[PPT] - Data Elevators Applying the Bundle Protocol in Delay Tolerant PowerPoint Presentation

SLIDE 1

Data Elevators

Applying the Bundle Protocol in Delay Tolerant Wireless Sensor Networks

Wolf-Bastian P¨

ttner, Felix B¨

usching, Georg von Zengen, Lars Wolf

IEEE MASS 2012, 2012-10-09

SLIDE 2

Introduction Bundle Protocol Data Elevator Network Capacity Conclusion

Motivation

ZebraNet

(a) (b) (d) (c)

Vineyard Computing SeNDT

Observation

Delay Tolerance is widely used (and needed) in sensor network research

Wolf-Bastian P¨

ttner | Data Elevators | 2

SLIDE 3

Introduction Bundle Protocol Data Elevator Network Capacity Conclusion

Common Requirements

Measurement

Periodic sampling of sensor values

Networking

Multi-hop data delivery Disrupted links, changing topologies Delay is not important, reliability is    Store, carry and forward

Hardware

Long lifetime Minimal installation effort Few maintenance cycles → Low-power → Wireless → Robust

Wolf-Bastian P¨

ttner | Data Elevators | 3

SLIDE 4

Introduction Bundle Protocol Data Elevator Network Capacity Conclusion

Wireless Sensor Networks (WSNs)

Wireless Sensor Networks

Multi-hop wireless Battery powered

Wireless Sensor Nodes

Based on microcontrollers IEEE 802.15.4 radios

App. 16 kB RAM, app. 128 kB ROM

Low-power hardware Storage (flash, SD, ...) INGA T-Mote Sky

Wolf-Bastian P¨

ttner | Data Elevators | 4

SLIDE 5

Introduction Bundle Protocol Data Elevator Network Capacity Conclusion

Outline

Introduction Bundle Protocol in Delay Tolerant Wireless Sensor Networks Data Elevator Application Scenario Capacity of Delay Tolerant Wireless Sensor Networks Conclusion

Wolf-Bastian P¨

ttner | Data Elevators | 5

SLIDE 6

Introduction Bundle Protocol Data Elevator Network Capacity Conclusion

Protocols for Wireless Sensor Networks

Predominant WSN Protocols

6LoWPAN: IPv6 over low-power WPAN Contiki’s and TinyOS’ proprietary protocols → Not delay tolerant (not store, carry and forward)

Store, Carry and Forward Protocols

ZebraNET (non-standardized) Vineyard Computing (non-standardized) Seal-2-Seal (non-standardized) Bundle Protocol (RFC 5050)

Wolf-Bastian P¨

ttner | Data Elevators | 6

SLIDE 7

Introduction Bundle Protocol Data Elevator Network Capacity Conclusion

Benefits and Drawbacks of Standard Protocols

Benefits

Seamless integration Lower entry barrier Generic solutions

Drawbacks

Not optimized for use case Higher overhead

Benefits of the Bundle Protocol

Flexibility: Variable length header fields, extension blocks, etc. Overlay Protocol: Works on top of heterogeneous technologies Well suited: Designed for unstable links and changing topologies Q: Is the Bundle Protocol too heavy for nodes?

Wolf-Bastian P¨

ttner | Data Elevators | 7

SLIDE 8

Introduction Bundle Protocol Data Elevator Network Capacity Conclusion

Bundle Protocol Overhead Comparison

Overhead [Bytes]

802.15.4 802.15.4 802.15.4 802.15.4 802.15.4 IPv6 TCP 6LoWPAN TCP 6LoWPAN TCP CL Bundle Protocol Bundle Protocol 6LoWPAN 6LoWPAN UDP UDP

TCP / IPv6 UDP / 6LoWPAN TCP / 6LoWPAN BP / UDPCL / 6LoWPAN BP / TCPCL / 6LoWPAN

802.15.4 IPv6 UDP

UDP / IPv6 57 69 23 38 44 61

802.15.4 Bundle Protocol

BP / IEEE 802.15.4 CL 31

802.15.4 RIME

Contiki‘s RIME 19

IEEE 802.15.4 maximum frame size is 127 bytes

A: Protocol overhead is higher but manageable

Wolf-Bastian P¨

ttner | Data Elevators | 8

SLIDE 9

Introduction Bundle Protocol Data Elevator Network Capacity Conclusion

Bundle Protocol Complexity Comparison

160 180 200 220 240 260 280 300 10 20 30 40 50 60 70 Parsing of Incoming Data [us] Payload Length [bytes] 6LoWPAN HC06 / UDP Bundle Protocol CBHE

5000 10000 15000 20000 25000 8 16 24 32 SDNV Operations per Second Integer Size [bit] Decode Encode

Run on INGA at 8 MHz

A: Computational complexity is comparable

Wolf-Bastian P¨

ttner | Data Elevators | 9

SLIDE 10

Introduction Bundle Protocol Data Elevator Network Capacity Conclusion

How can we implement the Bundle Protocol on nodes?

Literature

Bundle Protocol as overlay protocol over 6LoWPAN

Our Approach on the Nodes: µDTN

BP in IEEE 802.15.4 data frames Cross-layer, avoiding layers 3 and 4 Implementation based on Contiki OS

Our Approach on the PC

IEEE 802.15.4 radio attached to PC IBR-DTN software extension to handle radio

Wolf-Bastian P¨

ttner | Data Elevators | 10

SLIDE 11

Introduction Bundle Protocol Data Elevator Network Capacity Conclusion

Data Elevator Application Scenario

Opening Question

How can we get temperature readings from the rooftop into our lab?

Concept

Node with sensor on rooftop Elevator is data mule Delay tolerant network

Setup

1 sensor, 3 relays, 1 sink µDTN with RAM storage

Rooftop #1 #2 #3 #4 Elevator: 1

st

14thFloor #5 Building A Building B 15

th Floor

3

rdFloor

14

thFloor

Temperature Sensor Wolf-Bastian P¨

ttner | Data Elevators | 11

SLIDE 12

Introduction Bundle Protocol Data Elevator Network Capacity Conclusion

Evaluation: Temperature and Delay (Weekend)

1 2 3 4 5 6 7 8 2 4 6 8 10 12 14 16 18 20 22 3 6 9 12 15 18 21 24 Bundle Delay [h] Temperature [Degree Celsius] Time of Day [h] Bundle Delay Temperature

Wolf-Bastian P¨

ttner | Data Elevators | 12

SLIDE 13

Introduction Bundle Protocol Data Elevator Network Capacity Conclusion

Evaluation: Delay Distribution (Weekend)

10 20 30 40 50 60 70 80 90 100 >0-5 >10-15 >20-25 >30-35 >40-45 >50-55 >60-65 >70-75 >80-85 >90-95 >100 Number of Occurrences [%] Bundle Delay Bins [m] Delay Distribution Accumulated Delay Distribution

Wolf-Bastian P¨

ttner | Data Elevators | 13

SLIDE 14

Introduction Bundle Protocol Data Elevator Network Capacity Conclusion

DT-WSN Capacity Model

Storage(Capacity((

SCap,Send SCap,Recv

Bundles(in(Storage(

SSend,i SRecv,i

Sender( Receiver(

BundleRatei Channel Capacity: Transmitted Bundles: Storage Sender: Storage Receiver: Ci,j = Durationi · BundleRatei Ti = min(SSend,i, Ci,j) SSend,i = min(SSend,i−1 − Ti−1 + Ni, SCap,Send) SRecv,i = min(SRecv,i−1 + Ti, SCap,Recv)

Wolf-Bastian P¨

ttner | Data Elevators | 14

SLIDE 15

Introduction Bundle Protocol Data Elevator Network Capacity Conclusion

Evaluation: Capacity Model

10 20 30 40 50 60 70 80 50 100 150 200 250 300 Permanently Lost Bunldes [%] Sample Interval [s] Storage: 100 Bundles Storage: 300 Bundles Storage: 500 Bundles unlimited

Wolf-Bastian P¨

ttner | Data Elevators | 15

SLIDE 16

Introduction Bundle Protocol Data Elevator Network Capacity Conclusion

Conclusions

Wolf-Bastian P¨

ttner

poettner@ibr.cs.tu-bs.de http://www.ibr.cs.tu-bs.de/projects/mudtn

Protocols

Standard protocols are generic solutions to common problems BP is de facto standard in DTNs and should be in DT-WSNs

µDTN

Bundle Protocol implementation for Contiki Overhead is comparable to 6LoWPAN Integration into existing DTNs via transparent gateway nodes

Data Elevator

Data is delivered with delay but without loss

Wolf-Bastian P¨

ttner | Data Elevators | 16

SLIDE 17

Wolf-Bastian P¨

ttner | Data Elevators | 17

SLIDE 18

Introduction Bundle Protocol Data Elevator Network Capacity Conclusion

Evaluation: Temperature and Delay (Weekday)

1 2 3 4 5 6 7 8 2 4 6 8 10 12 14 16 18 20 22 3 6 9 12 15 18 21 24 Bundle Delay [h] Temperature [Degree Celsius] Time of Day [h] Bundle Delay Temperature

Wolf-Bastian P¨

ttner | Data Elevators | 18

SLIDE 19

Introduction Bundle Protocol Data Elevator Network Capacity Conclusion

Evaluation: Delay Distribution (Weekday)

10 20 30 40 50 60 70 80 90 100 >0-5 >10-15 >20-25 >30-35 >40-45 >50-55 >60-65 >70-75 >80-85 >90-95 >100 Number of Occurrences [%] Bundle Delay Bins [m] Delay Distribution Accumulated Delay Distribution

Wolf-Bastian P¨

ttner | Data Elevators | 19