Information about http://w3.antd.nist.gov/pubs/aswn05.pdf
On the Use of Wireless Network Technologies in …
Tags: cable replacement, fundamental questions, gaithersburg maryland, healthcare environments, hospital environments, interference, medical applications, nada, national institute of standards, national institute of standards and technology, network technology, parameter choices, personal area network, relevance, rf, sensors, suitability, wireless personal area, wireless personal area network, wireless technology,Pages: 6
Language: english
Created: Thu Jul 28 11:26:31 2005
Display cached document
Page 1
Page 2
Page 3
Page 4
Page 5
Page 6
On the Use of Wireless Network Technologies in
Healthcare Environments
Nicolas Chevrollier Nada Golmie
National Institute of Standards and Technology
Gaithersburg, Maryland 20899
Abstract-- In this article, we investigate the suitability of wire- Personal Area Network technology as specified in the IEEE
less technologies in healthcare/hospital environments. We focus on 802.15.4 standard [1] and the Bluetooth [2] technology for
Wireless Personal Area Network technologies, namely, Bluetooth cable replacement and short range connectivity. We try to
and the low-rate specifications described in the IEEE 802.15.4
standard. We evaluate the relevance of each technology for answer the following fundamental questions. What are the pro-
supporting medical applications and examine related scalability tocol parameters that are used in mapping medical applications
issues. Moreover, we consider heterogeneous wireless technology onto wireless technologies? What are the parameter choices
environments and quantify the interaction between Bluetooth that would make this mapping optimal? How scalable are the
devices and IEEE 802.15.4 devices when they operate in the wireless technologies chosen and how well can they support
same environment.
multiple sensors used on a patient's body?
I. I NTRODUCTION Most likely, multiple wireless technolgies will be used
simultaneously in the same area. As they share the same RF
As many hospitals are faced today with increasingly higher spectrum, the interference level between them is a matter
wiring cost to plug more devices on the network, there is a of concern in such unforgiving environments. Thus, after
practical opportunity to replace wires by wireless technologies. evaluating technologies independently, we investigate whether
This approach offers significant benefits in terms of reduc- they can coexist by quantifying the impact of any potential
ing deployment costs and providing safer care to patients. interference.
However, prevailing over wires by switching to wireless tech- The remainder of this paper is organized as follows. Sec-
nologies requires a careful analysis of some of the candidate tion II gives a brief overview of the two selected potential
technologies available in order to find out which ones are best wireless technologies and describes the characteristics and
suitable for such demanding environments. requirements of several medical applications. Section III con-
As a step in this direction, the IEEE 1073 group is cur- siders wireless technologies separately and focus on their scal-
rently developing guidelines for using wireless technologies ability. In Section IV, we examine a heterogeneous wireless
for medical device communications in various healthcare technology environment and provide an evaluation on the
environments. In fact, from a patient's hospital bedside to a impact of interference on performance. The final section offers
doctor's office, there is a wide range of potential applications concluding remarks and future research directions.
and use case scenarios. Medical applications such as real-
time waveform delivery, alarm notifications, asset tracking II. W IRELESS T ECHNOLOGY C ANDIDATES AND M EDICAL
or e-prescription have very strict requirements in terms of A PPLICATIONS
accuracy or latency as data lost or delayed have life and death In this section, we describe two potential wireless tech-
implications but usually have very low data rates. Sensors nology candidates for medical applications and give a brief
carrying these applications may be deployed in high density overview of the characteristics and requirements of these
on a patient's body and at the patient's bedside. Other types applications.
of applications can also be found in the clinical domain.
Queries to hospital databases and Internet access require A. Bluetooth
a fully deployed network infrastructure connecting different The Bluetooth technology is considered a Wireless Personal
departments or hospitals and stress the need for high-speed Area Network (WPAN), intended for cable replacement and
links to carry bandwidth-hungry applications. short distance ad hoc connectivity. Bluetooth operates in the
The many constraints imposed by the variety of applications ISM frequency band starting at 2.402 GHz and ending at
and use case scenarios make the choice of a single fit-all 2.483 GHz in the USA. 79 RF channels of 1 MHz width
wireless technology difficult if not impossible. Therefore, it are defined. The raw rate is defined at 1 Mbit/s and a Time
is expected that many wireless technologies will have to be Division Multiplexing technique divides the channel into 625
used in order to support different application requirements. µs slots. Transmission occurs in packets that occupy an odd
In this article, as we focus primarily on low-rate medical number of slots (up to 5). Each packet is transmitted on a
applications deployed at a patient's bedside, we consider two different hop frequency with a maximum hop frequency rate
potential candidates, namely, the emerging low-rate Wireless of 1600 hops/s.
Master TX Master RX Master TX Master RX Master TX Master RX
· 10 channels in the (902 to 928)MHz band providing 40
kbit/s each
(a) · 16 channels in the (2400 to 2483.5)MHz band providing
Slave RX Slave TX Slave RX Slave TX Slave RX Slave TX
250 kbit/s each
We focus our effort on the unslotted version in the 2450
625us
MHz band as it provides the highest data rate combined with
Slot Time
the least overhead (i.e., no beacon frames).
C. Application Requirements
beacon beacon
In this section, we describe the nature of some medical
applications and their requirements that have life or death
CCA
implications when data is lost, corrupted, or delayed. This is
Contention Period Frame Transmission
(b)
unlike most other environments where these types of require-
ments are mainly financial.
As part of the framework evaluation, the IEEE 1073 group
Time slots
has defined a number of potential medical applications and
Fig. 1. WPAN structure: (a)Bluetooth (b)Slotted 802.15.4 usage cases. Each medical application is defined in terms of
a data rate (raw data needed to be transported), end-to-end
latency (potential packetization and transmission delays), and
According to the specifications, up to 8 bluetooth devices expected coverage area (radio distance between two commu-
can actively participate in a small network, a so-called piconet. nicating devices).
Communications inside a piconet occur between a unique An example of a medical application is the electrocar-
predominant device, the master, and subordinate devices, so- diogram (ECG) monitoring. It uses a star topology where
called slaves. Upon connection establishment, a slave synchro- multiple sensors communicate with a unique collector. An
nizes its time and frequency hopping to the master's and waits ECG is an electrical recording of the heart used in the
to be polled by the master to transmit. In this manner, a slave's investigation of heart disease. It can identify abnormalities
packet always follows a master's packet. Figure 1 describes the in the heart's electrical conduction system. The data stream
transmission of one-slot type packet between a master and one resulting from the digitized analog signal is sent to a control
slave. monitor, available on either a nurse's personal digital assistant
(PDA) or a nurse's personal computer (PC). As part of an
B. IEEE 802.15.4 ECG system, a Personal Worn Device (PWD) defined by the
Another candidate for carrying medical applications is the IEEE 1073 group (i.e., a wireless electrode) generates 4 kbit/s
low rate IEEE Std. 802.15.4-2003. Designed for low cost of data and requires that the addition of the latency introduced
products, it supports very limited battery consumption, a short by the packetization of the samples and the transmission delay
range operation (10m) and low rate applications. Two different remain below 500 ms.
variations of the technology can be found: the slotted and the The goal of our evaluation is to determine how well
unslotted channel structure. Bluetooth and IEEE 802.15.4 support the IEEE 1073 medical
· The slotted channel structure described in Figure 1 uses application described above and identify any scalability issues.
synchronization between devices enabled by beacon ex-
changes and a slotted CSMA/CA mechanism as described III. S CALABILITY I SSUES WITH M EDICAL S ENSOR
in [4]. When a device wishes to send data frames, it waits D EPLOYMENT
for a random number of slots. Then, the medium idleness In this section we focus on the scalability issues pertaining
is evaluated during a CCA (Clear Channel Assessment) to the use of the Bluetooth and the IEEE 802.15.4 wireless
period of time. If the medium is still idle at the end of technologies for medical sensors. We compare the perfor-
this period, the packet can be sent at the beginning of the mance obtained with each technology to support an ECG
next time slot, otherwise the procedure is restarted from application.
the beginning.
· In the unslotted channel structure, if a device wishes to A. Topology and Simulation Parameter Setting
send data frames, it waits for a random period of time. We focus our attention on the ECG system which requires
Then if the medium is still idle after a CCA period of the deployment of multiple electrodes on a patient's body,
time, the frame transmission can start. each of them carrying a low data rate application. We use
The physical layer describes three different frequency the topology depicted in Figure 2. The distance between the
bands: communicating devices remains within the constraints of a
· 1 channel in the (868 to 868.6)MHz band providing 20 room. Since up to 16 leads can be used on a patient's body, this
kbit/s may represent a scalability issue for the technology considered.
Scenario 1: or Scenario 2: TABLE I
Unique monitor Multiple monitors
S IMULATION PARAMETERS
1
2 3 4 802.15.4 Sensor Bluetooth Sensor
Transmitted power (mW) 1 1
5 6 7 Packet header (bit) 72 174
Payload size (bit) 944 936
2m Packet interarrival time (s) 0.236 0.23375
8 9 10
11 12 13
Patient Bed mercial network simulation package OPNET1 . Our simulation
14 15 16
environment is based on detailed MAC, physical layer (PHY)
and channel models. The parameters used in the simulations
2m 2m to model an electrode are summarized in Table I.
Monitor B. Simulation Results
Sensor
We use multiple parameters to evaluate performance of a
given scenario. The performance metrics include:
Fig. 2. Patient bed topology
· End-to-end delay as presented in Figure 3.
· Packet loss at the MAC sublayer of the receiver node
In addition to the configuration requirements, the medical (i.e., the monitor) as shown in Figure 4.
· Efficiency, representing the number of successful data
application data stream needs to be formatted into packets and
mapped onto the baseband framing available. Among the many packets received at the receiver's application layer divided
choices available, we try to minimize the overhead while using by the number of data packets generated by all the
the full payload of each packet. transmitter's application layers related to this receiver as
plotted in Figure 5.
In the case of the IEEE 802.15.4 technology, 944 bits per
packet are available at the application layer. Thus, to achieve In order to comply with medical requirements, the end-
at least the minimum rate required of 4kbit/s, a packet has to to-end delay combined with the latency introduced by the
be generated every 0.236s. packetization has to be below 500 ms and the efficiency has
In the case of the Bluetooth technology, we can choose to be equal to 1. Every application layer packet generated has
either an Asynchronous Connection-Less or a Synchronous to be transmitted and received.
Connection-Oriented (SCO) link. The latter is a symmetric The topology depicted in Figure 2 includes two scenarios. In
point-to-point connection between a master and a specific slave scenario 1, a single access point is used for the central monitor
where a master sends an SCO packet to the slave at regular in order to collect data from all devices placed on the patient's
time intervals. The time interval can either be 2, 4 or 6 time body. When using a single monitor, serious limitations exist
slots for HV1, HV2, or HV3, respectively. SCO links are in terms of scalability, especially for Bluetooth. In fact, due
primilary designed to support voice traffic, thus we use an to the protocol specifications, only 7 slaves can be part of a
ACL link, intended for data communications. An ACL link single piconet, thus allowing only 7 sensors or electrodes to
is an asymmetric point-to-point link between a master and be deployed on the patient's body. In this case, as sensors
active slaves in the piconet using retransmissions to ensure have to be polled every 374 slots, there is enough bandwidth
data integrity. Among all available packet formats on an ACL to acccomodate all 7 slaves. The application delay increases
Link (i.e., DM1, DM3, DM5, DH1, DH3, DH5), DM5 and gradually to reach 0.00975s when 7 slaves transmit data to the
DH5 are not pertinent as the amount of data carried by a full central monitor. This is only due to the round robin mechanism
DM5 or DH5 packet (i.e., 1760 bits and 2680 bits respectively) used to poll alternatively each slave.
implies a significant packetization delay. DH3 and DM3 are To overcome the protocol's limitations in Bluetooth, another
the next available framings as we want to minimize overhead. option is to use a different piconet per sensor/central monitor
We select DM3 as its payload size of 936 bits is very similar pair. This is referred to as scenario 2, in Figure 2, where
to the 944 bits available in an IEEE 802.15.4 packet, which each lead uses a different piconet to send data to its own
makes a comparison between the use of these two technologies central monitor. In this case, the interference resulting from
more practical. multiple Bluetooth piconets operating in close proximity may
Given the 625 µs time slot and the 936 bits payload size of lead to a higher packet loss at the central monitors as seen
a DM3 packet, a device needs to be polled every 374 slots in Figure 4. In Figure 5, we see that up to 13 sensors,
to achieve at least the minimum rate required of 4 Kbit/s 1 Certain equipment, instruments, or materials are identified in this paper
for the PWD application. The interarrival between two packet in order to specify the experimental procedure adequately. Such identification
generations is then 0.23375s (374*625µs) which represents is not intended to imply recommendation or endorsement by the National
Institute of Standards and Technology, nor is it intended to imply that the
the packetization delay. materials or equipment identified are necessarily the best available for the
We develop models for both technologies using the com- purpose
0.1 0.7
IEEE 802.15.4, Scenario 1, UnAcknowledged Service IEEE 802.15.4, Scenario 1, UnAcknowledged Service
IEEE 802.15.14. Scenario 1, Acknowledged Service IEEE 802.15.4, Scenario 1, Acknowledged Service
IEEE 802.15.4, Scenario 2, UnAcknowledged Service IEEE 802.15.4, Scenario 2, UnAcknowledged Service
IEEE 802.15.4, Scenario 2, Acknowledged Service 0.6 IEEE 802.15.4, Scenario 2, Acknowledged Service
Bluetooth, Scenario 1 Bluetooth, Scenario 2
Bluetooth, Scenario 2
0.5
End-to-end Delay (s)
Packet Loss
0.4
0.01
0.3
0.2
0.1
0.001 0
0 2 4 6 8 10 12 14 16 0 2 4 6 8 10 12 14 16
Number of Sensors Number of Sensors
Fig. 3. Delay as a function of number of sensors Fig. 4. Packet Loss as a function of number of sensors
1
the efficiency equals 1 with an end-to-end delay (Figure 3)
that combined with the packetization latency (i.e., packet
interarrival) does not exceed the 500ms limit. Each data point 0.95
in the curve labeled "Bluetooth, Scenario 2" is an average over
the number of devices considered. Beyond 13 piconets, the
packet loss in Figure 4 is so high that the efficiency decreases 0.9
Efficiency
dramatically and the additional of the end-to-end delay and
the packetization latency exceeds the required limit. Thus, for
this specific scenario, up to 13 electrodes can be supported by 0.85 IEEE 802.15.4, Scenario 1, UnAcknowledged Service
IEEE 802.15.4, Scenario 1, Acknowledged Service
dedicating a piconet to a pair of electrode-monitor. Another IEEE 802.15.4, Scenario 2, UnAcknowledged Service
IEEE 802.15.4, Scenario 2, Acknowledged Service
design we can try is to use multiple piconets, each of them Bluetooth, Scenario 2
0.8
carrying data from multiple electrodes. For example, at least 3
piconets can be used to support 16 electrodes. In this case, 10
electrodes can be split on two piconets (5 on each), while the
0.75
third piconet will have 6 electrodes. When all 16 devices are 0 2 4 6 8 10 12 14 16
Number of Sensors
running at the same time, an efficiency of 1 can be achieved.
Although feasible, the main disavantage of this approach lies Fig. 5. Efficiency
in the added configuration and deployment complexity.
We repeat the same two scenarios using IEEE 802.15.4. In
scenario 1, multiple IEEE 802.15.4 sensors communicate with packet on the transmitter side. Even if the drop rate is small
a single access point or central monitor. For the unacknowl- (about 1% when 3 or more sensors are used), it is a major
edged mode in Figure 5 the efficiency starts to drop when disadvantage when using medical applications with strict loss
3 or more devices belong the same network. There are two requirements.
explanations for this phenomenom. In fact, even if devices Using the acknowledged service solves the issue of packet
start their transmission randomly at the beginning of each collisions by retransmitting lost packets but does not provide
simulation, at some point they might end up trying to access any solution against packets dropped at the transmitter. In fact,
the medium at the same time which leads to packets colliding packets can also be dropped at the transmitter side when 3
at the receiver. In this unacknowledged mode, packets that are transmissions have failed to receive a proper acknowlegment.
lost are not retransmitted. Figure 4 shows a significant packet We notice that the drop rate is about 2% when more than
loss when using more than 8 devices. The second explantation 3 sensors are used. The percentage of packet loss is slightly
is that even if a transmitter senses the medium to be busy, higher than in the case of the unacknowledged service since
it has only 4 attempts to access the medium. Between each more packets are exchanged due to (multiple) retransmissions.
attempt, a random backoff has to be performed. As we use For scenario 2, we set each a pair of electrode-monitor
fairly long packets compared to the possible backoff window, on a different channel using up all 16 channels available.
these attempts can be unsuccessful resulting in dropping a First, we run the unacknowledged service. We realize that, for
Patient Bed
our particular topology, there is a significant packet loss as
1
shown in Figure 4 due to co-channel interference making this
service not suitable for medical applications. In addition, using 2 3 4
the acknowledged service does not improve performance.
5 6
As packets are retransmitted, more packets are exchanged 1.5m
resulting in a higher packet loss. Efficiency does not improve
either. In fact, a transmission on a specific channel is unaware
of transmissions on adjacent channels. Therefore, when two
packets are about to be sent on adjacent channels, there
2m
is no mechanism to avoid transmission overlap and packet 802.15.4 Monitor Bluetooth Monitor
collisions. In this case, retransmissions will occur almost
802.15.4 Sensor Bluetooth Sensor
simultaneously and considering the significant packet size,
packets will most likely collide again. After 3 tries, packets
Fig. 6. Heterogeneous Wireless Environment
will eventually be dropped at the transmitter's side.
Considering our particular topology and our specific map-
pings of medical applications onto wireless technologies, it and IEEE 802.15.4 devices when they are present in the same
appears that scalability is not guaranteed. Both Bluetooth and environment.
the IEEE 802.15.4 technologies have severe limitations for
deploying multiple sensors on a patient's body. On one hand, A. Topology and Simulation Setting
protocol specifications and interference strongly limit the de- We extend scenario 2 described in Figure 2 by mixing
ployment of Bluetooth sensors and on the other hand, limited different technologies in the same environment. Six IEEE
bandwith and MAC protocol design limits the use of 802.15.4 802.15.4 sensors (i.e., sensor1, sensor2, sensor3, sensor4,
equipments. Moreover, by using these two technologies, the sensor5, sensor6) are spread over a patient's body according
2450 MHz frequency band is occupied and most likely it will to the topology shown in Figure 6 and transmit medical
not be interference-free. information following the description in Table I. To avoid
any interference between the IEEE 802.15.4 devices, non-
IV. C OEXISTENCE OF B LUETOOTH AND IEEE 802.15.4 IN overlapping channels are used and carriers are set 15MHz
THE SAME ENVIRONMENT apart, thus using 6 channels. We use the acknowledged service
which can potentialy overcome packet loss. Meanwhile, in the
While the first part of this article focuses on scalability same room, a nurse carrying her PDA sends information via
issues for the Bluetooth and the IEEE 802.15.4 technologies a Bluetooth connection to a Bluetooth access point located
and their deployment in a medical environment, we turn our close to the IEEE 802.15.4 monitors (All monitors are located
attention next to investigating how well they can coexist in close to the patient's bedside less than 0.5 meters away). We
the same environment. In fact, the deployment of wireless use the definition of the PDA application from the IEEE 1073
technologies has already started in many hospitals. From group which requires 60kbit/s to be sent. Using DH3 framing,
internet access to file transfer, WLAN is used heavily in a packet is sent every 0.02375s. As we add, IEEE 802.15.4
healthcare environments. Nurses carrying PDAs equiped with sensors one by one, we examine the interactions between the
Bluetooth connectivity exchange patients' information on a two technologies.
regular basis. As time progresses, we can envision using IEEE
802.15.4 sensors on a patient's body in order to collect critical B. Simulation Results
information and sending it to a central monitor located at the First, we look at the impact of a Bluetooth transmission
patient's bedside. on the IEEE 802.15.4 communications. Figure 7 shows the
Interference between Bluetooth and WLAN devices and average packet loss for different IEEE 802.15.4 sensors labeled
its impact on performance has been well documented in the "1" to "6". Differences between sensors are significant due
literature [6][7] and coexistence solutions have been pro- to their location and position with respect to the Bluetooth
posed [8][9]. In addition to these evaluations and coexistence transmitter. Thus, each is impacted differently by the Bluetooth
schemes, Golmie et al. [5] started to examine the interaction transmission. Nevertheless, a factor remains constant; they are
between WLAN and 802.15.4 devices. They noticed that a all severely impacted as the packet loss ranges between 26%
WLAN device can significantly impact IEEE 802.15.4 devices. for sensor6 to 62% for sensor5. During the transmission of a
In some experiments presented in [5], communication between single IEEE 802.15.4 packet, the Bluetooth device would have
IEEE 802.15.4 devices was simply not possible. Nevertheless, hopped on 10 different frequencies, causing errors in the IEEE
both IEEE 802.15.4 and WLAN use spread spectrum tech- 802.15.4 packet being received that then must be dropped. For
niques and one could argue that coexistence between these most of the sensors, using the acknowledged service does not
technologies is only a matter of choosing adequately non- overcome packet loss as retransmissions suffer the same fate
overlapping channels. Consequently, in this article, we focus as the initial transmission. Thus, efficiency drops below 1 to
our effort on evaluating the interactions between Bluetooth reach at most 0.83 for sensor6.
1.2
Efficiency First, the scalability of these technologies is not a given fea-
Packet loss
ture. Both Bluetooth and IEEE 802.15.4 have major constraints
1 in terms of supporting topologies consisting of multiple med-
ical sensors. The protocol specifications (i.e., the limitation of
0.8 7 slaves in a Bluetooth piconet) and the interference between
multiple piconets significantly limit the use of Bluetooth for
0.6
medical sensors. Meanwhile, limited bandwidth and MAC pro-
tocol design restrict the use of the IEEE 802.15.4 technology.
Very specific topologies need to be carefully designed in order
0.4
to support a high sensor density area using either one of these
two technologies.
0.2
Moreover, both technologies use the same RF spectrum and
using them simultaneously leads to severe interference and
0
1 2 3 4 5 6
performance degradation. In our experiments, 802.15.4 devices
Sensor Number are strongly impacted by a nearby Bluetooth communication.
Fig. 7. Packet Loss and Efficiency per IEEE 802.15.4 Sensor Our results show that these technologies are unable to meet
very strict application requirements under certain assumptions
0.03
Heterogeneous Wireless Environment chosen in this paper and thus, their usage in a healthcare
environment may require careful configuration design and even
0.025 protocol enhancements. Future simulations using different
assumptions (i.e., packet size) will help us to quantify a trade-
0.02
off between packet loss, latency and overhead.
Prevailing over wires in healthcare environments by using
Packet Loss
wireless technologies implies searching for mechanisms to
0.015
overcome interference between different technologies. For the
Bluetooth technology, we will adapt existing mechanisms such
0.01
as Adapative Frequency Hopping (AFH) [7], originally devel-
oped to mitigate interference between WLAN and Bluetooth,
0.005 in order to enable coexistence between IEEE 802.15.4 and
Bluetooth devices. In the meantime, our plan is to run complex
0 scenarios with multiple wireless technology devices operating
0 1 2 3 4 5 6
Number of Sensors simultaneously in the same area to explore the impact of
interference among them.
Fig. 8. Bluetooth Packet Loss as a function of the number of IEEE 802.15.4
sensors R EFERENCES
[1] IEEE Std. 802.15.4-2003, "Standard for Telecommunications and In-
formation Exchange Between System - Local Area Metropolitan Area
On the other hand, Bluetooth devices are relatively less Networks - Specific requirements - Wireless Medium Access Control
impacted. Figure 8 shows the packet loss recorded at the (MAC) and Physical Layer (PHY) Specifications for Low Rate Wireless
Personal Area Networks (WPAN)," 2003.
MAC layer of the monitor. As more IEEE 802.15.4 sensors [2] Bluetooth Special Interest Group, "Specifications of the Bluetooth Sys-
are added, the packet loss increases steadily to reach about tem, vol. 1, v.1.0B 'Core' and vol. 2 v1.0B 'Profiles'," Decembre 1999.
2.5%. There are several explanations for this behavior. In this [3] IEEE Std. 802-11, "IEEE Standard for Wireless LAN Medium Access
Control (MAC) and Physical Layer (PHY) Specification," June 1997.
case, the choice of the topology plays an important role in the [4] G. Lu, B. Krishnamachari and C. Raghavendra, "Performance Evalu-
simulation results. The IEEE 802.15.4 sensors are relatively ation of the IEEE 802.15.4 MAC for Low-Rate Low-Power Wireless
far from the Bluetooth monitor in order to cause significant Networks," in IEEE IPCCC EWCN Workshop, April 2004.
[5] N. Golmie, D. Cypher and O. Rebala, "Performance Analysis of low rate
interference. On the other hand, the IEEE 802.15.4 monitors wireless technologies for medical applications," in to appear in Elsevier
that are closer to the Bluetooth monitor and could cause Computer and Communications, 2005.
more interference, mostly receive data sent by the sensors. [6] N. Golmie, N. Chevrollier, and O. Rebala, "Bluetooth and WLAN Co-
existence: Challenges and Solutions," in IEEE Wireless Communications
In the acknowledged service, the IEEE 802.15.4 monitors Magazine, Vol. 10, No. 6, December 2003.
return ACK packets that are relatively short and do not cause [7] N. Golmie, R. E. Van Dyck, A. Soltanian, A. Tonnerre, and O. Rebala,
significant packet loss on the Bluetooth monitor as seen in "Interference Evaluation of Bluetooth and IEEE 802.11b Systems," in
ACM Wireless Networks, Vol. 9, pp. 202-211, 2003.
Figure 8. [8] N. Golmie, "Bluetooth Dynamic Scheduling and Interference Mitigation,"
in ACM Mobile Networks, MONET Vol. 9, No. 1, February 2004.
V. C ONCLUSION AND F UTURE W ORK [9] Carla F. Chiasserini, and Ramesh R. Rao, "Coexistence mechanisms for
In this article, we investigate the use of two wireless tech- interference mitigation between IEEE 802.11 WLANs and Bluetooth", in
Proceedings of INFOCOM 2002, pp. 590-598.
nologies, namely, Bluetooth and IEEE 802.15.4, in healthcare
environments. Our findings are summarized as follows.