hardware-assisted ieee 802.15.4 transmissions and why to … · evaluate performance of 6lowpan...

Hardware-AssistedIEEE 802.15.4 Transmissions and

Why to Avoid ThemAndreas Weigel, Volker Turau

IDCS 2015September 2nd, 2015

TUHHTUHHInstitute of TelematicsInstitute of TelematicsHamburg University of TechnologyHamburg University of Technology

IntroductionIntroduction

What is this about?

� Wireless Sensor Networks / IoT

� Evaluate performance of 6LoWPANfragmentation for large datagrams inmulti-hop scenarios

� Simulation and testbed experiments

CoAP

UDP

IPv6

6LoWPAN

RPL

802.15.4

Andreas Weigel Hardware-Assisted IEEE 802.15.4 Transmission and Why to Avoid ThemAndreas Weigel Hardware-Assisted IEEE 802.15.4 Transmission and Why to Avoid Them 11


Hardware-Assisted Transmissions

� ATmega256RFR2 “Extended Operating Mode” hashardware support for...

� automatic backoff and CCA; automatic retries (TX_ARET)� automatic acknowledgements (RX_AACK)

� Nice-to-have, because of:� less complex code (LOC: 371 vs 1847)� smaller binaries (5280 bytes ROM and 578 bytes RAM less)



Simulation vs. Testbed 0 21FD 21FE

21FC

21FF

2201

2204

2205

2214

2217

2203

2219

2206

221A

100 400 1,20020

40

60

80

100

Payload [Byte]

PR

R[%

]

Assembly Direct

100 400 1,20020

40

60

80

100

Payload [Byte]



6LoWPAN – Forwarding of FragmentedDatagrams

v1 l0 v2 l1 v3 l2 v4m1m2

...

mnm1m2

...

mnm1m2

...

mn

Assembly Mode

v1 l0 v2 l1 v3 l2 v4

m1m2

...mn

m1m2

...

mn

m1m2

...

mn

Direct Mode



Extended Operating Mode

2

1

2

1

no-Rx-while-TX

Time

· · ·

· · ·

RX-while-TX

Time

Node

IDLE RX TX BO RX TX PENDING TX RADIO

Assumption: “no-RX-while-TX” property of Extended OperatingMode is responsible for losses!


Experimental SetupExperimental Setup

Approach

� Approaches to verify the suspicion� Analytical model� Adapt simulation model� Realtime observation of testbed

� Two wireless stacks for CometOS� AACK MAC (using Extended Operating Mode)� Software MAC (ported TinyOS mac layer)


Experimental SetupExperimental Setup

� Record sequence of MAC layer states of each node:� Data transport and time synchronization per wire

⇒ Instrument code to signal events via GPIO

1 2

3

4

RaspPi FRDM-K64F

PC Basestation

PC

Ethernet/TCP

UART

UART

Radio Link/Static RouteGPIO connection cable


EvaluationEvaluation

Direct Mode - AACK MAC

4000 4060 4120 4180 4240 4300 4360 4420 4480 4540 4600

4

3

2

1

Time [ms]

Node

IDLE RX TX BO TX RADIO



Direct Mode - Software MAC

0 45 90 135 180 225 270 315 360 405 450

4

3

2

1

Time [ms]

Node




Results - PRR

Direct

0

20

40

60

80

100

120

21.6

97P

RR

in%

AACK MAC Software MAC



Results - Fragment CountsSum of fragment transmissions of nodes 3 and 4

Software MAC AACK MAC

Requests (theo. max) 720 720Successful; Dest TX 518 0Successful; Dest non-TX 198 399Failure; Dest TX 82 424Failure; Dest non-TX 23 4Failure; Total 105 428



Simulation vs. Testbed revisited 0 21FD 21FC

21FE21FF 2201

2203

2205

22062204

2214 2217

2219

221A

100 400 1,20020

40

60

80

100

Payload [Byte]

PR

R[%

]

Assembly Direct Direct EOM

100 400 1,20020

40

60

80

100

Payload [Byte]



Conclusion

� Verified assumption in small testbed� Extended Operating Mode responsible for nearly all

fragment losses in multi-hop traffic scenarios� Usage of similar-working mechanisms not advisable (in

similar scenarios)


Hardware-AssistedIEEE 802.15.4 Transmissions and

Why to Avoid ThemAndreas Weigel, Volker Turau

IDCS 2015September 2nd, 2015

Andreas Weigel

Phone +4940 428 78 3746

e-Mail [email protected]

http://www.ti5.tuhh.de/staff/weigel

TUHHTUHHInstitute of TelematicsInstitute of TelematicsHamburg University of TechnologyHamburg University of Technology

mailto:[email protected]

http://www.ti5.tuhh.de/staff/weigel

MAC Config

minBe maxBe csmaBackoffs5 8 5

maxFrameRetries CCA mode CCA threshold7 0 -90 dBm


Results - PRR

Direct Direct-ARR

0

20

40

60

80

100

120

21.6

79.5

97 99.6

PR

Rin

%

AACK MAC Software MAC


Results - Fragment Counts

Software MAC AACK MAC

fragRequests (theo. max) 720.00 720.00fragSuccessDstTx 518.48 ± 6.4 0fragSuccessDstNonTx 198.38 ± 5.9 398.76 ± 16.75fragFailDstTx 81.96 ± 3.8 423.08 ± 14.58fragFailDstNonTx 23.44 ± 2.7 4.44 ± 0.64fragFailTotal 105.40 ± 6.5 427.52 ± 15.22


IDLE

TX BO

TX RF TX ACK

RXRXTX PENDINGTX PENDING

TXSTART TX

BOFAIL

RF TX END

RFTX

START

TX REQUEST

RX DONE

RF TX END

RF TX END

TXRETRY

RF RX START

TX FAIL,TX SUCCESS

TX REQUEST

RF RX START

RF TX END

IDLE

RX

TX BO

TX RADIO


IDLE

TX BO

TX RF TX WAIT TX ACK

TX DONE

RXRX

RX DLRX DL RX ACKRX ACK

RX TX PDRX TX PD

RX DL TX PDRX DL TX PDRX ACK TX PDRX ACK TX PD

RX DONE

RF TX END

TX STARTRX DONE

RF TX END

RF RX END

TX BO FAIL

RX DONE

TXRE

TRY

RFRX

END

RX DONE,RX DROPPED

RF TX END

RX DONE

RFTX

STAR

T

RFRX

START

TXBO

FAIL

TX START

RF TX END

RX DROPPED,RX DONE

TXRE

TRY

RF RX START

TX FAIL,TX SUCCESS

TXFAI

L,

TXSUC

CESS

TX RETRY

RF RX DROPPED,RX DROPPED

RX DONE,RX DROPPED,

RF RX DROPPED

RF TX START

RF RX START

TX BO FAIL

RX DONE,RX DROPPED,RF RX DROPPED

TX START

RF TX START

RF RX START

TX FAIL, TX SUCCESS,RX DROPPED, RX DONE

TXSTART

RF RX START



Direct-ARR Mode - AACK MAC (Success)

20000 20060 20120 20180 20240 20300 20360 20420 20480 20540 20600

4

3

2

1

Time [ms]

Node



Direct-ARR Mode - AACK MAC (Failure)

8000 8060 8120 8180 8240 8300 8360 8420 8480 8540 8600

4

3

2

1

Time [ms]

Node



hardware-assisted ieee 802.15.4 transmissions and why to … · evaluate performance of 6lowpan...

Documents