PERFORMANCE

OF SENSOR MAC PROTOCOLS FOR MEDICAL

ICT USING IR-UWB TECHNOLOGY

Flavia Martelli

Wilab, CNIT, Universit

a di Bologna, V.le Risorgimento 2, 40136 Bologna, Italy

Leonardo Goratti, Jussi Haapola

CWC, University of Oulu, Finland, P.O. Box 4500 FIN-90014, University of Oulu, Finland

Keywords:

Medium Access Control, Ultra wideband, Capture effect, Medical applications, IEEE 802.15.4a.

Abstract:

In this paper the feasibility of contention-based medium access control (MAC) protocols using impulse radio

technology for medical ICT scenarios is shown. The particular scenario refers to a hospital waiting room in

which people enter wearing a number of sensors for continuous monitoring of health status. The evaluation

of the feasibility is founded on the implementation of slotted Aloha (S-Aloha) and preamble sense multiple

access (PSMA) MAC protocols, the physical layer characteristics and the non-coherent receiver scheme for

impulse radio ultra wideband (IR-UWB) in the network simulator Opnet. Simulated throughput and delay

performance under Poisson trafﬁc assumption are compared with analytical results for validation purposes.

The simulation of the medical scenario, not easily tractable by analysis, accounts for important characteristics

like capture effect and different bit rates of the sensors that monitor various vital functions. The results show

that the used MAC protocols are scalable within the scenario constraints and PSMA exhibits a better delay

performance than S-Aloha.

1 INTRODUCTION

Medical information and communication technology

(ICT) applications have been gaining increasing im-

portance in recent years (Arnon et al., 2003; Istepa-

nian et al., 2004). The need of continuous monitor-

ing of ill or elderly people is increasing, together with

the requirement of low costs for healthcare systems.

Wireless technologies enable non-invasive and ubiq-

uitous solutions, but the choice of the proper physical

layer (PHY) is required.

Due to its low power spectral density, ultra wide-

band (UWB) is suitable for transmissions near to

the human body in the hospital environment, with-

out causing harmful impact to patients or interfering

with other medical equipment. Therefore, UWB is a

promising PHY technology for these kinds of appli-

cations (H

ainen et al., 2008).

UWB received great attention from academy and

industry in latter years, though this technology was

already used in military applications. The develop-

ment of new types of wireless networks, as wireless

personal area networks (WPANs), led to the possibil-

ity of exploiting UWB also for civilian applications,

especially after 2002, when the Federal Communica-

tions Commission (FCC) gave the ﬁrst regulatory on

the power emitted by hand-held devices using UWB

transmissions (FCC, 2002).

Among the different possible technical imple-

mentations for UWB, only impulse radio UWB (IR-

UWB) is considered in this work. The IR-UWB

offers a high time resolution of multipath compo-

nents, making it an appealing technology for in-

door applications, and it has been chosen as al-

ternative PHY for the IEEE 802.15.4a Amendment

(IEEE-802.15.4a, 2007) of the IEEE 802.15.4 Stan-

dard (IEEE-802.15.4, 2006). The IEEE 802.15.4 fo-

cuses on low-rate WPANs (LR-WPANs), character-

ized by low costs and low power consumption.

According to (IEEE-802.15.4a, 2007), both coher-

ent and non-coherent receiver solutions are possible.

Even if the coherent approach can lead to optimal per-

formance, we consider a simple non-coherent receiver

based on energy detection (ED) since our target is an

extremely simple, cheap, though efﬁcient implemen-

tation solution for sensor devices.

Martelli F., Goratti L. and Haapola J. (2010).

PERFORMANCE OF SENSOR MAC PROTOCOLS FOR MEDICAL ICT USING IR-UWB TECHNOLOGY.

In Proceedings of the Third International Conference on Health Informatics, pages 13-20

DOI: 10.5220/0002691300130020

 SciTePress

Figure 1: Reference architecture of the energy detection re-

ceiver.

In this paper we focus on evaluating a medical

ICT scenario through simulations using multiple sen-

sors with different service requirements. First, the

IR-UWB PHY is modeled to a degree where the pe-

culiarities of the technology are captured in sufﬁcient

detail. Second, the model is then fed to the Opnet

network simulator and it is included in the analytical

framework developed in (Haapola et al., 2009). Third,

two different MAC protocols, tailored for IR-UWB,

are implemented in the simulator: slotted Aloha (S-

Aloha) and preamble sense multiple access (PSMA).

The simulation results are compared with the ones of

the analytical framework. Finally, the medical ICT

scenario is developed and simulated.

The paper is organised as follows. Section 2 de-

scribes the characteristics of IR-UWB physical layer,

whereas Section 3 illustrates the S-Aloha and the

PSMA MAC protocols. The simulation scenario and

the applications are presented in Section 4. Perfor-

mance evaluation in terms of throughput and delay in

the described scenario is carried out in Section 5, and

Section 6 concludes the paper.

2 IR-UWB PHYSICAL LAYER

IR-UWB is based on the transmission of a series

of, temporally extremely short, pulses with very low

emitted power. The main features of UWB are noise-

like power emission and very accurate time resolution

of multipath components. The basics of IR-UWB sig-

nal are described in the pioneer work of (Win and

Scholtz, 1998). Due to its carrier-less nature, IR-

UWB enables simple transceiver implementations. In

this work we refer to the non-coherent receiver based

on energy detection (ED) shown in Fig. 1. The en-

ergy of the received signal, r(t), which arrives at the

receiving antenna, is ﬁltered by the bandpass zonal

ﬁlter (BPZF) and integrated for a certain time dura-

tion, T

int

, and based on that the decision variable, Y ,

is made. The amount of integrated energy depends

on the integration time, which has to set a trade-off

between integration of noise and of the useful signal.

The ED approach is simple and has a low power con-

sumption, but it can be strongly inﬂuenced by mul-

tiuser interference.

IR-UWB uses several modulation schemes for

transmission. One of the most used is the pulse posi-

tion modulation (PPM). The IEEE 802.15.4a deﬁnes

a variation of the standard PPM that is summarized

in the next section. Furthermore, IR-UWB uses time

hopping (TH) codes to address the problem of multi-

user capability.

2.1 IEEE 802.15.4a

The IEEE 802.15.4a (IEEE-802.15.4a, 2007) is

an amendment to IEEE 802.15.4 Standard (IEEE-

802.15.4, 2006), where two alternative physical lay-

ers (PHYs) are deﬁned to add location capability and

to improve performance in terms of rates, range, and

power consumption. In this paper we will consider

only the UWB PHY. According to the amendment,

data is transmitted as a burst of pulses with Burst Po-

sition Modulation (BPM). BPM modulation scheme

is similar to the PPM but with the difference that the

symbol interval is divided into two slots. One slot

is for bit zero and the other one is for bit one. The

preamble sequence, added before the physical service

data unit (PSDU), provides packet synchronization

and enables channel estimation. The preamble uses

Ternary sequences with perfect periodic autocorrela-

tion properties.

2.2 False Alarm and Miss-detection

A device assessing the status of the channel

(idle/busy) can encounter into two erroneous events:

false alarm and miss-detection. Let H

be the statis-

tical hypothesis of no preamble in the channel, and

let H

be the hypothesis of presence of a preamble.

The false alarm refers to the event that the pream-

ble is found, under hypothesis H

, and the probabil-

ity of this event is indicated with P

. Miss-detection

refers to the event in which the preamble is missed

under hypothesis H

and the probability of this event

is indicated with P

. Both events can affect MAC

performance (Ramachandran and Roy., 2006), when

performing clear channel assessment (CCA) before

transmission or in the reception of a packet.

Considering an additive white Gaussian noise

(AWGN) channel and conditioning on a multipath re-

alization, the decision variable Y after the integrator

follows a Chi-square distribution with k = L

2q de-

grees of freedom, where L

is the number of symbols

in a preamble, N

is the number of pulses in a sym-

bol, and q is the time-bandwidth product. Under hy-

pothesis H

the decision variable is central Chi-square

distributed, while under hypothesis H

it has a non-

centrality parameter given by λ = L

SNR, with SNR

being the received Signal to Noise Ratio. When k is

HEALTHINF 2010 - International Conference on Health Informatics

large enough, the Chi-square distribution can be ap-

proximated by a Gaussian distribution, and P

and

can be expressed as:

= Pr{Y > ε|H

} = Q

, (1)

= Pr{Y < ε|H

} = 1 −Q

ε − L

SNR

4q + 4L

SNR

where ε is the threshold of energy on the channel for

deciding the presence of a preamble.

2.3 Capture Effect for IR-UWB

Techology

MAC protocol performance is classically analyzed by

assuming that when a collision occurs none of the

packets involved can be successfully received. The

phenomenon of one packet prevailing at the receiver

with respect to others is well known in the literature

as capture effect. For narrowband systems, several

works (e.g. (Zhang and Pahlavan, 1992)) have shown

how taking this realistic effect into account affects

MAC performance. The capture probability, P

, can

be generally deﬁned for different n (number of pack-

ets arriving at the receiver) as

(n) =







1, if n = 1

0, if n → ∞

∈ (0, 1), otherwise.

(2)

For IR-UWB systems, a capture effect model has

to consider how the pulses of the transmitted pack-

ets overlap in time as well as the received powers.

By assuming the point of view of a reference packet,

in the ED receiver, packets coming from other trans-

mitters will interfere with the reference transmission

if their pulses fall into the integration time of the re-

ceiver. Simultaneous transmissions do not necessary

lead to complete information loss. This depends on

the amount of integrated energy in both BPM po-

sitions corresponding to different bits. In the cur-

rent study the capture effect in the presence of intra-

piconet interference is considered, assuming that all

the users use the same preamble sequence and the

same TH code for the BPM transmission, and the data

pulse structure of the IEEE 802.15.4a. To simplify the

problem, the capture effect is considered only in the

presence of two simultaneous transmissions, such that

(n) =







= 1 −P

, if n = 1

∈ (0, 1), n = 2

0, if n > 2.

(3)

With n = 1 the probability of capture is just the prob-

ability of detecting the packet P

. The exact value of

(2) depends on the bit error probability (BEP) of

the channel:

BEP = Q

2(E

−E

)

4(E

+ E

) +4qN

, (4)

with E

and E

the integrated energy of the reference

and the interfering signal, respectively, and N

the

power spectral density of the thermal noise. The fully

analytical model is given in (Haapola et al., 2009).

According to (IEEE-802.15.4a, 2007), we con-

sider a Reed-Solomon RS

(63, 55) code, with a cor-

rection capability of t = 4 symbols. Let L

be the

length of a coded packet, and b

= 63 symbols the

length of a block of the coded packet. The probability

of accepting one of the x = d

e blocks of the packet

is:

∑

i=0

SEP

(1 −SEP)

−i

, (5)

where SEP = 1 − (1 − BEP)

is the symbol error

probability. A packet is correct if all the blocks are

correct, therefore P

(2) = (P

)

3 MAC PROTOCOLS

According to (IEEE-802.15.4, 2006), a WPAN shall

always include a coordinator, which is responsible

of setting up and maintaining the WPAN. Among

its tasks, the coordinator controls the access to the

medium, choosing between a beacon-enabled mode

or a non beacon-enabled mode. In the beacon-enabled

mode, the coordinator broadcasts a periodic beacon

frame, containing information about the WPAN. The

period between two consecutive beacons deﬁnes a su-

perframe structure. The superframe consists of an ac-

tive and an optional inactive period; devices commu-

nicate only during the active period and should enter

a low-power mode during the inactive one. The active

part can be further split into a Contention Access Pe-

riod (CAP), and an optional Contention Free Period

(CFP). During the CAP data is transferred using Car-

rier Sense Multiple Access with Collision Avoidance

(CSMA-CA) algorithm, while the CFP is composed

of reserved slots, the Guaranteed Time Slots (GTSs),

that the coordinator can assign to devices running ap-

plications with quality of service (QoS) constraints.

According to the CSMA-CA, when a device has

data to send, it performs CCA after a random initial

backoff delay. If the channel is found idle during two

consecutive CCAs, the device transmits the data; oth-

erwise it will repeat the procedure after another ran-

dom backoff obeying the binary exponential backoff

PERFORMANCE OF SENSOR MAC PROTOCOLS FOR MEDICAL ICT USING IR-UWB TECHNOLOGY

(BEB) rules. Full description of the algorithm can be

found from (IEEE-802.15.4, 2006).

Since the IR-UWB is a carrier-less technology,

traditional CCA methods valid for narrowband sys-

tems are not suitable. The IEEE 802.15.4a (IEEE-

802.15.4a, 2007) introduces Aloha as default chan-

nel access strategy, because it is suitable for lightly

loaded networks where the probability of collision is

reasonably small. Given the high processing gain of

the IR-UWB the transmission can have high interfer-

ence resilience. Alternative CCA methods based on

the sensing of the preamble are deﬁned. The standard

deﬁnes an optional modiﬁed frame structure, in which

preamble symbols are multiplexed with data symbols

in the frame. The multiplexed preamble enables CCA

at any time during frame transmission, but at a cost of

signiﬁcant overhead and according to (Haapola et al.,

2009) it attains the worst performance for the trafﬁc

loads targeted in this paper and it will not be consid-

ered further.

In this paper we will investigate slotted Aloha (S-

Aloha) and preamble sense multiple access (PSMA)

(Haapola et al., 2006), which proposes an alterna-

tive channel access method to utilise the CCA in

IR-UWB environment and it is compatible with the

IEEE 802.15.4a PHY. We will consider a typical IEEE

802.15.4 star topology network in beacon-enabled

mode. The access protocols analysed will thus refer

to the CAP of the superframe; no GTSs are present.

The analysis of the average throughput is carried

out using renewal theory, with the classical formula

S ,

B +I

, (6)

where U is the average useful period, B is the average

busy period, I is the average idle period, and B + I is

referred to as the average cycle. For the theoretical

analysis and for the validation of the simulator, the

overall trafﬁc is considered Poisson distributed with

mean arrival rate g. The probability of generating n

packets in a time slot T

can be expressed as

p(n) =

(gT

)

−gT

. (7)

The duration of T

is protocol dependent. The

throughput is studied as a function of the normalized

offered trafﬁc G, with normalization interval related

to T

and thus protocol dependent.

3.1 S-Aloha

In a beacon-enabled network the S-Aloha protocol

can be implemented as the synchronization is pro-

vided by the coordinator. The events of false alarm

and miss-detection as described in section 2.2 for the

−1

0.05

0.1

0.15

0.2

0.25

0.3

G [Erlang]

(S−ALOHA) Normalized Throughput, n = 80 nodes

Backoff delay = random[0, 31] slots (without capture)

Backoff delay = random[0, 31] slots (with capture)

Backoff delay = random[0, 3] slots (without capture)

Backoff delay = random[0, 3] slots (with capture)

Poisson analysis (without capture)

Poisson analysis (with capture)

Figure 2: S-Aloha normalized throughput as a function of

the normalized offered trafﬁc G = gT

−1

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18

0.2

G [Erlang]

Delay [sec]

(S−ALOHA) Delay, n = 80 nodes

Backoff delay = random[0, 31] slots (without capture)

Backoff delay = random[0, 31] slots (with capture)

Backoff delay = random[0, 3] slots (without capture)

Backoff delay = random[0, 3] slots (with capture)

Analysis, Backoff delay = random[0, 31] slots

Figure 3: S-Aloha delay as a function of the normalized

offered trafﬁc G = gT

CCA do not have a direct impact on the B for S-Aloha,

because the status of the channel is not assessed be-

fore the transmission of a packet. However, the P

affects also the reception of a frame: the receiver can

actually receive a packet only after detecting the cor-

responding preamble. Considering this, the useful pe-

riod can be expressed as U = BP

, where P

rep-

resents the probability of success. We take into ac-

count the overhead introduced by the standard, which

yields a decrease of 26% with respect to the maximum

throughput achievable for S-Aloha (0.368).

When capture effect is taken into account, the use-

ful period can be calculated as follows:

U = B

∞

∑

n=1

(n)

p(n) = B

∑

n=1

(n)

(gT

)

−gT

(8)

where P

is the probability to be in a busy period.

According to the Aloha protocol, in case of colli-

sion the packet is retransmitted after a random delay.

We have considered a backoff delay uniformly dis-

tributed between 0 and N time slots. The choice of N

HEALTHINF 2010 - International Conference on Health Informatics

has a strong impact on the performance. The lower

is N, the lower the throughput will be, since when a

collision occurs, two or more nodes will schedule a

retransmission within a shorter ﬁxed period, leading

to possible repeated collisions. On the other hand, a

high value of N means higher delay in the reception

of the packets. This effect can be seen from Fig. 2

and 3, where throughput and delay for S-Aloha are

shown. Simulations were run with Opnet Modeler,

considering 80 nodes, which proved to be a good ap-

proximation of the inﬁnite case. Results for N = 3

and N = 31 illustrate how both throughput and de-

lay are higher for the higher value of N. Through-

put obtained for N = 31 is very close to the classi-

cal inﬁnite population Poisson analysis. The dotted

curves are obtained considering the capture effect on

the channel, and show an improvement for both back-

off windows. The analytical curve of throughput with

capture has been obtained considering the interferer

relative distance equal to 2.02 meters and SINR of

2 dB. The matching of simulation results and theory

depends in fact on the relative distance between the

reference user and the interferer, since different rela-

tive distances cause different overlapping during the

receiver integration window leading to different per-

formances.

3.2 Preamble Sense Multiple Access

The PSMA protocol has been described and analyzed

by (Haapola et al., 2009; Haapola et al., 2006). When

a node has a data frame to transmit, it chooses an

initial random backoff according to the BEB value

range. When the backoff timer expires, the node per-

forms a CCA in terms of a preamble detection at the

beginning of a backoff boundary. If no preamble is

detected, the node transmits at the beginning of the

next backoff boundary with a preamble sequence fol-

lowed by the data. The data transmission and the

corresponding acknowledgment must be completed

within two time slots. If the CCA indicated “channel

busy”, the node performs a backoff according to the

BEB rules and re-attempts. The mechanism ensures

that once a transmission has started, it can continue

for two consecutive time slots without a collision. A

collision can occur only if two or more nodes per-

form a “channel free” CCA simultaneously. With the

above characteristics, PSMA classical, average Idle,

Busy, and Useful periods are

I =

1 −e

−gT

, B =

−gT

, (9)

U =

, P

−gT

1 −e

−gT

−1

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

Simulation throughput (No capture effect)

Simulation throughput (With capture effect)

Analysis throughput

−1

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

G [Erlang]

Delay [sec]

(PSMA) Throughput and Delay, n = 80 nodes

Simulation delay (No capture effect)

Simulation delay (With capture effect)

Analysis delay

Figure 4: PSMA normalized throughput and delay as a

function of the normalized offered trafﬁc G = 2gT

where P

is the probability of success and C represents

the fraction of useful data due to protocol overhead.

As the effects of BEB, P

, and P

are taken into

account, the following consequences can be perceived

for the classical throughput equations of Eq. (9). The

has no effect on the length of the idle period, but

the P

may increases it. If only one node performs

CCA during a slot, with a constant probability P

performs a backoff. When the BEB is taken into ac-

count, the P

can further increase the Idle period.

In the Busy period P

can increase the B if a node

performs the CCA during the ﬁrst slot of an existing

transmission and has a miss-detection. The B is ex-

tended by a T

for every P

, and it occurs if an arrival,

with proper scheduling in the next, 2nd, 3rd, and 4th

slot following the initial transmission, occurs. Multi-

ple P

are required to extend B by more than one T

With P

, a node performs a backoff, which reduces

the length of the busy period.

For the Useful period, the P

only affects the P

of Eq. (9), and the P

∝ P

. There can be a success-

ful transmission if and only if all but one of multile

CCAs have a P

in a slot. The mathematical deriva-

tion of the above has been carried out by (Haapola

et al., 2009).

Fig. 4 illustrates PSMA throughput and delay re-

sults. A good matching is shown between simulation

and the analysis described in (Haapola et al., 2009).

The dotted curves show the improvement obtained

when considering the capture effect. For offered traf-

ﬁc values up to 0.6 the performance of S-Aloha and

PSMA in terms of throughput are equivalent. For

higher values PSMA performs better. The delay of

PSMA is always lower than that of S-Aloha.

PERFORMANCE OF SENSOR MAC PROTOCOLS FOR MEDICAL ICT USING IR-UWB TECHNOLOGY

Table 1: Parameters of biomedical applications (Arnon

et al., 2003).

Biomedical

measure-

ments

Sample

rate (sam-

ples/s)

Resolution

(bits/ sam-

ple)

Information

rate (bit/s)

ECG 1250 12 15000

Heart rate 25 12 600

EEG 350 24 4200

Respiratory

rate

50 16 800

Temperature

of body

5 16 80

4 SIMULATION SCENARIO FOR

VITAL SIGNALS MONITORING

Aging of the population determines the need for re-

ducing the costs and the number of the hospitaliza-

tions that will otherwise burden the healthcare sys-

tems. On the contrary, there is an increased de-

mand for more accurate monitoring of vital signals

of a large variety of customers (e.g., in sports and

for chronically diseased people). Those two opposite

trends pose the demand for new technologies able to

fulﬁll both scenarios. Wireless technology can have

the requisites of low costs, ease of use and ease of re-

placement, without being invasive. Energy consump-

tion and reliability of the transmissions are the main

issues.

The target scenario is given by a number of peo-

ple wearing wireless sensors deployed on their body,

thus forming a WPAN. The IR-UWB technology suits

particularly well for transmissions near the body be-

cause of the low emitted power. The scenario under

investigation resembles at home or inside the hospital

monitoring of the vital signals summarized in Table 1

(Arnon et al., 2003). The study considers the aggre-

gated trafﬁc coming from different sensors that mon-

itor the same set of parameters. For each of the mon-

itored parameters samples are collected at a sampling

frequency of ﬁve times higher than the maximum fre-

quency (Arnon et al., 2003). Depending on the res-

olution, for every parameter the information rate is

obtained as: Information rate = Sample rate · Reso-

lution. Furthermore, a deterministic trafﬁc pattern is

assumed such that every second the aggregated traf-

ﬁc for each of the monitored parameters is generated.

This way of modeling the application has signiﬁcance

for example when patients need to be constantly mon-

itored. The MAC protocol encapsulates the gener-

ated trafﬁc into PHY protocol data units (PPDU), also

link-level fragmentation is used when needed.

This study assumes the beacon-enabled star topol-

ogy deﬁned in (IEEE-802.15.4, 2006) and the UWB

PHY of IEEE 802.15.4a, focusing on the performance

inside the WPAN. The coordinator, that maintains

the synchronization issuing the beacon, can be mains

powered. The wireless sensors transmit data to the co-

ordinator (multipoint-to-point communications) and

they are battery operated. The coordinator can act as

a gateway to connect to the Internet or the backbone

network. The scenario under investigation is a ward

of 10x10 m in which patients wear wireless sensors

and communication is initiated in a serialized man-

ner. Realistically, a number of patients up to ten has

been considered. As the number of patients increases,

the trafﬁc contending for the channel correspondingly

increases. The scenario is assumed to be realistic also

because the IR-UWB signal is not likely to interfere

with other adjacent rooms due to absorption and sig-

nal attenuation through the walls. This can be con-

cluded by considering the wavelength (smaller than

10 cm) of the UWB transmission and the wall atten-

uation as well as intentional lack of synchronization

between adjacent wards. Moreover, the use of differ-

ent preamble sequences (c.f. (IEEE-802.15.4a, 2007))

in adjacent rooms decreases the interference even fur-

ther.

Given that the reliability of the data is crucial

when monitoring vital signals the metrics under in-

vestigation are the throughput and the delay. Energy

consumption is essential but the main target of this

work is to prove that the use of IR-UWB technology

and random access can provide sufﬁcient data relia-

bility in realistic conditions. Random access is in fact

the default access in the IEEE Std 802.15.4. However,

as it is shown in (Haapola et al., 2009) the energy

consumption for PSMA per useful transmitted bit is

less than 100 nJ up to the channel capacity of offered

trafﬁc. The average throughput and delay are investi-

gated in a scenario with an increasing number of sen-

sors and including realistic capture effect at receiver

side. The monitored data are sent to the coordinator,

which acknowledges the correct reception of a PPDU

packet. In the case of large payloads, as for the ECG,

several PPDUs have to be sent to the coordinator, but

in the case that even a single PPDU is lost the entire

message is discarded.

5 SIMULATION RESULTS

MAC and PHY parameters assumed in the paper are

summarized in Table 2.

The transmitted power of 0.37 µW is compliant

HEALTHINF 2010 - International Conference on Health Informatics

Table 2: IEEE 802.15.4a PHY and MAC parameters.

Parameter Value

Max data MPDU length 127 bytes

Preamble length 64 preamble

symbols

Max ack waiting time 140 bits

Slot duration 1426 bits

PHY Bit rate 850 kbit/s

Center frequency 4492.8 MHz

Bandwidth 499.2 MHz

Transmitted power 0.37 µW

Transmitter power consumption 20 mW

Receiver power consumption 120 mW

Sleep power consumption 0.2 mW

with FCC ruling (FCC, 2002). The power consump-

tion values are taken from (Stoica et al., 2005).

The metrics under investigation, the throughput S,

in terms of bits per second correctly received, and the

average delay D, can be expressed as follows:

( j)

Bits

Sim

, D

( j)

∑

( j)

i=1

− T

Enq

)

( j)

, (10)

where the index j refers to the different applications,

( j)

is the total number of application messages re-

ceived per application, N

( j)

Bits

represents the message

size in bits, T

Sim

is the duration of a simulation, T

indicates the instant at which the acknowledgment for

the last packet of a message is received, and T

Enq

the time when the application message is enqueued at

the MAC layer. These metrics have been studied as a

function of the number of active nodes in the scenario.

Fig. 5 shows throughput results for applications

with higher data rates. The PSMA and the S-Aloha

achieve the same performance in the simulated sce-

nario when the capture effect is considered. The

PSMA curve without capture effect is not plotted be-

cause it practically overlaps with the one obtained

considering capture. For S-Aloha, instead, an im-

provement can be seen when the capture effect is in-

troduced in case of ECG and EEG. All curves exhibit

a linearly increasing behaviour as the number of pa-

tients increases. The behavior is due to the fact that

for the simulated trafﬁc the protocol is working in the

stability region and below the channel capacity.

The delay results are shown in Fig. 6 for the S-

Aloha and in Fig. 7 for the PSMA. For both of

the protocols, ECG and EEG achieve the highest de-

lay values, and this is due to their higher information

rate, which yields a higher number of MAC packets to

5 10 15 20 25 30 35 40 45 50

100

150

Number of active nodes

S [kbps]

Medical applications Throughput

ECG S−Aloha

ECG S−Aloha and PSMA (with capture)

EEG S−Aloha

EEG S−Aloha and PSMA (with capture)

Respiratory Rate S−Aloha and PSMA

Figure 5: Throughput for medical applications.

5 10 15 20 25 30 35 40 45 50

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18

0.2

Number of active nodes

Delay[sec]

(S−ALOHA) Medical applications Delay

ECG

ECG (capture)

EEG

EEG (capture)

Temperature

Temperature (capture)

Respiratory Rate

Respiratory Rate (capture)

Heart Rate

Heart Rate (capture)

Figure 6: S-Aloha delay for medical applications.

5 10 15 20 25 30 35 40 45 50

0.02

0.04

0.06

0.08

0.1

0.12

Number of active nodes

Delay[sec]

(PSMA) Medical applications Delay

ECG

ECG (capture)

EEG

EEG (capture)

Temperature

Temperature (capture)

Respiratory Rate

Respiratory Rate (capture)

Heart Rate

Heart Rate (capture)

Figure 7: PSMA delay for medical applications.

be transmitted. The other three applications achieve

lower delay, with very similar behavior, since their

messages ﬁt into one MAC frame. The impact of

the capture effect on the delay is quite relevant. The

PSMA delay appears to be lower than that of S-Aloha

for all applications. This concludes that the PSMA

performs better also in this case. Moreover, for the

PSMA the delay stays almost constant and predictable

while the number of active nodes are varied. We have

not found a suitable comparison from the literature.

The values obtained for the delay stay under the 125

PERFORMANCE OF SENSOR MAC PROTOCOLS FOR MEDICAL ICT USING IR-UWB TECHNOLOGY

ms threshold indicated in (Zhen et al., 2008) as a

generic maximum latency for medical applications in

body area networks (BANs), showing the feasibility

of the protocols under study in the scenario. In partic-

ular, a constant delay is an important feature of a pro-

tocol to use for transmitting health related data. The

proposed PHY/MAC solution, even in the presence

of interference (more patients in the room), exhibits

a stable behaviour, hence guaranteeing the mandatory

reliability of the transmitted information that doctors

use for diagnosis. Clearly, further work is required

to evaluate not only the average delay but also the jit-

ter, to meet the requirements of some critical real-time

application.

6 CONCLUSIONS

This paper has investigated average throughput and

delay performance of contention-based MAC proto-

cols in a realistic medical ICT scenario when consid-

ering different applications used for health monitoring

as: ECG, EEG, heart rate, respiratory rate and tem-

perature of the body. The performance has been eval-

uated using the network simulator Opnet, after vali-

dation with theoretical analysis.

The scenario addresses a wireless sensor network

for vital signal monitoring, in which the physical layer

technology is IR-UWB, as deﬁned in IEEE 802.15.4a

standard. The network architecture is the star topol-

ogy deﬁned by the IEEE 802.15.4 standard. Given

that the study targets cheap sensors implementations,

the considered receiver scheme is the non-coherent

energy detector. The MAC protocols investigated are

S-Aloha and PSMA, whose performance has been ob-

tained by varying the number of nodes in the scenario.

The events of false alarm and miss detection in the

sensing of the preamble, and the possibility of the re-

ceiver capturing a packet subjected to collision are the

realistic effects that have been included in both anal-

ysis and simulations.

The results show good performance in terms of

throughput for both protocols, but for the delay,

PSMA performs better. In addition, the results en-

lighten the importance of accounting for realistic

PHY effects and accurate modelling of the applica-

tion. The study showed that the average delay is be-

low the indicative maximum latency of 125 ms and

that the capture improves the performances. Given

the lack of explicit reference delay for these kinds of

applications, we can conclude that in a realistic envi-

ronment the network architecture used and the MAC

protocols investigated satisfy the requirements for en-

abling the concept of wireless health monitoring.

REFERENCES

Arnon, S., Bhastekar, D., Kedar, D., and Tauber, A. (2003).

A comparative study of wireless communication net-

work conﬁgurations for medical applications. IEEE

Wireless Communications.

FCC (2002). Revision of part 15: First report and order.

Recommendation, Federal Communications Commis-

sion.

Haapola, J., Goratti, L., Oppermann, I., and Rabbachin, A.

(2006). Preamble sense multiple access (psma) for im-

pulse radio ultra wideband sensor networks. In Em-

bedded Computer Systems: Architectures, Modeling,

and Simulation. Springer Berlin / Heidelberg.

Haapola, J., Rabbachin, A., Goratti, L., Pomalaza-R

aez, C.,

and Oppermann, I. (2009). Effect of impulse radio-

ultra wideband based on energy collection on mac

protocol performance. IEEE Transaction on Vehicu-

lar Technology, To appear, 58(9).

ainen, M., Pirinen, P., Iinatti, J., and

Taparugssanagorn, A. (2008). Uwb supporting

medical ict applications. In IEEE International

Conference on Ultra-Wideband (ICUWB 2008).

IEEE-802.15.4 (2006). Part 15.4: Wireless medium ac-

cess control (MAC) and physical layer (PHY) spec-

iﬁcations for low-rate wireless personal area networks

(LR-WPANs). Standard, The Institute of Electrical

and Electronics Engineers, Inc.

IEEE-802.15.4a (2007). Part 15.4:wireless medium ac-

cess control (mac) and physical layer (phy) speciﬁca-

tions for low-rate wireless personal area networks (lr-

wpans): Amendment to add alternate phy. Standard,

The Institute of Electrical and Electronics Engineers,

Inc.

Istepanian, R., Jovanov, E., and Zhang, Y. (2004). Guest

editorial introduction to the special section on m-

health: Beyond seamless mobility and global wireless

health-care connectivity. Information Technology in

Biomedicine, IEEE Transactions on.

Ramachandran, I. and Roy., S. (2006). On the im-

pact of clear channel assessment on mac perfor-

mance. In IEEE Global Telecommunications Confer-

ence (GLOBECOM ’06).

Stoica, L., Tiuraniemi, S., Oppermann, I., and Repo, H.

(2005). An ultra wideband impulse radio low com-

plexity transceiver architecture for sensor networks.

In IEEE International Conference on Ultra-Wideband

(ICU), 2005.

Win, M. and Scholtz, R. (1998). Impulse radio: how it

works. Communications Letters, IEEE.

Zhang, K. and Pahlavan, K. (1992). Relation between trans-

mission and throughput of slotted aloha local packet

radio networks. Communications, IEEE Transactions

on.

Zhen, B., Patel, M., Lee, S., Won, E., and Astrin, A. (2008).

Tg6 technical requirements document (trd). Technical

report, IEEE P802.15.

HEALTHINF 2010 - International Conference on Health Informatics