On Decentralized In-Network Aggregation in Real-World Scenarios - - PowerPoint PPT Presentation

▶
on decentralized in network aggregation in real world
SMART_READER_LITE
LIVE PREVIEW

On Decentralized In-Network Aggregation in Real-World Scenarios - - PowerPoint PPT Presentation

Aug 18, 2022 235 likes •459 views

On Decentralized In-Network Aggregation in Real-World Scenarios with Crowd Mobility M. Gregorczyk, T. Pazurkiewicz, K. Iwanicki University of Warsaw DCOSS 2014, Marina Del Rey, CA, USA, May 26th, 2014 Monitoring Crowds Image source:

slide-1
SLIDE 1

On Decentralized In-Network Aggregation in Real-World Scenarios with Crowd Mobility

  • M. Gregorczyk, T. Pazurkiewicz, K. Iwanicki

University of Warsaw

DCOSS 2014, Marina Del Rey, CA, USA, May 26th, 2014

slide-2
SLIDE 2

Monitoring Crowds

  • Our interest:

– utilize sensing, processing,

and communication capabilities of low-power wearable devices

– to monitor the behavior of

crowds from the inside.

  • Envisioned effect:

– Deeper understanding of

crowd behavior

– More informed planning (e.g.,

transportation, infrastructure)

– Ability to manage and control

crowds in real time

Image source: http://en.wikipedia.org/wiki/File:Crowd_in_street.jpg

slide-3
SLIDE 3

In-Network Aggregation

adapted from image: http://www.clker.com/clipart-smaller-crowd-rdc.html

  • Problem: data deluge.
  • One of the solutions:

decentralized in-network aggregation:

–

Each node senses its surroundings.

–

It communicates its observations via low-power radios to other nearby nodes.

–

The nodes collaboratively aggregate the readings to reduce the traffic volume to an external monitoring site.

  • We target basic aggregates:

–

AVG, COUNT, MAX, MIN, SUM

slide-4
SLIDE 4

Aggregation in Sensornets

Volumes of Aggregation Techniques

slide-5
SLIDE 5

Aggregation in Sensornets

Volumes of Aggregation Techniques Structured Aggregation Techniques Unstructured Aggregation Techniques

slide-6
SLIDE 6

Aggregation in Sensornets

Volumes of Aggregation Techniques Structured Aggregation Techniques Unstructured Aggregation Techniques Gradual Variance Reduction (GVAR) Order- and Duplicate-Insensitive Sketches (ODIS)

slide-7
SLIDE 7

Gradual Variance Reduction (GVAR)

  • To compute a global average:

–

Each node periodically selects another neighbor at random.

–

It exchanges its local value with the neighbor's local value.

–

Both nodes set their local values to the average of two values.

–

Over time, the local node values converge to the global average.

  • To count the number of nodes:

–

One node sets its initial local value to 1.

–

Others set their values to 0.

v1 v2 v1

  • 1. Select random

neighbor

  • 2. Exchange local values

v1

v2 v1

v2

  • 3. Average local values

va va va←(v1+v2)/2 va←(v1+v2)/2

  • 4. Repeat periodically
  • 3. Average local values

v2

slide-8
SLIDE 8

Order- & Duplicate-Insensitive Sketches (ODIS)

  • To count the number of nodes:

–

All nodes maintain local 16-bit bitmasks (initially all zeroes).

–

Each node sets one bit in the bitmask with the index drawn from a geometric distribution.

–

Repeatedly exchanges its bitmask with its neighbors OR-ing the received bitmasks with its own.

–

When all bitmasks have converged, the number of nodes is estimated as:

  • 1.2928 • 2 pos0, where pos0 is

the position of the least significant 0.

½ ¼ ⅛ ½ ½4

... 1 ...

½i+1

2 i 3 more probable less probable

½ 1 1 1

... 1 ... 2 i 3 pos0 estimated count: 1.2928 • 22 = 5.17 probability of selecting bit

slide-9
SLIDE 9

Base Simulated Scenario

Done in OMNeT++ with MiXiM extensions for wireless sensor networks. G1 = 333 nodes; G2 = 222 nodes; G3 = 444 nodes a node's aggregate = COUNT of nodes in the node's connected component.

slide-10
SLIDE 10

GVAR Results

slide-11
SLIDE 11

ODIS Results

slide-12
SLIDE 12

Improving Communication

  • To improve communication we adapt the Trickle algorithm for code

propagation to ODIS aggregate computation.

–

Normally, broadcast your bitmask randomly within every Tmax time units.

–

But, when your bitmask changes significantly shrink the interval to Tmin.

–

Each subsequent interval doubles up to Tmax.

–

Suppressing broadcasts when several similar bitmasks are received.

Effect: When the local bitmast quickly compute aggregates while minimizing traffic when the system is quiescent.

slide-13
SLIDE 13

ODIS with New Communication

slide-14
SLIDE 14

Improving Accuracy

  • Use many instances of sketches: smoothing.
  • Use more efficient sketches:

–

parameterless sketches:

  • can be used out-of-the-box, but
  • are not the most efficient ones (wrt. error / #bits).

–

parametrized sketches:

  • are very efficient, but
  • their accuracy depends on the final result.

– Solution: pipelining a parameterless sketch with a parametrized

  • ne.

Effect: The accuracy improves for the same number of bits.

slide-15
SLIDE 15

ODIS with Improved Accuracy

slide-16
SLIDE 16

Back to Real-World Experiments

  • We implemented the

algorithms as an aggregation service for TinyOS.

  • We conducted several real-

world deployments of the service.

– Up to 177 nodes.

  • Mostly on eZ430 Chronos

smart watches.

Image source: http://electronicdesign.com/content/14978/59382_fig_01.jpg

slide-17
SLIDE 17

Sample Scenarios

G1 = 20 nodes; G2 = 19 nodes; G3 = 15 nodes

slide-18
SLIDE 18

Results (Scenario 1)

G1 = 20 nodes; G2 = 19 nodes; G3 = 15 nodes

slide-19
SLIDE 19

Results (Scenario 2)

G1 = 20 nodes; G2 = 19 nodes; G3 = 15 nodes

slide-20
SLIDE 20

Conclusions

  • To be applied in real-world crowd-monitoring

scenarios, decentralized in-network aggregation algorithms for sensornets require considerable adaptation.

  • Applications have to be prepared that the

aggregates they see may exhibit errors.

  • We may need to revisit some of their

assumptions.

  • (Conducting real-world crowd-monitoring

deployments is challenging.)

slide-21
SLIDE 21

Thank You

Questions?

Supported by the (Polish) National Science Centre (NCN) within the SONATA programme under grant no. DEC-2012/05/D/ST6/03582.