Peer to Peer Video Streaming in Bluetooth

Overlays

Sewook Jung,1,2 Alexander Chang,1,3 and Mario Gerla 1,4

June 6, 2007

1Department of Computer Science, University of California, Los Angeles2e-mail: sewookj@cs.ucla.edu3e-mail: acmchang@cs.ucla.edu4e-mail: gerla@cs.ucla.edu

Abstract

As Bluetooth is available in most personal and portable terminals(eg, cellular phone, PDA, videocamera, laptop, etc) P2P video streamingthrough Bluetooth networks is now a reality. Camera equipped Bluetoothphones capture video and broadcast it to other Bluetooth devices andto the infrastructure. Traditionally, large scale Bluetooth networks weredesigned using scatternet concepts. However, many Bluetooth devices donot support Scatternet connections and, even if they support it, they pro-vide only very limited features suitable mostly for static environments.In high mobility situations, a traditional Scatternet design is not usefulbecause of frequent disconnections and reconnections. To overcome theseproblems, we propose Overlaid Bluetooth Piconets (OBP) and SimplifiedOverlaid Bluetooth Piconets (SOBP) that interconnect Piconets formingvirtual Scatternets. In OBP, every Piconet dynamically reconfigures tocollect metadata from neighboring Piconets. If metadata shows the ex-istence of useful data to transfer, an inter-piconet connection is made tocarry out the transfer. SOBP can be used instead of OBP once neighborpiconets have already discovered each other. In this paper, we comparevia analysis and simulation the throughput and efficiency of OBP, SOBPand Scatternet for video applications. We demonstrate the feasibility ofvideo over OBP and SOBP for a representative application.

KEYWORDS:: Bluetooth; Overlay; Peer to Peer; Video Streaming

1 Introduction

Bluetooth is a short-range wireless network technology that supports ad-hocnetwork. It initiates Wireless Personal Area Network (WPAN) and become themost popular WPAN device. Bluetooth uses 2.4Ghz globally unlicensed ISM

1

(a) Cellular Network Share

Bluetooth Links

Video Streams

3G (GPRS, UMTS)

(b) Mobile Video Broadcasting

Figure 1: Scenarios of Video streaming over Overlaid Bluetooth Piconets (OBP)

band and uses Frequency Hopping (FH) with pseudo-random ordering of 79 fre-quencies in the same band to reduce interference. The hopping pattern may beadapted to exclude a portion of the frequencies that are used by interfering de-vices. The adaptive hopping technique improves Bluetooth devices’ co-existencewith static (non-hopping) ISM systems such as 802.11 when these are co-located[6].

Many Bluetooth chips are produced and already installed in many personaldevices such as Laptop, PDA, and Cellular phone. With this proliferation, newBluetooth based applications are needed to cope with people’s demand. Withthe digital camera technology, people are easily making their own video clipsand want to share these with each other. Websites such as ”www.youtube.com”and ”myspace.com” are the main enablers of this phenomenon.

ComVu Media launched the world’s first mobile Webcasting solution in Feb.2005 [1]. ComVu’s PocketCaster software enables live video streaming fromone cellular phone to the other cellular phones. It uses cellular network and10 tier-1 Internet backbone network. All users of ComVu should register firstand download video streams with cellular network. SopCast is a simple, freeway to broadcast video and audio or watch the video and listen to radio onthe Internet [2]. Adopting P2P(Peer-to-Peer) technology, anyone become abroadcaster without the costs of a powerful server and vast bandwidth. SopCastuses wired network and connects all users with Point-to-Point (P2P) Network.

Usually, sharing of video stream happens through wired network. However,we can share video stream in the mobile environment with Bluetooth devices.Figure 1 (a) shows the usage of sharing video stream from 3G cellular network.With this scenario, one user pays for video stream and the other users can shareit for free or with paying less money. Some incentive method can be applied forthis scenario. 3G network enabled user collects money from the other sharingdevices and with informing this transaction to the server and receive incentives

2

from server. Figure 1 (b) shows the usage of sharing video stream from onevideo camera that has Bluetooth devices or cellular phone that has camera andBluetooth device. In a theater or a stadium, end row seated spectators cannotsee details of play. So, they usually use binoculars to see the detail. Play orgame play is captured by one user in the first row and transfer video stream withtree shape Bluetooth network. Data rate of video stream should be re-encodedas 1/2 of original stream to keep suitable delay.

Bluetooth scatternet can form tree or mesh type network but not apt formobile environment because of frequent disconnection and re-connection. More-over, support of Scatternet connection is defined as optional in all Bluetoothspecifications, therefore many Bluetooth chips do not support Scatternet. Evenif Scatternet connection is supported in Bluetooth devices, there is a limita-tion in the number of simultaneous masters a slave can connect to, and alsoforming and keeping Scatternet requires special applications. Because of thesereasons, temporary interconnection of Piconets is more useful than a permanentScatternet in mobile situations.

We proposed Overlaid Bluetooth Piconets (OBP) which enables network ser-vices for mobile users without Bluetooth Scatternet [13]. Bluetooth nodes firstform several Piconets, and OBP forms a virtual Scatternet later. OBP does notform a permanent interconnection of Piconets. Instead, it virtually intercon-nects Piconets when they are in the communication range. By using OBP, eachBluetooth Piconet can collect metadata from the Piconets in the communicationrange. Metadata contains information on transmission nodes, file names, andsynchronization times. If there is real data to transfer between Piconets, it willbe transferred after the metadata exchange. We propose Simplified OverlaidBluetooth Piconets (SOBP) to reduce overhead for already discovered piconets.

This paper has three main contributions. First, we propose Simplified Over-laid Bluetooth Piconets (SOBP). Second, we analyse and simulate SOBP forfeasibility test of video streaming over OBP and SOBP. Third, we show howvideo streaming is possible on top of OBP and SOBP with simple demo.

2 Bluetooth Overview

Bluetooth uses 2.4 GHz ISM band that is divided as 79 channels (1MHz each).Frequency Hopping Spread Spectrum (FHSS) jumps one channel to anotherwith a pseudo-random sequence. Devices that communicate each other shouldshare this channel and jump together. The hop rate is 1600 hops per secondand one channel is used for 625 μs and this is named as a slot.

Up to eight nodes are organized in a star-shaped cluster, called Piconet. Thecluster head is called master and the other nodes are called slaves. Two slavescannot transfer packets directly. So, master should intervene between two slaveswhen a slave transmits packets to the other slave. If more than one slave areconnected to one master, master can select one slave and transmit a packet withusing 1,3,5 slot. When slave receives a data from master, a slave can respondwith 1,3,5 slot packet or 1 slot NULL packet (for acknowledgment). If master

3

(a) Piconet

(b) Scatternet via Slave Bridge

Master Node

Slave Node

Bridge Node

(c) Scatternet via Master Bridge

Bluetooth Links

Figure 2: Piconet and Scatternet

does not have data to send to slaves, master should select one node and send 1slot POLL packet to that slave for enable data transfer for slave.

Piconets are interconnected through bridge nodes and interconnected Pi-conets form a Scatternet. Bridges are the nodes participating in more than onePiconet with a time-sharing method. When a node is acting as a master for acertain Piconet and acting as a slave for the other Piconet, we call it a masterbridge. When a node is acting as a slave for more than one Piconet at the sametime, we call it a slave bridge. Fig. 2 shows examples of Piconet and Scatternet.Fig. 2(a) shows two Piconets that has one master and two slaves. These twoPiconets are interconnected via slave bridge (Fig. 2(b)) and via master bridge(Fig. 2(c)).

Bluetooth data communication usually uses Asynchronous ConnectionlessLinks (ACL) that has time slots of 625μs. Data packets may use 1, 3, or 5 slotsand they may be Forward Error Coded (FEC). FEC packets are DM1, DM3, andDM5 (with the digits indicating the number of slots used). The non-error codedones are DH1, DH3, and DH5. The latest Bluetooth Specification 2.0 introducesEnhanced Data Rate (EDR) packets and they are 2-DH1, 2-DH3, 2-DH5, 3-DH1, 3-DH3, and 3-DH5. The 2-DH(1,3,5) and 3-DH(1,3,5) packets are similarto DH(1,3,5) but uses p/4-DQPSK and 8DPSK modulations, respectively [6].Bluetooth packet information is described in Table 1.

3 Related Works

3.1 Overlay network

In [9], overlay architecture is used to operate on top of the existing proto-col stacks in various network architectures and to provide a store-and-forwardgateway function between them when a node physically touches two or more

4

Type Payload FEC Symmetric Asymmetric Asymmetric(bytes) Max Rate Max Rate Max Rate

(Kbps) (Kbps) (Kbps)Forward Backward

DM1 0-17 2/3 108.8 108.8 108.8DH1 0-27 No 172.8 172.8 172.8DM3 0-121 2/3 258.1 387.2 54.4DH3 0-183 No 390.4 585.6 86.4DM5 0-224 2/3 286.7 477.8 36.3DH5 0-339 No 433.9 723.2 57.6

2-DH1 0-54 No 345.6 345.6 345.62-DH3 0-367 No 782.9 1174.4 172.82-DH5 0-679 No 869.7 1448.5 115.23-DH1 0-83 No 531.2 531.2 531.23-DH3 0-552 No 1177.6 1766.4 235.63-DH5 0-1021 No 1306.9 2178.1 177.1

Table 1: Bluetooth ACL Packets

dissimilar networks.

3.2 Opportunistic network

In ZebraNet [11], wireless sensor nodes are attached to animals and collectlocation data. This data is opportunistically transferred when the nodes are inthe radio range of base stations. They show the effect of mobile base stationsand sensor devices, and the use of two flooding-based routing protocols. InDataMules [20], ”mule” travels among low-power sensor nodes and providesnon-interactive messages periodically to allow sensor nodes to save power.

In Pocket switched Network [10], Bluetooth devices are used in conferencesituations and measure real-world mobility patterns. They used Intel iMoteBluetooth platform to find out human mobility patterns. They check contactand inter-contact time and show many characteristics such as contacts withgroup of nodes, distribution of contacts among nodes, and influence of the timeof day. These results are helpful to determine proper store-and-forward tech-niques.

In [13], Overlaid Bluetooth Piconets (OBP) is introduced and shows thepossibility of interconnecting Bluetooth Piconets without Scatternet. OBP con-tinuously changes its state and collects metadata from probed Piconet and usesmultiple transfers to increase throughput. This state change makes virtual Scat-ternet among Piconets and in the viewpoint of application layer, Piconets areseen as interconnected. In this paper, OBP shows better throughput and effi-ciency than Scatternet.

In [12], Blueprobe, a capacity measurement tool for TDMA protocol, mea-sures allocated capacity of a certain link or a multi-hop path. Moreover, capacityis compared in various situations and shows the effect of hop length and inter-connection types (master bridge or slave bridge). As a result, interconnectiontype affects more than hop length. Based on comparisons among capacitiesof multiple one-to-one connections, interconnection via master bridge, and in-terconnection via slave bridges, multiple one-to-one connections case has the

5

maximum capacity. Interconnection via master bridge is the second, and inter-connection via slave bridge is the last.

3.3 Bluetooth Scatternet

Forming a Scatternet requires special Scatternet formation algorithms. Even if aScatternet is formed, user’s mobility disconnects the initial Scatternet, and thusfrequent reconnections are needed. Many Scatternet algorithms [5, 16, 21, 19]are developed and they help keeping connectivity of each device. However,Scatternet connection increases the average hop length and the number of linksconnected to a certain node, therefore it decreases capacity [15]. To increasecapacity, Scatternet optimization method is needed [14]. Scatternet also has ascalability problem. As number of nodes increases, Scatternet is hard to main-tain because Scatternet maintenance algorithms often use centralized methods.Because of these problems, Scatternet connections are not always useful, espe-cially in high mobility situations.

4 Overlaid Bluetooth Piconets (OBP) and Sim-plified Overlaid Bluetooth Piconets (SOBP)

4.1 Overlaid Bluetooth Piconets (OBP)

Overlaid Bluetooth Piconets (OBP) does not require Scatternet connection. So,all Bluetooth devices used in the world can use OBP as a Piconet interconnectionmethod and form a virtual Scatternet, even if they do not support Scatternet.OBP can be used for the network that has challenging conditions, such as fre-quent disconnections, or long delays due to mobility of nodes. Instead of usingScatternet connection, OBP uses multiple one-to-one connections at the sametime. Because of the frequency hopping scheme, several one-to-one links canbe made and used to transfer at the same time without interference. And thisinterference-free feature increases total capacity.

Consider that we are using Scatternet unsupported Bluetooth devices. Whena Piconet is formed, slave nodes cannot communicate with outside Piconetnodes. Master nodes can do inquiry and look for free nodes (unconnected nodes)in the communication range. Slave nodes cannot do inquiry-scan after their con-nections to a master. So, to do an inquiry-scan or to be connected to anothermaster, a slave node should disconnect from its master node and become a freenode. Therefore, each Piconet continuously changes its stages. Slave stage,Probe stage, Return stage, and Transfer stage are used in this sequence, andthey form OBP Period as shown in Figure 3.

In Slave stage, every node keeps its original Piconet connection and intra-piconet transfers are made. Some nodes may not have any Piconet connection.These nodes remain as free nodes and are denoted as singleton nodes.

In Probe stage, one slave is randomly selected and disconnected from eachPiconet and performs inquiry-scan and we denote this slave as probe node.

6

SlaveStage

Probe is DisconnectedFrom orig. piconet

ProbeStage

ReturnStage

TransferStage

SlaveStage

Probe isConnectedTo visitedpiconet

Probe isDisconnectedfrom visitedpiconet

Probe isConnectedTo orig.piconet

FormOrig.piconet

FormOrig.piconet

Make 1-to-1connectionBetweenSrc. And Dst.

TransferNode isDisconnectedFrom orig. piconet

Overlaid Bluetooth Piconets Period

TransferNodeDisconnects1-to-1 connection

slaveT probeT returnT transferT

periodOBPT _

paget st stpagetttpagetpagetpagetinquiryt mtmt

slaveT

pagetinquiryt : Inquiry Time

: Page Time

mtttst : Slave Time

: Transfer Time

: Metadata transfer Time

Figure 3: Overlaid Bluetooth Piconets (OBP) Period

Master nodes perform inquiry and find out which probe nodes are available inthe communication range. If a master node finds a probe node, master connectsto it. Several probe nodes may be detected at the same time. In this case,master node should decide which one to choose among them. At the first Probestage, master node randomly chooses one probe node and connects to it. At thelater Probe stages, master chooses a probe node that is not connected before.If all probe nodes are connected before, master chooses the probe node that isconnected earlier than other nodes. Master node keeps probe node connectionlog (bd-address and connection timestamp). Singleton nodes have 50% chanceof doing an inquiry-scan (acting as a probe node) and 50% chance of doing aninquiry (acting as a master node). Thus in this stage, probe nodes are createdto be connected to other Piconets (probed Piconets). After the connection, aprobe node transfers metadata to nodes in the probed Piconet and finds outwhether there is useful data or not. If there is data to transmit, probe node andprobed Piconet nodes synchronize transfer start time and decide which nodewill send and receive.

In Return stage, probe nodes are disconnected from the probed Piconetsand return to their original Piconets. Inquiry is not included in this stagebecause master node already knows that probe node (that was slave of thismaster in slave stage) is in the communication range. So, master can connectto probe node with BD ADDR. After connection to the original Piconet, probenode conveys metadata received from the probed Piconet and information aboutwhich nodes are used in the Transfer stage and when it is started.

In Transfer stage, inter-piconet transfer related nodes are disconnected from

7

slaveT probeT returnT transferT

paget st stpagetttpagetpagetpagetinquiryt mtmt

slaveT

pagetinquiryt : Inquiry Time

: Page Time

mtttst : Slave Time

: Transfer Time

: Metadata transfer Time

slaveT probeT returnT transferT

paget st stpagetttpagetpagetpagetinquiryt mtmt

slaveT

NotSynchronized

Synchronized

Figure 4: Synchronization between Piconets

the original Piconets. If a master is related to this transfer, it will disconnect allof its slaves. After the disconnection, source nodes connect destination nodes(form a 1-to-1 connection) and transfer data. Inquiry is also not needed for thisbecause source nodes already know that destination nodes are in the communi-cation range and source node can connect to destination node with destinationnode’s BD ADDR. If inter-piconet transfer related nodes are no more in thecommunication range, those nodes lose the chance of transfer.

After Transfer stage, source and destination nodes return to their originalPiconets and OBP enters Slave stage. This returning is made almost same asReturn stage but at this time, more than one node may be returned to the samePiconet.

Two Piconets may not be synchronized in the Slave stage. However, aftera probe node is connected to the probed Piconet, the probe node will receiveexact synchronization point from the probed Piconet. Two Piconets can be syn-chronized after the Transfer stage. Figure 4 shows how to synchronize betweenPiconets in Probe stage and Return stage.

Each node in the Piconet changes its role according to stages in OBP Period.Figure 5 shows each stage. There are three application flows: from S1 to D1,from S2 to D2, and from S3 to D3. S1, S2, and S3 denote source nodes and D1,D2, and D3 denote destination nodes. Figure 5 (a) shows Slave stage. In thisstage, only intra-piconet transfer is possible because there is no link betweendifferent Piconet nodes. So, only the flow from S3 to D3 can be transferred.The flow will remain until Transfer stage is started because link from S3 to D3is remained as connected until Transfer stage. Figure 5 (b) shows Probe stage in

8

S1 D2

D1S3

S2 D3

(a) Slave Stage (b) Probe Stage

S1 D2

D1S3

S2 D3

S1 D2

D1S3

S2 D3

(c) Return Stage

S1 D2

D1S3

S2 D3

(d) Transfer Stage

Node1

Node2 Node3

Node4

Node5 Node6

Node1

Node2 Node3

Node4

Node5 Node6

Node1

Node2 Node3

Node4

Node5 Node6

Node1

Node2 Node3

Node4

Node5 Node6

Master Node

Slave Node

Probe Node

Free Node

S_ : Source

D_ : Destination

Bluetooth Links

Application Flows

Figure 5: Overlaid Bluetooth Piconets Stages

which probe nodes (node 3 and 5) are disconnected from their original Piconetsand are connected to probed Piconets. After these connections, the probe nodesand the nodes in the probed Piconets exchange metadata. Synchronized transfertime will be assigned at this time. Figure 5 (c) shows Return stage and theprobe nodes return to their original Piconets and convey the metadata to theirPiconet nodes. Figure 5 (d) shows Transfer stage. In this stage, source anddestination nodes are disconnected from their original Piconets. Source nodesmake connection to destination nodes and start inter-piconet transfers such asS1 → D1 and S2 → D2.

4.2 Simplified Overlaid Bluetooth Piconets(SOBP)

Among OBP stages, Probe Stage and Return Stage are used for discoveringneighbor piconets and collecting metadata. If we have already collected meta-data, and these two piconets are temporarily static, no further discovery isneeded.

So, we propose Simplified Overlaid Bluetooth Piconets (SOBP) for this casein which probe nodes have application flows (source or destination) of two pi-conets.

Figure 6 shows stages of SOBP and it has Original Stage and Visit Stage. InOriginal stage, only intra-piconet transfers are possible, whereas in Visit Stage,probe nodes can do inter-piconet transfer. In Figure 6 (a), there is one intra-piconet transfer (S1 → D1). In Figure 6 (b), there is one inter-piconet transfer(S2 → D2).

9

S1D3

D1D2

S2D4

S3S4

(a) Original Stage (b) Visit Stage

S1

D1D2

S2

S3S4

Node1

Node2 Node3

Node4

Node5 Node6

Node1

Node2 Node3

Node4

Node5 Node6

Master Node

Slave Node

Probe Node

S_ : Source

D_ : Destination

Bluetooth Links

Application Flows

Figure 6: Simplified Overlaid Bluetooth Piconets Stages

With proper buffering, one node can simultaneously play two video streamsfrom two sources, one in its own piconet and the other in different piconet. Forexample, Node 3 can play two video streams (from Node 1 and Node 4) at thesame time.

Two destination nodes can play same video stream from one common source.For example, Node 1 and Node 4 can download same video stream from Node5 and play at the same time.

5 Analysis

5.1 Throughput and Power Estimation

Throughput and Power are estimated to make comparison among OBP, SOBPand Scatternet.

5.1.1 Overlaid Bluetooth Piconet (OBP)

Slave stage, Probe stage, Return stage, and Transfer stage durations are denotedas (1)-(4) and OBP Period duration is the sum of all stages’ durations anddenoted as (5).

Tslave = tpage + ts (1)

Tprobe = tinquiry + tpage + tm (2)

10

Treturn = tpage + tm (3)

Ttransfer = tpage + tt (4)

TOBP period = Tslave + Tprobe + Treturn + Ttransfer (5)

tpage and tinquiry are page time and inquiry time, respectively. tm is meta-data transfer time in Probe stage and Return stage. ts is slave time in Slave stageand used only for intra-piconet transfer. tt is transfer time in Transfer stageand used for inter-piconet transfer. But, intra-piconet transfer is still possibleduring Transfer stage because not all the Piconet links are disconnected everytime. If source and destination nodes are not used for inter-piconet transfer,they can be used for intra-piconet transfer.

Intra-piconet throughput in OBP is calculated as follows.

θsdOBP intra = C · qsd · fsd · pi · ( tt

TOBP period+ (1 − pe) · Ttransfer

TOBP period) (6)

Intra-piconet transfer is possible during tt when source and destination arein the same Piconet. It is also possible during Ttransfer when source and des-tination remain in the same original Piconet because they are not used forinter-piconet transfer. C is the maximum capacity of a Bluetooth radio link,specified in Table 1. fsd is usage percentage of capacity. It is calculated by 1over the number of intra-piconet flows in one Piconet for intra-piconet case andis calculated by 1 over the number of inter-piconet flows located at same nodefor inter-piconet case. qsd is the Link Quality (LQ) of the link (s, d) that canbe obtained from the packet error rate (PER), as (7), while PER, denoted by r,can be calculated as a function of the bit error rate (BER), using the formulae(8) and (9), for DH and DM packet types, respectively [8].

q = 1 − r (7)

r = 1 − (1 − b)s (8)

r = 1 − ((1 − b)15 + 15b(1 − b)14)s/15 (9)

pi is the probability of intra-piconet (internal) flow existence and pe is theprobability of inter-piconet (external) flow existence.

Assume that N is the set of nodes in the conference room and F is the setof all flows in all nodes. In that case, |F | sources and |F | destinations exist. So,the possibility of having a source or a destination at a certain node is |F |/|N |.And then, pi and pe are calculated as follows.

11

pi =|F ||N | ·

nPiconet − 1|N | − 1

(10)

pe =|F ||N | ·

nprobed Piconet − 1|N | − 1

· pprobe (11)

nPiconet and nprobed Piconet are the number of nodes in original Piconet andin probed Piconet, respectively. pprobe is probability that at least one Piconet isprobed. It depends on the communication range and nodes’ moving range. If allnodes are in the communication range, all Piconets are in the same range. So,at least one Piconet detects probe node and connects to it. In this case, pprobe

is 1. If all nodes are not in the communication range, pprobe is communicationarea divided by moving area. Near the boundary, communication area will bedecreased because it is not a full circle. So, pprobe can be calculated as follows.When all nodes are in the communication range (12) is applied and when allnodes are not in the communication range (13) is applied.

pprobe = 1 (12)

pprobe � 1 − (1 − 102π

Xr · Yr)(|N |/nP iconet)−1 (13)

|N |/nPiconet is average number of Piconets, and (1 − 102πXr·Yr

)(|N |/nP iconet)−1

is the probability that all other Piconets are not in the communication range of10m in the moving area of Xr by Yr.

Inter-piconet throughput is calculated as follows.

θsdOBP inter = C · qsd · fsd · pe · ( tt

TOBP period) (14)

Total throughput is the sum of intra-piconet transfer and inter-piconet trans-fer and it is calculated as follows.

θOBP =∑

(s,d)∈F

(θsdOBP intra + θsd

OBP inter) (15)

Power consumption for OBP is calculated as follows.

POBP =∑

(s,d)∈F

(Pt + Pr) · hsd · fsd + POBP con (16)

hsd is the hop distance between source and destination. For the intra-piconettransfer, the hop distance is 1 (master and slave) or 2 (slave and slave), and forthe inter-piconet transfer, it is 1. In [15], Pt and Pr are assumed as transmittingand receiving power consumption at the full capacity of a radio link. POBP con

is the power consumed for connection and disconnection in various stages.

12

5.1.2 Simplified Overlaid Bluetooth Piconet (SOBP)

Original stage and Visit stage durations are denoted as (17)-(18) and SOBPPeriod duration is the sum of all stages’ durations and denoted as (19).

Torig = tpage + to (17)

Tvisit = tpage + tv (18)

TSOBP period = Torig + Tvisit (19)

tpage is page time defined in OBP case. to is original piconet transfer timein Original stage and used only for intra-piconet transfer. tv is visit piconettransfer time in Visit stage and used for inter-piconet transfer. But, intra-piconet transfer is still possible during Visit stage because not all the Piconetlinks are disconnected every time. If source and destination nodes are not usedfor inter-piconet transfer, they can be used for intra-piconet transfer.

Intra-piconet throughput in SOBP is calculated as follows.

θsdSOBP intra = C · qsd · fsd · pi · ( to

TSOBP period+ (1 − pe) · Tvisit

TSOBP period) (20)

Intra-piconet transfer is possible during tv when source and destination arein the same Piconet. It is also possible during Tvisit when source and destinationremain in the same original Piconet because they are not used for inter-piconettransfer. C, fsd, qsd, pi, and pe are same as in OBP case.

Inter-piconet throughput is calculated as follows.

θsdSOBP inter = C · qsd · fsd · pe · ( tv

TSOBP period) (21)

Total throughput is the sum of intra-piconet transfer and inter-piconet trans-fer and it is calculated as follows.

θSOBP =∑

(s,d)∈F

(θsdSOBP intra + θsd

SOBP inter) (22)

Power consumption for OBP is calculated as follows.

PSOBP =∑

(s,d)∈F

(Pt + Pr) · hsd · fsd + PSOBP con (23)

hsd is the hop distance between source and destination and used same asOBP case.

13

Number of Piconet member [1, 4](nP iconet or nprobed P iconet)

Number of Nodes(|N |) 50Number of Flows(|F |) 100

Capacity (C) 723KbpsInquiry time and Page time (4, 2) sec

(tinquiry, tpage)Slave time and Transfer time (5, 5) sec

(ts, tt)or

Piconet time and Scatternet time(tpi, tsc)

Table 2: Comparison Parameters

5.1.3 Bluetooth Scatternet

In [15], throughput is calculated as follows.

θscatter =∑

(s,d)∈F

C · min(i,j)∈(s,d)

(fsdij qij) (24)

min(i,j)∈(s,d)(fsdij qij) denotes the smallest usable bandwidth portion on the

links of a connection (s, d) (i.e the bottleneck), while qsd is the link quality (LQ)of the link (i, j).

In [15], power consumption is calculated as follows.

Pscatter =∑

(s,d)∈F

(Pt + Pr) · hsd · min(i,j)∈(s,d)

fsd + Precon (25)

Precon is the power consumed for reconnection of link when Scatternet ispartitioned.

5.1.4 Throughput comparison

Throughputs of OBP, SOBP, and Scatternet are calculated as (15), (22), and(24), respectively. We assume parameters as in Table 2. And then, pi and pe

are calculated as follows.

pi =10050

· 2.5 − 149

= 0.061224 (26)

pe =10050

· 2.549

· pprobe = 0.102041pprobe (27)

We assume Link Quality qsd as 0.25, and Usage Percentage fsd as 0.2 forintra- and inter-piconet transfers in OBP and SOBP. But for Scatternet case,it is set to lower values because Scatternet increases retransmission based ondisconnection. fsd is calculated by average number of flows in same Piconet orScatternet. Average number of nodes in the Piconet, nPiconet = 2.5, thereforeAverage number of Piconet is calculated as 50/2.5 = 20. So, number of flows

14

in each Piconets is calculated as 100/20 = 5. If all flows passes same node andthen fsd = 1/5 = 0.2. And then throughput of OBP is calculated as

θsdOBP intra = 723Kbps · 0.25 · 0.2 · 0.061224 · ( 5

23+ (1 − 0.102041 · pprobe) · 7

23)

= (1.154737878 − 0.068735pprobe)Kbps

(28)

θsdOBP inter = 723Kbps · 0.25 · 0.2 · 0.102041 · pprobe · 5

23= 0.801909pprobeKbps

(29)

θOBP = 100 · (θsdOBP intra + θsd

OBP inter)= (115.4737878 + 73.3174pprobe)Kbps

(30)

We assume Link Quality qsd as 0.25, and Usage Percentage fsd as 0.2 forintra- and inter-piconet transfers in SOBP, these are calculated same as thoseof OBP. And then, throughput of SOBP is calculated as

θsdSOBP intra = 723Kbps · 0.25 · 0.2 · 0.061224 · ( 5

14+ (1 − 0.102041 · pprobe) · 7

14)

= (1.897069371 − 0.1129217857pprobe)Kbps

(31)

θsdSOBP inter = 723Kbps · 0.25 · 0.2 · 0.102041 · pprobe · 5

14= 1.317421929pprobeKbps

(32)

θSOBP = 100 · (θsdSOBP intra + θsd

SOBP inter)= (189.7069371 + 120.4500143pprobe)Kbps

(33)

We assume Link Quality qsd as 0.2, and Usage Percentage fsd as 0.01 forScatternet. If all flows go through one node, then fsd = 1/100 = 0.01. There-fore, throughput of Scatternet is calculated as

θscatter = 100 · 723Kbps · 0.2 · 0.01= 144.6Kbps

(34)

Figure 7 shows throughputs of OBP, SOBP, and Scatternet versus probeprobability (pprobe). When probe probability is increased, throughput of OBP

15

100

150

200

250

300

350

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Thr

ough

put(

kbps

)

Probe probability

Throughput vs. Probe probability

OBPSOBP

Scatternet

Figure 7: Throughput vs. Probe probability

and SOBP are increased. Based on our assumption, in the higher probe proba-bility, OBP and SOBP show better performance than that of Scatternet. Mov-ing range and mobility affect probe probability. When moving range is small,all nodes are always discoverable therefore probe probability is 1. As movingrange is bigger, some part of nodes are undisdoverable. Node’s mobility increasechance of meeting out-of-range nodes and increase probe probability again. But,throughput will be decreased as nodes are moving.

5.2 Feasibility Test for Video Streaming over OBP andSOBP

If two piconets temporarily stay in the communication range, video streamingfrom two different piconets can be done. In this case, pi = 1 and po = 1 becausethere is one flow from each piconet. We can assume Usage Percentage fsd as 1for intra- and inter-piconet transfers for this special case because there is onlyone flow at a time. With using (7) and (8), throughputs of OBP and SOBP arecalculated as (35)-(38). DH5 packet has 339 Bytes and that is 2712 bit.

θsdOBP intra = 723Kbps · qsd · 1 · 1 · ( 5

23+ (1 − 1) · 7

23)

= 157.1714 · qsdKbps

= 157.1714 · (1 − (1 − (1 − b)s)Kbps

= 157.1714 · (1 − b)2712Kbps

(35)

θsdOBP inter = 723Kbps · qsd · 1 · 1 · 5

23= 157.1714 · qsdKbps

= 157.1714 · (1 − b)2712Kbps

(36)

16

0

50

100

150

200

250

300

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Thr

ough

put(

kbps

)

Link Quality

Throughput vs. Link Quality

OBPSOBP

Figure 8: Throughput vs. Rate

0

50

100

150

200

250

300

1e-10 1e-09 1e-08 1e-07 1e-06 1e-05 1e-04 0.001

Thr

ough

put(

kbps

)

Bit Error Rate (BER)

Throughput vs. Bit Error Rate

OBPSOBP

Figure 9: Efficiency vs. Rate

θsdSOBP intra = 723Kbps · qsd · 1 · 1 · ( 5

14+ (1 − 1) · 7

14)

= 258.2143 · qsdKbps

= 258.2143 · (1 − b)2712Kbps

(37)

θsdSOBP inter = 723Kbps · qsd · 1 · 1 · 5

14= 258.2143 · qsdKbps

= 258.2143 · (1 − b)2712Kbps

(38)

Let’s compare these results with those of two flows in single piconet case.We can assume a piconet that has one master and two slaves. There are twoflows from the master to each slave. And then, throughput is calculated as (39).Nf is number of flows in piconet, therefore it is 2. e is efficiency of link andassumed as 0.8. When the master transfers data to two slaves at the same time,some slots are missed because of switching between two slaves.

θsdpico = 723Kbps · qsd · fsd · 1

Nf· e

= 723Kbps · qsd · 1 · 12· 0.8

= 289.2 · qsdKbps

= 289.2 · (1 − b)2712Kbps

(39)

Throughput of OBP and SOBP for video streaming are shown as Figure 8and 9. For both OBP and SOBP cases, two video streams from two differentpiconets can be supported up to 110.02 Kbps and 180.75 Kbps, respectively,when qsd is 0.7.

17

Connection for one stream lasts only 5 seconds and it will be disconnectedfor 18 seconds for OBP case and 9 seconds for SOBP case. So, before playing thevideo, data should be buffered. Buffer amount BOBP and BSOBP are calculatedas (40) and (41). This amount can be estimated as 248 Bytes and 204 Bytesfor OBP (110.02 Kbps stream is used) and SOBP (180.75 Kbps stream is used),respectively, when qsd is 0.7.

BOBP = 18/8 · θOBP inter

= 353.6357 · qsdBytes

= 353.6357 · (1 − b)2712Bytes

(40)

BSOBP = 9/8 · θSOBP inter

= 290.4911 · qsdBytes

= 290.4911 · (1 − b)2712Bytes

(41)

6 Simulation

In this section, we present the simulation environment that we used for evalu-ating our approach.

6.1 UCBT Simulator

For evaluation purposes, we implemented OBP algorithms in the UCBT ns-2[3] based Bluetooth simulator [4], because it is the only publicly available opensource Bluetooth simulator that supports mesh-shaped Scatternets.

UCBT implements the majority of the protocols in the Bluetooth. The sim-ulator has recently added support for mesh-shaped Scatternets, but it assumesthat all nodes are in the communication range. Therefore, we also added toUCBT a simple Scatternet formation protocol (described in section 6.3), be-sides our OBP algorithms.

6.2 Mobility

We assume Bluetooth devices are used in a conference room that has fixedboundary. A group of people are moving together with specific waypoint. Forsimulating mobility, we use the revised random waypoint model and Nomadiccommunity mobility model in [7]. Because Piconets are moving together, weassume a Piconet master is moving according to the random waypoint modeland slaves are staying in the short range (< 3m) of their master. Therefore, allPiconet members are moving to randomly chosen direction and speed. Maxi-mum speed (0.0, 0.3, 0.6. 0.9, or 1.2 m/s) is predefined to limit node’s speed.1.2 m/s is selected because this speed is same as 4.32 km/h which is just abovewalking speed. To add random factor, direction is changed periodically with an

18

Moving Area(Xr, Y r) 15.1 × 15.1m2, 21.28 × 21.28m2

Number of Piconet member [1, 4](nP iconet or nprobed P iconet)

Number of Nodes(|N |) 50Number of Flows(|F |) 100

Moving speed of nodes(S) 0.0, 0.3, 0.6, 0.9, 1.2 m/sPacket type(P ) DH5, 2-DH5, 3-DH5

Inquiry time and Page time (4, 2) sec(tinquiry, tpage)

Slave time and Transfer time (5, 5) sec(ts, tt) (7, 7) sec

or (10, 10) secPiconet time and Scatternet time

(tpi, tsc)

Table 3: Simulation Parameters

offset in the range [-10, 10] degrees with respect to the original direction. Whena node reaches the boundary of the simulation area, it is mirrored back into thesimulation area.

6.3 Scatternet Formation

We implemented a Scatternet algorithm based on [5, 16, 15]. On the first phase,nodes execute inquiry or inquiry-scan with a probability of 1/4 and 3/4, respec-tively. When an inquiry node discovers an inquiry-scan node, it will page theinquiry-scan node. This way, the inquiry node becomes a master of the othernode in the newly formed Piconet. After this first phase, Piconets are formed.On the second phase, master nodes execute inquiry and slave nodes executeinquiry-scan. When master detects nodes that have hop distance longer thanMAX HOP distance (we define it as 4), master connects them and a Scatternetis formed.

Node’s mobility can disconnect certain link and make hop distance longerthan 4 (if partition is made, hop distance is set as ∞). For healing partition andlong hop distance, Scatternet reconfiguration procedure makes reconnectionsand reduces hop distance.

6.4 Parameters

Parameters are described in Table 3.

7 Results

We evaluate throughput and efficiency (throughput / power consumption) ver-sus speed, data rate, and time. We also check number of distinct probed Piconetsper second versus slave and transfer times. For all simulations, we set transfertime value same as slave time value, and Piconet time value same as Scatternettime value.

19

0

50

100

150

200

0 0.3 0.6 0.9 1.2

Thr

ough

put(

kbps

)

Speed (m/s)

Throughput vs. speed (15.1^2 m^2)

OBP ST = 5OBP ST = 7

OBP ST = 10Scatternet

Figure 10: Throughput vs. Speed

0

50

100

150

200

0 0.3 0.6 0.9 1.2

Effi

cien

cy(k

bit/J

)

Speed (m/s)

Efficiency vs. speed (15.1^2 m^2)

OBP ST = 5OBP ST = 7

OBP ST = 10Scatternet

Figure 11: Efficiency vs. Speed

7.1 Throughput vs. Speed

Figure 10 shows throughput vs. speed results. We use maximum moving speedvarying from 0 to 1.2 m/s to evaluate the throughput versus speed. DH5 packetsand 15.1×15.1m2 area are used for this test.

As the speed increases the throughput of Scatternet decreases. When nodesare moving, nodes can be moved out of communication range. At this time,supervision timeout will happen and therefore that link is disconnected. Discon-nection will make Scatternet partition and requires reconnection. Until recon-nection, application flow should be stopped. These frequent link disconnectionsand reconnections reduce throughput.

However, the throughputs of OBP cases stay the same or increase as thespeed increases. OBP uses opportunistic transfers, therefore meeting chance isthe most important factor of throughput. High mobility makes higher chance ofmeeting other Piconets, which produces more inter-piconet transfers in OBP andthus increases throughput. Moreover, OBP uses multiple one-to-one connectionsto fully utilize frequency hopping method. Frequency hopping method usespseudo random frequencies, and therefore multiple one-to-one transmissions viamultiple links can be possible without interference.

7.2 Efficiency vs. Speed

Figure 11 shows efficiency vs. speed results. The same testing environment isused as in section 7.1.

The power consumptions in OBP cases are higher than that of Scatternetbecause of higher throughput, frequent connections and disconnections, andmetadata transfers. Even though the power consumption is higher in OBP,the throughput is much higher than that of Scatternet, which results in betterefficiency in high mobility cases for OBP.

20

0

100

200

300

400

500

600

3-DH52-DH5DH5

Thr

ough

put(

kbps

)

Rate

Throughput vs. rate (15.1^2 m^2)

OBP ST = 5OBP ST = 7

OBP ST = 10Scatternet

Figure 12: Throughput vs. Rate

0

50

100

150

200

250

300

3-DH52-DH5DH5

Effi

cien

cy (

kb/J

)

Rate

Efficiency vs. rate (15.1^2 m^2)

OBP ST = 5OBP ST = 7

OBP ST = 10Scatternet

Figure 13: Efficiency vs. Rate

7.3 Throughput vs. Rate

Figure 12 shows throughput vs. rate results. For this test, DH5, 2-DH5, and3-DH5 packets are used. The speed is set to 1.2 m/s speed and the area is setto 15.1×15.1m2 for this test.

When higher capacity packets are used, throughput increases as we expectedin all cases. All OBP cases’ throughputs are better than those of Scatternetbecause of multiple one-to-one transfers in OBP. Throughputs of 2-DH5 and 3-DH5 are not increased as twice or three times of DH5 case because OBP requiresprobe and connection.

7.4 Efficiency vs. Rate

Figure 13 shows efficiency vs. rate results. With the same testing environmentas in section 7.3, the efficiencies of OBP and Scatternet do not vary a lot for aparticular rate. As the rate increases, the efficiencies increase as well followingthe same pattern of throughput in section 7.3, because the power consumptionsdo not vary very much among different rates.

7.5 Probe Rate vs. Speed

Figures 14 and 15 shows the number of distinct probed Piconets per second withvarying speeds in the areas of 21.28×21.28m2 and 15.1×15.1m2, respectively.When speed increases, the percentage of probed Piconets increases in both areas.And this increase reflects the increase in throughput shown in section 7.1. Also,in the larger area, the percentage increase between the speeds of 0 and 0.3 m/sis significant compared to other speed differences as expected, because nodesstart moving increases the chance of meeting other Piconets. This is not thecase for the smaller range as more Piconets are already in the communicationrange even if speed is 0 m/s. Among different slave times and Scatternet times,shorter ones have higher probe rate than longer ones as we expected, becausetotal OBP periods are directly proportional to the times and thus decreases thenumber of probe.

21

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.3 0.6 0.9 1.2

Num

ber

of p

robe

/sec

Speed (m/s)

Number of probe vs. speed (21.28^2 m^2)

OBP ST = 5OBP ST = 7

OBP ST = 10

Figure 14: Probe rate vs. Speed (21.28× 21.28 m2)

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.3 0.6 0.9 1.2

Num

ber

of p

robe

/sec

Speed (m/s)

Number of probe vs. speed (15.1^2 m^2)

OBP ST = 5OBP ST = 7

OBP ST = 10

Figure 15: Probe rate vs. Speed (15.1× 15.1 m2)

0

100

200

300

400

500

0 30 60 90 120 150 180 210 240 270 300

Thr

uput

(kb

ps)

Time (s)

Thruput vs. Time

OBP ST = 5OBP ST = 7

OBP ST = 10Scatternet

Figure 16: Throughput vs. Time

0

50

100

150

200

250

300

350

400

0 30 60 90 120 150 180 210 240 270 300

Effi

cien

cy (

kb/J

)

Time (s)

Efficiency vs. Time

OBP ST = 5OBP ST = 7

OBP ST = 10Scatternet

Figure 17: Efficiency vs. Time

7.6 Throughput vs. Time

Figure 16 shows every 10 seconds’ average throughput. We use 1.2 m/s speed,2-DH5 packets, and 15.1×15.1m2 range for this test.

In OBP, throughput varies a lot during the test time, because inter-piconettransfers (which is the main part of the throughput) are only possible duringTransfer stage. During this stage, the throughput is high and in other stages itis low, and this is reflected in the oscillation of the throughputs in the figure.Shorter slave time one has shorter Transfer stage and thus has shorter oscillationperiod where as the longer one has longer period.

In Scatternet, node’s movement disconnects some links, and thus decreasesthroughput at certain times, and reconnection regains the throughput.

7.7 Efficiency vs. Time

Figure 17 shows every 10 seconds’ average efficiency. Same parameters in section7.6 are used. During Probe and Return stages of OBP, power for inter-piconettransfers disappears, instead, power for connections and disconnections is con-

22

Path Length

Bluetooth Links

Video Streams

361.5Kbps

180.75Kbps

90.37Kbps

Level 1

Level 2

Level 3

Figure 18: Video Streaming over Scat-ternet

(a) Odd Level Link Transfer

(b) Even Level Link Transfer

Bluetooth Links

Video Streams

Level 1

Level 2

Level 3

Level 1

Level 2

Level 3

Path Length

Path Length

361.5Kbps

90.37Kbps

180.75Kbps

Figure 19: Synchronous Video Stream-ing over SOBP

Bluetooth Links

Video Streams

A

B C

C1 C2B1 B2

… … … …

A

B C

C1 C2B1 B2

… … … …

A

B C

C1 C2B1 B2

… … … …

(a) Time slot 1

(b) Time slot 2 (c) Time slot 3

Figure 20: Asynchronous Video Streaming over SOBP

sumed. So, power consumption does not decrease as throughput decreases. InScatternet, power consumption is almost constant throughout the simulations.Thus, the efficiency follows the throughput pattern in section 7.6.

7.8 Throughput vs. Path Length

Consider the scenario shown in Figure 1 (a). Imagine that you are in a soccerstadium watching a national championship tournament game. In a champi-onship tournament, another 10 or 15 soccer games are played at the same timeacross the Nation. The networks often broadcast a single live video streamcombining all games and summarizing the most important actions/events ineach game (For example, the cellular network provider may broadcast the videostream over its 3G service at 100Kbps). Each spectator at the stadium can

23

individually receive the 3G stream on the smart phone, for a price. However, ifAll spectators indeed tried to access 3G at the same time, most of them wouldbe blocked due to limited 3G cellular capacity.

With Bluetooth, another option is available that greatly relieves this bottle-neck. Namely, a limited number of users receive the 3G stream and re-broadcastit to their neighbors using protocol. Many video streaming P2P protocols havebeen deployed in the fixed Internet (eg, SopCast [2]). These protocols are allinspired to the BitTorrent overlay concept, and can be applied to the SOBP envi-ronment with proper modifications (as already demonstrated in the CarTorrentexample [18]). Namely, a user interested in the soccer video stream must firstdiscover a neighbor already involved in the download and then download fromit.

In this SOBP example, we want to address the feasibility of such a downloadin a stadium environment. We will assume that nodes are static and each nodehas N neighbors within range that are interested in downloading. In fact, asshown in 1(a), we assume that the potential downloaders form a tree with Nchildren nodes at each level. For given N (say N = 2), and for a maximumtolerable latency T (say T = 30s), we want to compute the maximum numberof customers served by a single feed.

We will use a very simplified analytic model, leveraging the data reportedearlier in this section. The model depicted in Figure 19 assumes layer by layersynchronization in the distribution tree. We will further simplify by assumingasynchronous pull operation (as in BitTorrent and CarTorrent).

In Figure 20, we show the originator A (connected to 3G). We assume N =2. Users B and C download from A. The download cycle = 3s. First B connectsto A and downloads at max speed = 723Kbps, for an interval t = 1s. Then, Cconnects to A and also downloads @ 723Kbps (the maximum peak rate whenDH5 packet is used), for a total interval t = 1s. For another 1 second there isno download (Note: at least 2 seconds are used for lower level transfer). Thenthe cycle resumes. Each child has downloaded at the nominal rate 241Kbps. Itis easy to see that this procedure applies recursively to the lower levels of thetree. For example, the children of B, say B1 and B2 download from B duringtime slot 2 and 3. And so on. Each peer must buffer the video stream receivedover a 3 sec cycle, i.e. 723Kbits or 90.38KB.

Each cycle takes 2 seconds. Thus, if the latency must be less than 30 seconds,the depth of the tree must be < 10 and the number of video-fed nodes is up to1000. There is no synchronization requirement as each peer at the lower layersrequests data asynchronously, when it needs it. The peer will be able to getthe upper peer’s attention when the latter is done downloading from its ownupstream peer or uploading the peer’s sibling.

With the above model, each originator (i.e. seed) can feed up to 1000 nodes.This represents a major relief for the 3G network. In fact, if there is a crowdof 300,000 people at the stadium and 10% are interested in the download, only30 seeds are required, and thus only 30 3G simultaneous video connections - areasonable proposition for a 3G network. At issue of course is the incentive forthe seed to participate in the download. A possible solution would be to rotate

24

0

50

100

150

200

250

300

350

400

1 2 3 4 5

Thr

ough

put(

kbps

)

Path Length (hops)

Throughput vs. Path Length (DH5)

ScatternetSOBP(Sync)

Analysis(Sync)SOBP(Async)

Analysis(Async)

Figure 21: Throughput vs. Path Length(DH5)

0

200

400

600

800

1000

1 2 3 4 5

Thr

ough

put(

kbps

)

Path Length (hops)

Throughput vs. Path Length (3-DH5)

ScatternetSOBP(Sync)

Analysis(Sync)SOBP(Async)

Analysis(Async)

Figure 22: Throughput vs. Path Length(3-DH5)

the seed role in the tree among 3G subscribers. Then, assuming that 10% cellusers subscribe to 3G, the load will be shared among 100 users, again a veryreasonable proposition. However, at each turnover both old and new seed mustreceive the 3G video simultaneously for 30 s in order to avoid data loss.

For simulation, we use scatternet, synchronous SOBP, and asynchronousSOBP models as Figure 18 - 20. As depth increases, throughput of each levelreduced as a half of upper level for synchronous SOBP case and scatternet. But,asynchronous SOBP case keep same throughput for every level. In scatternet,all links are always connected and each node re-transfer video stream to thechild nodes after receiving from its parent as Figure 18. In synchronous SOBP,odd level links and even level links are used for different stages because it usespiconet only. For odd level link stage, only odd level links are used and formpiconets as Figure 19 (a). for even level link stagem, only odd level links areused and form piconets as Figure 19 (b). In asynchronous SOBP, each nodeconnect 1/3 of time with parent, left child, and right child, each.

Figure 21 and 22 are simulation results of video streaming over Scatternetand SOBP, when using DH5 packet and 3-DH5 packet, respectively. WhenPath Length is 1, that use only level 1 and form only one piconet for scatternetand SOBP cases. So, results are same. As path length increases, SynchronousSOBP shows better performance than scatternet. It will more parallelized aspath length increases and shows better performance. Whereas, in scatternet, aspath length increase, long path increase much switching time of piconets anddecrease throughput. Asynchronous SOBP shows somewhat less performancewhen path length is 2 because level 2 nodes do not have a child node, thereforeparallelization is limited. Whereas, path length is longer than 2, AsynchronousSOBP degrade less than other cases and shows the best performance.

25

Figure 23: Video Streaming Demo Setting

OBPApplication

PingInformation

MplayerApplication

MplayerVideo

Output

Figure 24: Video Streaming Demo Application

26

8 Video Streaming over SOBP

Dual video streaming are performed over SOBP as Figure 23. Two laptops areacting as master node of each piconet. One laptop is acting as a slave node.This nodes are periodically change its master with SOBP application. Masternodes connect to this slave (connection lasts for 5 seconds) and then disconnectthis slave after 5 seconds duration. For continuous video streaming duringdisconnected state, video stream data are buffered when link is connected.

Figure 24 shows linux windows that performs video streaming over SOBP.SOBP application is running in upper left window. It continuously changeits stage (Original stage and Visit stage) as Figure 23.This application doesconnection and disconnection. Bottom left windows shows current connectivity.When connected, ping information shows the delay, but when not connected,ping shows disconnection state. Upper right windows runs Mplayer [17] that ispopular video player for linux environment and support most of video streams.Bottom right windows shows the real video stream output generated by Mplayer.

9 Conclusion

In this paper, we presented several approaches to interconnect Bluetooth Pi-conets without using a permanent Scatternet in mobile environments. OverlaidBluetooth Piconets (OBP) shows resilience to mobility compared to traditionalScatternet and produces significantly higher throughput. Simplified BluetoothPiconets (SOBP) can be used in the temporary static state and shows higherthroughput than that of OBP. Scatternet requires Scatternet formation and ref-ormation as nodes are moving. OBP and SOBP creates virtual Scatternets thatdo not require persistent connections. Even more, OBP always uses multipleone-to-one connections therefore routing protocol is not needed. Thus, it is verywell suited for mobile environments. OBP and SOBP use only piconet connec-tions and they are applicable to all currently available Bluetooth devices evenif they do not support Scatternet.

With these features, OBP and SOBP can support video streaming. Analyti-cal performance of SOBP is comparable to that of single piconet and simulationresult of SOBP shows better performance than scatternet. With demonstration,we show that when SOBP is used, dual video streams can be played simultane-ously with normal video player.

We plan to implement mobile Point-to-Point (P2P) based video streaming inour future work. Bluetooth nodes can move individually and want to share videostream with each other. To supporting mobile P2P video streaming, efficientdiscovery and connection method are needed.

References

[1] Comvu - live mobile video broadcasting & user-generated video sharing.http://www.comvu.com.

27

[2] Sopcast - free internet iptv. http://www.sopcast.org/.

[3] The network simulator - (ns-2). http://www.isi.edu/nsnam/ns/, Last ac-cessed: December 2004.

[4] UCBT simulator. https://www.ececs.uc.edu/ cdmc/ucbt/, Last accessed:December 2004.

[5] S. Basagni and C. Petrioli. A scatternet formation protocol for ad hocnetworks of bluetooth devices. In IEEE Vehicular Technology Conference(VTC).

[6] Bluetooth specification v2.0, 2004. Bluetooth SIG.

[7] T. Camp, J. Boleng, and V. Davies. A survey of mobility models for adhoc network research. In IEEE Mobicom, 2002.

[8] L.-J. Chen, R. Kapoor, M. Y. Sanadidi, and M. Gerla. Enhancing BluetoothTCP throughput via link layer packet adaptation. In ICC, Paris, France,June 2004.

[9] K. Falls. A delay-tolerant network architecture for challenged internets. InACM SIGCOMM, 2003.

[10] P. Hui, A. Chaintreau, J. Scott, R. Gass, J. Crowcroft, and C. Diot. Pocketswitched networks and human mobility in conference environment. In SIG-COMM Workshops, 2005.

[11] P. Juang, H. Oki, Y. Wang, M. Maronosi, L. Peh, and D. Rubenstein.Energy-efficient computing for wildlife tracking: Design tradeoffs and earlyexperiences with zebranet. In ASPLOS, 2002.

[12] S. Jung, A. Chang, and M. Gerla. Comparison of bluetooth interconnec-tion methods using blueprobe. In The Second International Workshop OnWireless Network Measurement (WiNMee), 2006.

[13] S. Jung, A. Chang, and M. Gerla. Performance comparison of overlaidbluetooth piconets (obp) and bluetooth scatternet. In IEEE Wireless Com-munications and Networking Conference (WCNC), 2006.

[14] S. Jung, C. K. Kallo, M. Gerla, and M. Brunato. Decentralized optimizationof dynamic bluetooth scatternets. In Proceedings of the Second AnnualInternational Conference on Mobile and Ubiquitous Systems: networkingand services (MobiQuitous), 2005.

[15] C. K. Kallo, S. Jung, L.-J. Chen, M. Brunato, and M. Gerla. Through-put, energy and path length tradeoffs in bluetooth scatternets. In IEEEInternational Conference on Communications (ICC), 2005.

[16] C. Law, A. K. Mehta, and K.-Y. Siu. A new Bluetooth scatternet formationprotocol. Mobile Networks and Applications, 8:485–498, October 2003.

28

[17] Mplayer movie player, http://www.mplayerhq.hu.

[18] A. Nandan, S. Das, G. Pau, M. Sanadidi, and M. Gerla. Cooperativedownloading in vehicular ad hoc networks.

[19] C. Petrioli, S. Basagni, and I. Chlamtac. Bluemesh: Degree-constrainedmulti-hop scatternet formation for bluetooth networks. In mobile Networksand Applications, 2004.

[20] R. Shah, S. Roy, S. Jain, and W. Brunette. Data mules: Modeling a three-tier architecture for sparse sensor networks. In IEEE SNPA Workshop,May 2003.

[21] I. Stojmenobic. Dominating set based bluetooth scatternet formation withlocalized maintenance. In Proceedings of the 16th International Paralleland Distributed Processing Symposium, 2002.

29