International Journal of Emerging Trends & Technology in Computer Science (IJETTCS)

Web Site: www.ijettcs.org Email: [email protected], [email protected]

Volume 7, Issue 3, May - June 2018 ISSN 2278-6856

A Mobility Model and Dynamic Scatternet

Formation for Wireless Personal Area Networks
with Portable Bluetooth Devices
Chorng-Horng Yang* and Yi-Sheng Chen
Department of Information Management, I-Shou University
No. 1, Sec. 1, Syuecheng Rd., Dashu District, Kaohsiung City 84001, Taiwan R.O.C.
*
Corresponding author

Abstract: Bluetooth is a mature technology for a variety of

personal computing devices, which are interconnected to form a 2. RELATED WORK
wireless personal area networks (WPANs). The topology of the
WPAN is known as scatternet, which consists of the portable Many scatternet formation protocols were proposed in the
Bluetooth devices. This paper presents a proposed mobility literature. These protocols can be classified into two types.
model for the portable Bluetooth devices. Based on the mobility One is the big-scatternet formation protocol, which
model a dynamic scatternet formation is developed for WPANs constructs a big scatternet and makes the Bluetooth devices
that interconnects portable Bluetooth devices. The proposed always connect to the scatternet. The big-scatternet
approach constructs a scatternet within three phases: scatternet formation protocols include the distributed formation [3],
backbone formation, device movement detection, and scatternet
tree scatternet formation (TSF) [4], MIT Scatternet
reformation. Finally, the paper demonstrates the approach with
a practical case and then discusses the topics for further study. Formation Algorithm [5], BlueStars [6], Loop scatternet
[7], etc.
Keywords: WPAN, scatternet, mobility, Bluetooth. The other one is the scatternet-route structure [8], which
constructs a control scatternet (i.e., a backbone) and
establishes a route on demand. The scatternet-route
1. INTRODUCTION structure does not maintain a big scatternet, so that the
Bluetooth [1] is a promising technology that provides number of control messages for maintaining the scatternet
wireless interconnection between electronic devices for can be reduced and the power can be saved. Since most of
wireless personal area networks (WPANs). A piconet is the the Bluetooth devices are battery-driven, operating these
basic block for constructing WPANs. In a piconet, one devices in a power-efficient mode is essential. Thus, many
device should act as the master and the other devices are researchers investigate the related issues of the scatternet-
slaves. The number of active slaves in a piconet can be up route structure in recent years.
to seven. Devices in a piconet share the same hopping In [9], we have studied the performance of scatternet
channels, which the channel is selected from the 79 routes. Moreover, most of the Bluetooth devices are
hopping channels according to the FH sequence. The portable and mobile, and the users of these Bluetooth
identity and system clock of the master determines the FH devices may take with these devices and walk (roam)
sequence. Each device selects the same hopping channel, among (in) offices. Thus, a scatternet formation protocol
and there creates a unique channel in a piconet. Moreover, has to take not only the power consumption but also the
a number of piconets can interconnect by using the bridge device mobility into account. Unfortunately, there are few
devices to form a scatternet. papers that deal with the mobility issue in the literature. We
There are few papers discuss the mobility of Bluetooth have proposed a mobility model for Bluetooth PANs in [2].
devices in the literature. We proposed a mobility model for On basis of the proposed mobility model, we can further
Bluetooth PANs in [2]. On the other hand, many develop the dynamic scatternet formation protocol to cope
researchers propose the scatternet formation protocols [3- with both the power consumption and the device mobility
8]. The existing scatternet formation protocols do not deal issues. In the following Section, we briefly introduce the
with the device mobility. Thus, we develop a dynamic proposed mobility model.
scatternet formation protocol that takes the device mobility
into account and can format the scatternet in three phases:
(i) constructing a backbone, (ii) detecting the device 3 PROPOSED MOBILITY MODEL
mobility for joining/leaving the scatternet, and (iii)
reformatting the scatternet. The mobility model [2] for Bluetooth scatternets in the
The rest of this paper is organized as follows. Section 2 wireless PAN includes topology model, movement model,
introduces the related work. Section 3 presents the device residence time model, and call arrival model. We describe
mobility. In Section 4, we propose the scatternet formation these models as follows.
protocol. Section 5 summarizes this paper.

Volume 7, Issue 3, May – June 2018 Page 32

 International Journal of Emerging Trends & Technology in Computer Science (IJETTCS)
Web Site: www.ijettcs.org Email: [email protected], [email protected]
Volume 7, Issue 3, May - June 2018 ISSN 2278-6856

3.1 Model of Topology Change

In the literature, there are many kinds of scatternet
topologies be proposed. These topologies include star, ring,
loop, tree, and scatternet-route [3-8]. The topology model
presents the possible topology changes due to devices
movement in a scatternet. We summarize the possible Created using UNREGISTERED Top Draw 4/12/104 7:09:23 AM
topology changes in Table 1. (a) Piconet topology (b) Scatternet topology
From the topology model shown in Table 1, the number change change
of device(s) that move(s) to cause topology change can be Figure 1. Mobility of Bluetooth devices
either single device or multiple devices. For the single
device movement and the moved device is a slave, the Bluetooth devices into low mobility and high mobility
topology changes only occur in the two piconets (see Fig. scenarios. Table 2 summarizes the movement model. Low
1-(a)), which one piconet is the device moves from and the mobility scenario describes that a device moves within a
other piconet is the device moves to. However, if a device small area in an indoor environment. For example, one
whose role is either a master or a bridge moves in the brings Bluetooth devices and moves between offices on the
scatternet, the topology of the scatternet would change. To same floor. High mobility scenario represents the
deal with such a scatternet topology change, the scatternet movement pattern that devices move faster in a larger area.
formation procedure must be run to select another device to For example, one brings Bluetooth devices walking
replace the moved master/bridge, so that not only the between several buildings or moving in a factory by riding
piconet topology but also the scatternet topology changes. on a low-speed car.
We refer to such a topology change as a scatternet topology Moreover, we classify the moving behaviors into tree
change. Note that when a Bluetooth device uses up its patterns: (i) enter/leave a scatternet, (ii) roam among the
battery energy the device would lose its connection to other
piconets that connect to a scatternet, and (iii) turn off the
devices. We also regard such an event as a single device
device temporarily. For the enter/leave pattern, users may
movement.
bring Bluetooth devices to enter/leave to/from a scatternet
For the movement of multiple devices, there are two
possible cases. One is that the moved devices are all slaves, during rush hours. Such a movement may usually be a
so that the masters and bridges within the scatternet remain piconet movement and pass through a particular
unchanged and the topology changes occur only at entrance/exit point (or a particular piconet in the scatternet).
piconets. The other case is that the moved devices The movement pattern of roaming among offices describes
constitute a piconet and such a piconet topology must be that users bring Bluetooth devices and roam among the
retained after devices moving. We refer to such multiple offices during office hours. For this pattern, users usually
devices movement as piconet movement (see Fig. 1-(b)). move to another piconet to stay there for a while and then
Obviously, the scatternet topology would change if a return to the home piconet. For example, users move from
piconet moves. From the point of view of users, the piconet their offices to the meeting room and return to their offices
movement may occur frequently. For example, one may after the meeting. The pattern of turning off device
bring a cell phone (master), a headset (slave) and a PDA temporarily represents the event that a device runs out
(slave) together and moves to a meeting room. And she/he battery power or be turned off by users. We propose the
would like to retain such a piconet topology after moving to associated characteristics for each movement pattern in
the meeting room. Note that there is another case that Table 2. These characteristics are number of moved
multiple devices which include masters or bridges but do devices, residence time (i.e., residence time model), and
not form a piconet move in a scatternet. Such movements geographical range of the movement. The length of
of multiple devices can be regarded as many single-device residence time and distance of movement range we propose
movements occur simultaneously. in Table 2 are considered for the academy in Taiwan.
Table 1. Topology model
Number of Types of Types of
3.3 Model of Device Residence Time
devices device topology
moved moving change The residence time is an important parameter to identify
single slave piconets mobility characteristics. It anticipates the time that a
Single Bluetooth device resides in a piconet. We further define
Master or
device scatternet two parameters for residence time model: (i) T p : The
bridge
multiple piconet residence time represents the length of time that a
Multiple piconets
slaves Bluetooth device resides in a piconet. (ii) Tx : The transition
devices
a piconet scatternet time represents the length of the time period from a
Bluetooth device leaving a piconet to the device settling
down in a new piconet. The distributions of T p and Tx are
3.2 Model of Device Movement
dependent on many factors, such as mobility patterns, etc.
Movement model describes the patterns of Bluetooth The generalized Gamma distribution can provide the best
device movements. We distinguish the movement of
Volume 7, Issue 3, May – June 2018 Page 33
 International Journal of Emerging Trends & Technology in Computer Science (IJETTCS)
Web Site: www.ijettcs.org Email: [email protected], [email protected]
Volume 7, Issue 3, May - June 2018 ISSN 2278-6856

Table 2. Movement model

Characteristic
Movement patterns
Number of devices Resid. time Movement range
Long
Enter/leave Multiple Within 100m
3~8 hrs
Low mobility Roam among offices Single / Multiple Medium 1~3 hrs Within 100m
Turn off device Short
Single Within 100m
temporarily <30mins
Long
Enter/leave Multiple Within 2000m
3~8 hrs
High mobility Roam among offices Single / Multiple Medium 1~3 hrs Within 2000m
Turn off device
Single Short <30mins Within 2000m
temporarily

approximation for residence time in cellular networks.

However, it is for further study to validate the distribution 4.1 Formation of Scatternet Backbone
for Bluetooth scatternets. In the literature, there are lots of papers that present their
approaches of scatternet backbone formation. In this paper,
we propose an approach that is similar to the approach
3.4 Model of Device Call Arrival proposed in [8], which does not provide the details of their
In this paper, Call arrival model gives the probability of approach. During formatting the backbone, the state of the
having incoming calls during the time that a Bluetooth Bluetooth device may be in one of the following states. We
device resides in a piconet. The Poisson distribution is describe these states as follows.
often used to model the calls. Thus, we can also employ
(i) Inquiry (Discovering) and Inquiry Scan States: A
Poisson distribution to model call arrivals in a Bluetooth
Bluetooth device that tries to find other nearby devices
scatternet. Given a rate of  , the call arrival is then defined
is known as an inquiring device. It can actively send
by
inquiry requests and waits for the responses from other
C (t )   e   t (1) devices. On the other hand, the Bluetooth device that is
available to be found is known as a discoverable device.
It listens to the inquiry request and sends the response
back to the inquiring device. Moreover, a pair of
4 DYNAMIC SCATTERNET FORMATION Bluetooth devices that one successfully inquires the
other one can form the simplest piconet, which consists
The proposed dynamic scatternet formation involves three of a master and a slave.
procedures: (i) backbone formation, (ii) device movement
detection, and (iii) scatternet reformation. Figure 2 depicts (ii) Page (Connecting) and Page Scan States: After
the flow of the proposed dynamic scatternet formation. discovering the nearby devices, the master can execute
The first step is to construct the scatternet backbone, the paging procedure for creating connections. The
which consists of the masters and the bridges of the procedure is asymmetrical and requires that one
piconets that connect to the scatternet. After the backbone Bluetooth device carries out the page (connection)
formation, the procedure of the device movement detection procedure while the other Bluetooth device is
is responsible for searching the nearby Bluetooth device connectable, in other words, in page scanning state.
that would like to join the scatternet and detecting any (iii) Connected State: After a successful connection
device that has already connected to the scatternet but now procedure, the devices are physically connected to each
loses its connection to the scatternet. If a new device or a other within a piconet.
leaving device is detected, the procedure of scatternet
reformation starts dealing with the topology change, which Suppose that the Bluetooth devices are brought to
is either a piconet topology change or a scatternet topology offices in an ad-hoc fashion. Each active device runs the
change. After the scatternet reformation, the status of the above procedures to join a piconet to be either a master or a
scatternet can be a scatternet, a scatternet with backbone slave. After the piconets forming, the scatternet is
partitioned, or a vanished scatternet. If the scatternet is constructed by several piconets that interconnect by bridge
maintained well, the flow goes to device movement devices. The bridge device can be either a slave or a
detection. However, if the backbone is partitioned, the flow master/slave (double-role) device. In a scatternet, different
goes to backbone formation procedure. Finally, if the piconets share a common bridge device. The bridge device
scatternet vanishes, the procedure goes to the end. In the then alternates its participation to the two interconnected
following subsections, we describe the three main piconets. After creating a scatternet in such a distributed
procedures, backbone formation, device movement manner, the scatternet backbone can be further constructed.
detection, and scatternet reformation, as follows.

Volume 7, Issue 3, May – June 2018 Page 34

 International Journal of Emerging Trends & Technology in Computer Science (IJETTCS)
Web Site: www.ijettcs.org Email: [email protected], [email protected]
Volume 7, Issue 3, May - June 2018 ISSN 2278-6856

Figure 3 depicts a scatternet that can be further constructed

as the backbone. Backbone
After connecting the piconets together to form a m2 b2 m4
m1
scatternet, each master in the scatternet has to gather the b1 m3/ b3
information of the backbone for routing. The master first
creates a piconet topology table, which stores the physical
addresses and roles (i.e., master, bridge, and slave) of the
devices in the piconet. Then, the master sends its piconet
topology table to its bridge(s), which further send(s) the s1.1 s1.2 s1.3 s2.1 s2.2 s3.1 s4.1 s4.2
piconet topology table to the other master(s) that it connects master slave bridge
to. A master receives the piconet topology tables of other
piconets in the scatternet from the bridges, and then
constructs a scatternet topology table, which links all the Figure 3. Scatternet backbone
piconet topology tables together to provide the master with
the complete information of the scatternet. joining the scatternet as a slave, the new device may change
Table 3 shows the scatternet topology table for the its role to be a bridge or a master during the scatternet
scatternet of Fig. 3. (Note that the items in the table reformation.
represent both the physical addresses and the roles of the For detecting a leaving device, there are two possible
Bluetooth devices in the scatternet. And the piconet address approaches. One is an active approach, which the device
(e.g., p1) is the same as the physical address of the master actively reports the leaving (in this case, leaving means a
(e.g., m1). In order to maintain the scatternet topology device with low power [2]), and the other one is the passive
table, the master periodically sends its piconet topology approach, which the scatternet detects the device movement
table to other masters. Generally speaking, since the (i.e., users bring the devices out of the range of the piconet
Bluetooth PAN naturally consists of a few piconets (e.g., 3 and/or the scatternet [2]). For the active approach, a
or 4 piconets), such a topology information exchange can Bluetooth device that is either a slave or a bridge can be
be done quickly. And the topology information can be configured to report its low power status to the master.
maintained correctly and consistently. After receiving such a leaving report from the slave or
By employing the scatternet topology table, an on- bridge, the master can activate the scatternet reformation
demand route can be established easily and quickly. We procedure to resolve such a device movement. We discuss
demonstrate an example by using the scatternet shown in the scatternet reformation in the following subsection. If the
Fig. 3. If s1.1 wants to connect to s4.2, the device of s1.1 master would like to leave due to low power, the master
sends a Route Discovery Protocol (RDP) packet [8] to its can activate the scatternet reformation by itself to deal with
master m1. The master m1 looks up its scatternet topology the master movement.
table and finds that there are four possible paths to s4.2. For the passive approach of detecting device movement, we
Then, the master m1 sends the RDP packets to bridges b1 discuss according to the role of the device in the scatternet
and b2. When the RDP packets finally arrive at the master as follows.
of piconet 4, i.e., m4, the master m4 obtains the permission (i) Slave or bridge: The master periodically polls the slave
of establishing connection from the slave s4.2, and then and bridge in the piconet not only for data transmission
selects a route and sends back a Route Reply Protocol but also for status report. Let the period of the master
(RRP) packet [8] to the master m1. When the master m1 polling any device in the piconet be T poll seconds, and
receives the RRP packet from m4, the master m1 informs
the master waits for the response from the slave or
the slave s1.1 and starts data transmission.
bridge for Tw seconds. If a slave or a bridge does not
In general, the above proposed scatternet backbone can
successfully support on-demand routing for a static respond to the master’s poll for consecutive k times,
scatternet, in which the Bluetooth devices do not move
frequently. In order to cope with the device mobility in the Table 3. Scatternet topology table
Bluetooth PANs, the device movement detection is
m1 s1.1, s1.2, s1.3, b1, b2
required. We describe the device movement detection as p1
follows. b1: p2 b2: p4

m2 s2.1, s2.2, b1, b2, b3

4.2 Device Movement Detection p2
The device movement detection is responsible for searching b1: p1 b2: p4 b3: p3
a new device that would like to join the scatternet and
detecting the device that loses connection to the scatternet m3 s3.1, b3
p3
due to movement or power off. For searching a new device, b3: p2, p3
the masters in the scatternet could periodically transmit the
inquiry control message to find a new device as discussed m4 s4.1, s4.2, b2, b3
in the previous subsection. The new device that joins to an p4
existing scatternet has to be a slave in a piconet. After b2: p1, p2 b3: p3

Volume 7, Issue 3, May – June 2018 Page 35

 International Journal of Emerging Trends & Technology in Computer Science (IJETTCS)
Web Site: www.ijettcs.org Email: [email protected], [email protected]
Volume 7, Issue 3, May - June 2018 ISSN 2278-6856

the master decide that the slave or bridge has moved out last case is the worst one that no device can be a new
of the range of the piconet/scatternet or powered off. bridge. Thus, the scatternet reformation can activate the
Thus, we can calculate the average time ( Td ) of backbone formation if the backbone is partitioned due
detecting a slave or a bridge moving as follows. to the bridge moving. Similarly, such a topology update
has to be sent to the other piconets by the master. So,
1 the on-demand routing still works successfully.
Td  T poll  k  Tw (2)
2
(iii) Master: When a master will move due to the low
(ii) Master: Similarly, if a slave or a bridge waits for the power, the master can select one of its slaves to be a
master polling and does not receive any poll from the new master. After role switching [1], the master can
master for Tidle seconds, the slave or bridge knows that leave successfully and the piconet topology update can
the master has moved or powered off. Note that the be sent to the other piconets by the new master through
value of Tidle can be set to the value that satisfies the bridges. However, if a master actually moves out of
Tidle  n  T poll . the range of the scatternet, one of the slaves or bridges
will first detect the movement. There are two possible
The average time required to detect the device solutions for resolving the master moving: one is local
movement depends on the above parameters. For different reformation and the other one is globally backbone
device mobility, we can adjust the values of the parameters formation. For the local reformation, the first device
to detect the device movement quickly. However, setting that detects the master moving can be a new master and
improper values for the above parameters may lead to inquire the devices that lose their master to join its new
wrong decision on device movement. Thus, the adaptive piconet. After local reformation, the new master has to
approach for setting the values of the parameters is send the new piconet topology table to the other
important for the scatternet with the devices that have piconets for updating their scatternet topology tables.
varying mobility. After detecting the device movement, the Such a local reformation may fail and the (globally)
scatternet reformation is activated to deal with the device backbone formation is activated to resolve such an
movement. We discuss the scatternet reformation in the issue.
following subsection.
4.4 Performance Evaluation
4.3 Scatternet Reformation
IN THE PROPOSED MOBILITY MODEL [2], TWO TYPES OF
Refer to the simulation model proposed in [10]. A
simulation study was conducted to evaluate the
MOBILITY ARE PROPOSED: ONE IS THE SINGLE DEVICE
performance of scatternet formation. The area is a 50
MOVEMENT AND THE OTHER ONE IS THE MOVING OF A
GROUP OF DEVICES. IN THIS PAPER, WE FOCUS ON THE
meters x 50 meters square for the simulation study. The
SINGLE DEVICE MOVEMENT AND THE GROUP MOBILITY ISSUE
number of nodes is from 25 to 55. The other parameters are
IS FOR FURTHER STUDY. WE DISCUSS THE SCATTERNET
as follows. T poll is set to 0.5s, and Tw is set to 0.75s as well
REFORMATION ACCORDING TO THE ROLE OF THE MOVED as k is set to 3. The generalized Gamma distribution is
DEVICE. employed to provide the best approximation for residence
time. The performance is evaluated with respect to the ratio
(i) Slave: When a slave moves (either be powered off or
of the number of nodes connected to the scatternet to the
actually moves) and be detected by its master, the
total number of nodes. The simulation result is shown in
master updates its piconet topology table and then sends
Fig. 4.
the piconet topology update to the bridge(s). Such a
piconet topology update will propagate to the other For the case of fixed devices, the ratio is higher than
piconets in the scatternet. As discussed in the previous 0.96. For the case of low-mobility devices, the ratio is
subsection, the number of piconets in a scatternet is a higher than 0.91. However, for the case of high-mobility
few. Thus, such a piconet topology update can take devices, the ratio is lower than 0.86. Besides, the ratio
effect quickly.
(ii) Bridge: For the case that a bridge moves and be
detected by its masters, the master with the smallest 1
value of the addresses is responsible for searching an
existing device to replace the moved bridge. There are 0.9
three possible outcomes of executing the scatternet Ratio
reformation. (1) The new bridge completely replaces the 0.8
moved one. In other words, all piconets that the moved
bridge connects to are connected by the new bridge. (2) 0.7
A new bridge is found but not all of the piconets that the 25 35 45 55
moved bridge connects to are connected by the new Number of nodes
bridge. In such a case, the scatternet reformation will
activate the backbone formation, if the backbone is
high mobility low mobility fixed
partitioned. On the other hand, if the backbone is not
partitioned, the scatternet reformation is done. (3) The Figure 4. Simulation result
Volume 7, Issue 3, May – June 2018 Page 36
 International Journal of Emerging Trends & Technology in Computer Science (IJETTCS)
Web Site: www.ijettcs.org Email: [email protected], [email protected]
Volume 7, Issue 3, May - June 2018 ISSN 2278-6856

increases as the number of nodes increases. It is obvious Scatternet-Route Structure for Multihop Ad Hoc
since the density of nodes could increase the connectivity Networks,” IEEE J. on Sel. Areas in Commu., vol. 21,
of the scatternet. Based on the above performance no. 2, pp. 229-239, Feb., 2003.
evaluation, the high-mobility devices may lose their [9] C.-H. Yang, S.-C. Hsu, and H.-J. Tsai, “Performance
connections to scatternet. Thus, more sophisticated Study on Scatternet Routes in Bluetooth Personal
approach should be developed for detecting device Area Networks,” in Proc. IASTED Int’l Multi-Conf.
movement and reconnecting device to scatternet. on Wireless and Optical Communications, pp. 656-
661, July, 2004.
5 SUMMARY [10] E. M. Arpitha and T. M. Ramesh, “Performance
Analysis of Low Rate WPAN Topologies,” Int’l J. of
In this paper, the proposed mobility model for the Innovative Research in Computer and Communication
Bluetooth WPANs was presented. The mobility model Engineering, pp. 4283-4288, May 2014.
consists of the topology model, movement model, residence
time model, and call arrival model. The mobility model can AUTHOR
represent the behavior of the portable Bluetooth devices.
With the proposed mobility model, a dynamic scatternet Chorng-Horng Yang received the B.S. degree in
formation approach was developed and presented. The Electrical Engineering in 1992 from National
approach includes three major procedures: backbone Taiwan University of Science and Technology, and
formation, device movement detection, and scatternet M.S. and Ph.D. degrees in Electrical Engineering
reformation. Moreover, a practical case study was from National Chung Cheng University, Taiwan, in
employed for demonstrating the proposed approach. For 1994 and 2001, respectively. In 1999, he was a
further study, the following topics are interesting. Firstly visiting scholar at the Ohio State University, USA; and from 2005
to 2007 he was the Chief of Networking Section of Computer
the adaptive device-movement detection can facilitate the
Center in I-Shou University, Taiwan. He is currently an assistant
device movement detection. Secondly, the performance of professor at the Department of Information Management, I-Shou
the proposed dynamic scatternet formation can be evaluated University. His research interests include information systems,
on basis of the proposed mobility model. Finally, the wireless networks, performance analysis, and protocols.
scatternet reformation must be further refined for resolving
the movement of a group of devices that would like to Yi-Sheng Chen had graduated from the Department of
retain their roles. Information Management, I-Shou University, Taiwan.

References
[1] Bluetooth SIG, Specification of the Bluetooth
System-Core v1.2, 2003 (www.bluetooth.com).
[2] C.-H. Yang, Y.-S. Chen, J.-W. Ruan, and W.-P. Ho,
“A Mobility Model for Bluetooth Scatternets in
Wireless Personal Area Networks,” in Proc. Int’l
Conf. on Wireless Networks (WNET 2004), pp. 140-
143, June, 2004.
[3] V. Verma and A. Chandak, “Distributed Bluetooth
Scatternet Formation Algorithm,” in Proc. IEEE
Global Telecommunications Conference
(GLOBECOM ’03), vol. 3, pp.1274-1278, December,
2003.
[4] G. Zaruba, S. Basagni, and I. Chlamtac, “Bluetrees-
Scatternet Formation to Enable Bluetooth Based Ad
Hoc Networks,” in Proc. IEEE Int’l. Conf. on
Communications (ICC ’01), vol. 1, pp.273-277, June,
2001.
[5] C. Law and K.-Y. Siu, “A Bluetooth Scatternet
Formation Algorithm,” in Proc. IEEE GLOBECOM
’01, vol. 5, pp. 2864-2869, 2001.
[6] C. Petriol, S. Basagni, and I. Chlamtac, “Configuring
BlueStars: Multihop Scatternet Formation for
Bluetooth Networks,” IEEE Transactions on
Computers, vol. 52, no. 6, pp.779-790, June, 2003.
[7] H. Zhang, J. Hou, and L. Sha, “A Bluetooth Loop
Scatternet Formation Algorithm,” in Proc. IEEE Int’l.
Conf. on Communications (ICC ’03), vol. 2, pp.1174-
1180, June, 2003.
[8] Y. Liu, M. Lee, and T. Saadawi, “A Bluetooth
Volume 7, Issue 3, May – June 2018 Page 37