Data-Centric Storage in Sensornets with GHT, A Geographic Hash Table

1
Data-Centric Storage in Sensornets with GHT, A
Geographic Hash Table

• Presented by Yan Lu
• Date 04/03/2003

2
Outline

• Background
• Existing Schemas
• Data-Centric Storage
• Performance
• Conclusion
• References

3
Background

• Sensornet
• ? A distributed sensing network comprised of a
  large number of small sensing devices equipped
  with
• processor memory radio
• ? Great volume of data
• Data Dissemination Algorithm
• ? Scalable
• ? Self-organizing
• ? Energy efficient

4
Observations/Events/Queries

• Observation
• ? Low-level output from sensors
• ? E.g. detailed temperature and pressure
  readings
• Event
• ? Constellations of low-level observations
• ? E.g. elephant-sighting, fire, intruder
• Query
• ? Used to elicit the event information from
  sensornets
• ? E.g. locations of fires in the network
• Images of intruders detected

5
Existing Schemas

• External Storage (ES)
• Local Storage (LS)
• Data-Centric Storage (DCS)

6
External Storage (ES)
7
ES Problems
8
Local Storage (LS)
9
Local Storage (LS)
10
Data-Centric Storage (DCS)

• Events are named with keys
• DCS provides (key, value) pair
• DCS supports two operations
• ? Put (k, v) stores v ( the observed data )
  according to the key k, the name of the data
• ? Get (k) retrieves whatever value is stored
  associated with key k
• Hash function
• ? Hash a key k into geographic coordinates
• ? Put() and Get() operations on the same key k
  hash k to the same location

11
DCS Example
(11, 28)
(11,28)Hash(elephant)
12
DCS Example
Get(elephant)
(11, 28)
(11,28)Hash(elephant)
13
DCS Example contd..
elephant
fire
14
Geographic Hash Table (GHT)

• Builds on
• ? Peer-to-peer Lookup Systems
• ? Greedy Perimeter Stateless Routing

GHT
GPSR
Peer-to-peer lookup system
15
Peer-to-peer Lookup System

• P2P Model
• ? Each user stores a subset of data
• ? Each user can access data of all the other
  users in the system
• ? More scalable than centralized / hierarchy
  networks
• Schema based on hashing techniques
• ? Content Addressable Networks (CAN)

16
Content Addressable Networks

• Dynamically partition entire coordinate space,
  every node owns its distinct zone
• Each node stores entire hash table, and adjacent
  zones information
• Data Storage
• Data Retrieval

(k,v)
17
Greedy Perimeter Stateless Routing

• The position of destination is known
• Nodes know neighbors positions
• Routing decision is made based on the position of
  the neighbors and a packets destination
• Greedy Forwarding
• Perimeter Forwarding

18
GPSR Greedy Forwarding
19
GPSR - void
20
GPSR Perimeter Forwarding
Right Hand Rule Each node to receive a packet
forwards the packet to the next link
counterclockwise about itself from the ingress
link
2
X
Z
3
1
Y
21
Home Node and Home Perimeter

• Hash function is ignorant of the placement of
  individual nodes in the topology
• Home Node
• ? Geographically nearest the destination
  coordinates of the hashed location
• Home Perimeter
• ? Uses GPSR perimeter mode
• ? Packet traverses the entire perimeter that
  enclose the destination, before returning to the
  home node

22
Problems

• Not robust enough
• ? Nodes could move (new home node?)
• ? Home nodes could fail
• Not scalable
• ? Home nodes could become communication
  bottleneck
• ? Storage capacity of home nodes

23
Solutions

• Perimeter Refresh Protocol
• ? Extension for robustness
• ? Handles nodes failure and topology change
• Structured Replication
• ? Extension for scalability
• ? Load balance

24
Perimeter Refresh Protocol
(replica)
(replica)
E
D
? Key stored at location L. ? Home node A. ?
Replicas D and E on the home perimeter
L
F
A
(home)
C
B
25
PRP contd..

• Consistency
• ? Every Th seconds, the home node generates a
  refresh packet, it will take a tour of the
  current home perimeter
• ? If the receiver is closer to the destination,
  it consumes that refresh packet, and initiates
  its own
• ? If not, forwards the refresh packet in
  perimeter mode
• ? Ensure the node closest to a keys hash
  location will become home node

26
PRP contd..

• Persistency
• ? Replica node receives a refresh packet,
• caches the data in the packet
• sets a takeover timer Tt for that key
• ? The timer expires, replica node initiates a
  refresh for that key and its data, addressed to
  the keys hashed location
• ? When home node fails, its replica nodes step
  forward to initiate refreshes

27
PRP contd..
(replica)
(replica)
E
D
? Some time after node A fails, replica D
initiates a fresh for L
L
F
C
B
28
PRP contd..
(replica)
(replica)
E
D
? Node F becomes the new home node ? Node F
recruits replicas B, C, D and E
L
F
(home)
C
(replica)
(replica)
B
29
Structured Replication (SR)

• Too many events with the same key are detected,
  keys home node could become a hotspot
• Hierarchical decomposition of key hashed location

? d, hierarchy depth ? mirrors, 4d -1 e.g.
d 2
30
SR contd..

• Storage cost reduces
• ? A node stores detected event at the mirror
  closest to its location
• Query cost increases
• ? Route queries to all mirror nodes recursively
• ? Responses traverse the same path, reverse
  direction

31
Comparison Study

• Metrics
• ? Total Messages
• total packets sent in the sensor network
• ? Hotspot Messages
• maximal number of packets sent by any
  particular node

32
Comparison Study - contd..

• Assume ? n is the number of nodes
• ? Asymptotic costs of O(n) for floods
• O(n 1/2) for point-to-point
  routing

ES LS DS
Cost for Storage O(n 1/2) 0 O(n1/2)
Cost for Query 0 O(n) O(n1/2)
Cost for Response 0 O(n1/2) O(n1/2)
33
Comparison Study -contd..

• Dtotal, the total number of events detected
• Q , the number of event types queries for
• Dq, the number of detected events of event
  types
• No more than one query for each event type, so
  there are Q queries in total.
• Assume hotspot occurs on packets sending to the
  access point.

34
Comparison Study contd..
ES LS DCS
Total
Hotspot

• DCS is preferable if
• Sensor network is large
• Dtotal gtgt maxDq, Q

35
Performance
Total Messages, varying queries
36
Performance contd..
Hotspot Messages, varying queries
37
Performance contd..
Total Messages, varying network size
38
Performance contd..
Hotspot Messages, varying network size
39
Conclusion

• In DCS, relevant data are stored by name at nodes
  within the sensornets.
• GHT hashes a key k into geographic coordinates,
  the key-value pair is stored at a node in the
  vicinity of the location to which its key hashes.
• To ensure robustness and scalability, DCS uses
  Perimeter Refresh Protocol (PRP) and Structured
  Replication (SR).
• Compared with ES and LS, DCS is preferable in
  large sensornet .

40
References

• 1 Sylvia Ratnasamy, Brad Karp, Scott Shenker,
  Deborah Estrin, Ramesh Govindan, Li Yin and
  Fang Yu, Data-Centric Storage in Sensornets with
  GHT, A Geographic Hash Table
• 2 Sylvia Ratnamy, Paul Francis, Mark Handley,
  Richard Karp, Scott Shenker, A Scalable
  Content-Addressable Network
• 3 Ion Stoica, Rober Morris, David Karger, M.
  Frans Kaashoek, Hari Balakrishnan, Chord A
  Scalable Peer-to-peer Lookup Service for Internet
  Application
• 4 C.Intanagonwiwat, R.Govindan, and D.Estrin,
  Directed Diffusion A Scalable and Robust
  Communication Paradigm for Sensor Networks.
• 5 Philippe Bonnet, Johannes Gehrke, Praveen
  Seshadri, Towards Sensor Database Systems

41
References

• 6 John Heidemann, Fabio Silva, Chalermek
  Intanagonwiwat, Ramesh Govindan, Deborah Estrin,
  Deepak Ganesan, Building Efficient Wireless
  Sensor Networks with Low-Level Naming
• 7 Sri Kumar, David Shepherd, and Feng Zhao,
  Collaborative Signal and Information Processing
  in Micro-Sensor Networks
• 8 Brad Karp, H.T.Kung, GPSR Greedy Perimeter
  Stateless Routing For Wireless Networks
• 9 Li Yin, Fang Yu, Presentation slides for A
  Scalable Routing Schema Based on Hashing
  Technique for P2P Wireless Ad Hoc Networks
• 10 Chengdu Huang, Presentation slides for Data
  Storage Schemas in Sensor Networks

42
Questions and Comments
Thank you!