Ling-Jyh Chen1, Chih-Wei Sung2, Hao-Hsiang Hung1, Tony Sun3, Cheng-Fu Chou2

1Academia Sinica2National Taiwan University

3PacketMotion Inc.

Path Capacity Estimation in Time-Slotted Wireless

Networks

OutlineIntroductionOverview of Bluetooth SystemTime-Slotted Path Capacity Estimation

Comparison of Popular Capacity Estimation Tools

AnalysisTSProbe

Proposed ApproachAlgorithm

EvaluationsConclusion

IntroductionKnowing the end-to-end path capacity is

important for wireless applications

Time-slotted wireless techniques are increasingly popularWiMAX (IEEE 802.16), Bluetooth (IEEE

802.15.1) and ZigBee (IEEE 802.15.4)

An effective path capacity estimation tool in time-slotted wireless systems is still lacking

IntroductionMultimedia stream adaptationRate controlOverlay designQoS applications, etc.

Overview of Bluetooth System

master

slave 1

slave 2625 µ sec

0 1 2 3

1600 hops/sec

4 5

DH1/DM1

DH3/DM3

DH5/DM5

1 2 3 4 5 6 time

Data Packet

ACK Packet

Bluetooth – ACL Mode

DH : Data High Mode ACL: Asynchronous Connectionless Link

DM : Data Medium Mode FEC: Forward Error Correction

Mode FEC

Packet SymmetricThroughput

(Kbps)

AsymmetricThroughput

(Kbps)Size

(bytes)Length(slots)

DM1 yes 17 1 108.8 108.8 108.8

DM3 yes 121 3 258.1 387.2 54.4

DM5 yes 227 5 286.7 477.8 36.3

DH1 no 27 1 172.8 172.8 172.8

DH3 no 183 3 390.4 585.6 86.4

DH5 no 339 5 433.9 723.2 57.6

Overview of Bluetooth System Sender Receiver

BNEP: Bluetooth Network Encapsulation Protocol

L2CAP: Link Layer Control and Adaptation Layer Protocol

Networking Applications

TCP/UDP

IP

BNEP (3 bytes)

L2CAP (4 bytes)

Bluetooth Baseband

Bluetooth Radio

Networking Applications

TCP/UDP

IP

BNEP (3 bytes)

L2CAP (4 bytes)

Bluetooth Baseband

Bluetooth Radio

Evaluation of Capacity Estimation ToolsThe setting of capacity estimation tools

AdHoc ProbeProbing packet size = 1500 bytes5 probes per second

Pathrate and IperfDefault setting

One-hop Bluetooth testbedAverage of results of 10 experiment runs

Comparison – Average Link Capacity Estimates

Unit: Kbps

ToolBluetooth Packet Type

DH5 DH3 DH1 DM5 DM3 DM1

Theoretical 723.2 585.6 172.8 477.8 387.2 108.8

AdHoc Probe 643 541 138 459 370 84

Pathrate 608 537 137 432 360 84

Iperf - UDP 541 469 129 415 328 82

Iperf - TCP 539 462 126 397 321 80

Comparison – Average Capacity Estimation Time

Unit:mm’ss

ToolBluetooth Packet Type

DH5 DH3 DH1 DM5 DM3 DM1

AdHoc Probe 0’40 0’40 0’40 0’40 0’40 0’40

Pathrate 18’33 18’41 18’38 18’41 18’43 0’34

Iperf - UDP 0’12 0’12 0’17 0’13 0’13 0’22

Iperf - TCP 0’11 0’12 0’16 0’12 0’12 0’20

Analysis

The effective link capacity is dependent on the employed packet size

C

L

PE

LHP

Maximum Data Throughput

Capacity Estimation

Probing Packet Size

Baseband Packet Size

Bluetooth Stack Header

Analysis – AdHoc Probe Estimates

Analytical Results Simulation Results

Unit: Kbps Unit: Kbps

ModeProbing Packet Size (bytes)

1500 1300 1100 900

DH5 640.0 693.3 586.7 640.0

DH3 533.3 520.0 502.9 576.0

DH1 171.4 169.8 171.7 169.4

DM5 457.1 462.2 469.3 383.9

DM3 369.2 378.2 352.0 360.0

DM1 107.9 108.1 106.7 106.7

ModeProbing Packet Size (bytes)

1500 1300 1100 900

DH5 640.0 693.3 586.7 640.0

DH3 533.3 520.0 502.9 576.0

DH1 171.4 169.8 171.7 169.4

DM5 457.1 462.2 469.3 383.9

DM3 369.2 378.2 352.0 360.0

DM1 107.9 108.1 106.7 106.7

Analytic AdHoc Probe Results

TSProbe: Concept

LHP

LHP

k

L

C

P

ER

L-H 2L-H 3L-H 4L-H 5L-H P-H

k/5k/4k/3k/2k/1R=

0

TSProbe: Proposed Approach

LHp

LHp ]0[)1]1[(

1)1]0[(]0[

LHp

LHp

1]0[]1[

LHp

LHp

P[1] P[0]

Algorithm

EvaluationsSimulations

NS-2 v2.28UCBT v0.9.8.2

ExperimentsPacket-pairs: 100 samples, 4 packet-pairs per

second10 TSProbe runs on each link mode

Initial ParametersMTU = 1500 bytesp_start = 1500 (bytes)Init_range = 64 (bytes)

Evaluation – Bluetooth Topologies

Simulation Results – One-hop

Mode

Peak[0] Peak[1] ResultsTheoretical

CP[0] E[0] C[0] P[1] E[1] C[1] L H C

DH5 1349 719.0 722.8 1010 718.1 723.1 339 7 722.92 723.2

DH3 1457 583.8 585.6 1274 582.2 585.4 183 7 585.51 585.6

DH1 1478 171.5 172.3 1451 172.7 173.5 27 7 172.89 172.8

DM5 1337 476.0 478.5 1113 475.3 478.2 224 7 478.35 477.8

DM3 1445 385.8 387.7 1324 385.3 387.3 121 7 387.50 387.2

DM1 1489 108.7 109.2 1472 108.9 109.5 17 7 109.33 108.8

Simulation Results – Multi-hop

ModePiconet (Kbps) Scatternet (Kbps)

1 hop 2 hops 2 hops 3 hops 4 hops

DH5 722.92 361.46 351.13 361.46 361.46

DH3 585.51 292.76 266.08 292.75 292.75

DH1 172.89 86.45 83.79 86.44 86.44

DM5 478.35 239.18 244.76 239.18 239.18

DM3 387.50 193.75 180.56 193.75 193.75

DM1 109.33 54.67 51.70 54.66 54.66

Experiment Results – Multi-hop

ModePiconet (Kbps) Scatternet (Kbps)

1 hop 2 hops 2 hops 3 hops 4 hops

DH5 726.03 343.69 308.91 271.82 246.91

DH3 594.02 282.19 262.08 237.61 217.62

DH1 139.61 78.47 82.93 67.99 29.13

DM5 479.39 228.72 237.32 165.42 158.68

DM3 387.56 161.86 191.71 150.12 149.85

DM1 X X X X X

Internet Experiment Scenario

Internet

WorkstationSender

(linux1.cise.ntu.edu.tw)

Laptop 1Router

(218.161.124.93)(10.0.0.1)

Bluetooth Connectivity

Laptop 2Router

(10.0.0.2)

Robustness Experiments Testbed Experiments

Internet Experiments

Mode DH5 DH3 DH1 DM5 DM3 DM1

Success Rate 100% 100% 80% 100% 100% -

Time Consumption 15 min. 12 min. 10 min. 14 min. 13 min. X

Minimum Steps 27 24 20 26 23 x

Mode DH5 DH3 DH1 DM5 DM3 DM1

Success Rate 100% 80% 10% 90% 80% -

Time Consumption 17 min. 14 min. 10 min. 18 min. 12 min. X

Minimum Steps 27 24 20 26 23 x

ConclusionAn analytical model presents the

relationship between the link capacity estimation and channel utilization

TSProbe can adapt to the properties of link layer in time-slotted systems for estimating the end-to-end capacity

TSProbe is an accurate, robust and reliable capacity estimation tool

This study is also applicable to other time-slotted systems

Thanks for your patience!

http://www.iis.sinica.edu.tw/~cclljj/http://www.iis.sinica.edu.tw/~cclljj/

http://nrl.iis.sinica.edu.tw/http://nrl.iis.sinica.edu.tw/