in ComputerApplied Technology from Northeastern Univer-sity, China, in 2009 and 2011, respectively.He received the Ph.D. degree in Electrical andComputer Engineering from the University ofWaterloo, Canada, in 2016. He was also a post-

doctoral fellow with the Broadband Communications Research (BBCR)group, Department of Electrical and Computer Engineering, University ofWaterloo, Canada, from 2016-2017. His research interests include securityand privacy for mobile social networks, e-healthcare systems, cloud/edgecomputing and cyber physical systems.

Nan Cheng (S’12-M’16) received the Ph.D.degree from the Department of Electrical andComputer Engineering, University of Waterlooin 2015, and B.E. degree and the M.S. degreefrom the Department of Electronics and Infor-mation Engineering, Tongji University, Shang-hai, China, in 2009 and 2012, respectively. Heis currently working as a postdoctoral fellowwith the Department of Electrical and ComputerEngineering, University of Toronto and Depart-ment of Electrical and Computer Engineering,

University of Waterloo under the supervision of Prof. Ben Liang andProf. Sherman (Xuemin) Shen. His current research focuses on big datain vehicular networks and self-driving system. His research interests alsoinclude performance analysis, MAC, opportunistic communication, andapplication of AI for vehicular networks.

Ning Zhang (M’15) is an Assistant Professorat Texas A&M University-Corpus Christi, USA.He received the Ph.D degree from Universityof Waterloo, Canada, in 2015. After that, hewas a postdoctoral research fellow at Universityof Waterloo and University of Toronto, Canada,respectively. He serves/served as an associateeditor of IEEE Access, IET Communication,and Journal of Advanced Transportation; an areaeditor of Encyclopedia of Wireless Networks(Springer) and Cambridge Scholars; and a guest

editor of Wireless Communication and Mobile Computing, InternationalJournal of Distributed Sensor Networks, and Mobile Information System.He also served as the workshop chair for the first IEEE Workshop onCooperative Edge and IEEE Workshop on Mobile Edge Networks andSystems for Immersive Computing and IoT. He is a recipient of theBest Paper Awards at IEEE Globecom 2014 and IEEE WCSP 2015,respectively. His current research interests include next generation mobilenetworks, physical layer security, machine learning, and mobile edgecomputing.

Dr. Xuemin (Sherman) Shen (M’97-SM’02-F’09) received the B.Sc.(1982) degree fromDalian Maritime University (China) and theM.Sc. (1987) and Ph.D. degrees (1990) fromRutgers University, New Jersey (USA), all inelectrical engineering. He is a University Pro-fessor and the Associate Chair for GraduateStudies, Department of Electrical and ComputerEngineering, University of Waterloo, Canada.Dr. Shens research focuses on wireless resourcemanagement, wireless network security, social

networks, smart grid, and vehicular ad hoc and sensor networks. Heis the elected IEEE ComSoc VP Publication, was a member of IEEEComSoc Board of Governor, and the Chair of Distinguished LecturersSelection Committee. Dr. Shen served as the Technical Program Commit-tee Chair/Co-Chair for IEEE Globecom16, Infocom14, IEEE VTC10 Fall,and Globecom07, the Symposia Chair for IEEE ICC10, the Tutorial Chairfor IEEE VTC11 Spring and IEEE ICC08, the General Co-Chair for ACMMobihoc15, Chinacom07 and QShine06, the Chair for IEEE Communica-tions Society Technical Committee on Wireless Communications, and P2PCommunications and Networking. He also serves/served as the Editor-in-Chief for IEEE Internet of Things Journal, IEEE Network, Peer-to-Peer Networking and Application, and IET Communications; a FoundingArea Editor for IEEE Transactions on Wireless Communications; andan Associate Editor for IEEE Transactions on Vehicular Technologyand IEEE Wireless Communications, etc. Dr. Shen received the IEEEComSoc Education Award, the Joseph LoCicero Award for ExemplaryService to Publications, the Excellent Graduate Supervision Award in2006, and the Premiers Research Excellence Award (PREA) in 2003 fromthe Province of Ontario, Canada. Dr. Shen is a registered ProfessionalEngineer of Ontario, Canada, an IEEE Fellow, an Engineering Instituteof Canada Fellow, a Canadian Academy of Engineering Fellow, a RoyalSociety of Canada Fellow, and a Distinguished Lecturer of IEEE VehicularTechnology Society and Communications Society.

Performance Analysis of IEEE 802.15.6-Based Coexisting ... · Carrier Sense Multiple Access with...

Documents

Transcript of Performance Analysis of IEEE 802.15.6-Based Coexisting ... · Carrier Sense Multiple Access with...