{[ promptMessage ]}

Bookmark it

{[ promptMessage ]}

slides.0317.2011 - TAG A Tiny Aggregation Service for...

Info icon This preview shows pages 1–10. Sign up to view the full content.

View Full Document Right Arrow Icon
TAG: A Tiny Aggregation Service for Ad-Hoc Sensor Networks Samuel Madden, et. Al OSDI 2002 Presented by: Curtis Wang 1 12/9/11
Image of page 1

Info icon This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
What is Tiny AGgregation? Service that allows for efficient, distributed aggregation in wireless sensor networks (WSN) Provides a SQL-like declarative interface Ultimately, want to be able to treat WSN like a database, where each mote is equivalent to a tuple Significantly reduces communication between motes 2 12/9/11
Image of page 2
Motivation Most sensor network applications use aggregations instead of raw sensor data. Sending data using radio is energy- intensive compared to computation. Centralized approach requires lots of data transmission Transmitting one bit of data is equivalent to 800 instructions Programming in TinyOS is difficult. 3 12/9/11
Image of page 3

Info icon This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Design Considerations Limited battery Ad-hoc networks: Changing topology High link-level losses during radio communication (~20%) Bandwidth limitations 4 12/9/11
Image of page 4
Query Model SQL-style queries over a single table (called sensors). Epoch – time interval for computing an aggregate record Differences from SQL: Stream of values vs. single value; <group id, aggregate value> per group per epoch 5 12/9/11 Example Query: Monitoring occupancy in conference rooms on a certain floor General Query
Image of page 5

Info icon This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Tree-based Routing A B C D F E 6 Base Station Propagate Query: nodeID, level Set level Set parent Rebroadca st 12/9/11 Figure from Author’s Slides
Image of page 6
Structure of Aggregates Aggregate consists of three functions: Partial State Record: Intermediate record Ex: AVERAGE; <x> = <SUM,COUNT > Merging function: f(<x>,<y>) = <z> f(<S1, C1>, <S2, C2>) = < S1 + S2, C1 + C2> Initializer: i(x) = <x> i(x) = <x, 1> Evaluator: e(<x>) = y e(<S, C>) = S/C 7 12/9/11
Image of page 7

Info icon This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Taxonomy of Aggregates TAG is not restricted to SQL Aggregates Classify aggregates according to relevant properties 12/9/11 8 Property Examples What does it affect? Duplicate Sensitivity MAX/MIN: No AVERAGE: Yes Routing redundancy Exemplary/Summary MAX/MIN: Exemplary COUNT, SUM: Summary Effects of loss Monotonicity COUNT / SUM: Monotonic MEDIAN: Not monotonic Hypothesis testing, Snooping Partial State Record Size MAX/MIN: O(1) MEDIAN: O(n) Performance of TAG
Image of page 8
Phases of TAG Two Phases: Distribution, Collection 9 12/9/11
Image of page 9

Info icon This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Image of page 10
This is the end of the preview. Sign up to access the rest of the document.
  • Spring '08
  • Gupta
  • Wireless sensor network, fault tolerance, Sensor Network Applications, partial state, Short Listen Problem, TinyOS packet w/

{[ snackBarMessage ]}