Date post: | 31-Dec-2015 |
Category: |
Documents |
Upload: | garry-hutchinson |
View: | 212 times |
Download: | 0 times |
Data centric Storage In Sensor networks
Based on
Balaji Jayaprakash’s slides
Overview of the Seminar
Introduction Keywords and Terminology Existing Schemes Why Data centric Storage? Assumptions Geographic Hash table Comparitive Study Conclusion
Introduction
Sensornet ♦ A distributed network comprised of a large number of small sensing devices equipped with • Computation • Communication • Storage
♦ Great volume of data
Data Dissemination Algorithm ♦ Energy efficient
♦ Scalable ♦ Self-organizing
Keywords and Terminology
Observation ♦ low-level readings from sensors
♦ e.g. Detailed temperature readings
Events ♦ Predefined constellations of low-level observations
♦ e.g. temperature greater than 75 F
Queries ♦Used to elicit information from sensor network
Total Usage /Hotspot Usage
Total Usage Total number of packets sent in the Sensor
network Hotspot Usage The maximal number of packets send by a
particular sensor node
Existing schemes for Storage
External Storage (ES) Local Storage (LS) Data Centric Storage (DCS)
External Storage (ES)
External
storage
event
Local Storage (LS)
EventData
EventData
event
event
Why do we need DCS?
Scalability Robustness against Node failures and Node
mobility To achieve Energy-efficiency
Assumptions in DCS
Large Scale networks whose approximate geographic boundaries are known
Nodes have short range communication and are within the radio range of several other nodes
Nodes know their own locations by GPS or some localization scheme
Communication to the outside world takes place by one or more access points
Data Centric Storage
Relevant Data are stored by “name” at nodes within the Sensor network
All data with the same general name will be stored at the same sensor-net node.
e.g. (“elephant sightings”) Queries for data with a particular name are
then sent directly to the node storing those named data
Data centric Storage
EventData
Elephant Sighting
source:lass.cs.umass.edu
Geographic Hash Table
Events are named with keys and both the storage and the retrieval are performed using keys
GHT provides (key, value) based associative memory
Geographic Hash Table Operations GHT supports two operations ♦ Put(k,v)-stores v (observed data) according to the
key k ♦ Get(k)-retrieve whatever value is associated
with key k Hash function ♦ Hash the key in to the geographic coordinates ♦ Put() and Get() operations on the same key “k”
hash k to the same location
Storing Data in GHT
Put (“elephant”, data)
(12,24)
Hash (‘elephant’)=(12,24)
source:lass.cs.umass.edu
Retrieving data in GHT
Get (“elephant”)
Hash (‘elephant’)=(12,24)
(12,24)
Geographic Hash Table
PDA
Node A
Node B
Algorithms Used By GHT
Geographic hash Table uses GPSR for Routing
(Greedy Perimeter stateless routing)
PEER-TO-PEER look up system
(data object is associated with key and each node in the system is responsible for storing a certain range of keys)
Algorithm (Contd)
GPSR- Packets are marked with position of destinations and each node is aware of its position
Greedy forwarding algorithm Perimeter forwarding algorithm
A
B
A
B
Home Node and Home perimeter
In GHT packet is not addressed to specific node but only to a specific location, hence only perimeter mode is used
The packet will traverse the entire perimeter that encloses the destination
before being consumed at the home node (the node closest to destination)
Problems
Robustness could be affected» Nodes could move (i.d. of Home node?)» Node failure can Occur» Deployment of new Nodes
Not Scalable» Storage capacity of the home nodes» Bottleneck at Home nodes
Solutions to the problems
Perimeter refresh protocol
Structured Replication
Perimeter refresh protocol
Replicates stored data for key k at nodes around the location to which k hashes, and ensures that one node is chosen consistently as the home node for that “K” –consistency & persistence
By hashing keys, GHT spreads storage and communication load between different keys evenly throughout the sensornet
Perimeter Refresh Protocol
E
F
B
D
A
C
L
ED
F
C
B
L
home
Replica
ReplicaReplica
Replica
home
Replica
Replica
Time Specifications
Refresh time (Th)
Take over time (Tt)
Death time (Td)
General rule
Td>Th and Tt>Th
In GHT Td=3Th and Tt=2Th
Characteristics Of Refresh Packet Refresh packet is addressed to the hashed
location of the key Every (Th) secs the home node will generate
refresh packet Refresh packet contains the data stored for
the key and routed exactly as get() and put() operations
Refresh packet always travels along the home perimeter
Structured Replication
Too many events are detected then home node will become the hotspot of communication.
Hierarchical decomposition of the key space Structured replication reduces the cost of storage
and is useful for frequently detected events.
Comparative Study
Comparison based on Cost
Comparison based on Total usage and Hot spot usage
Assumptions in comparison
Asymptotic costs of O(n) for floods and O( n) for point to point routing
Event locations are distributed randomly Event locations are not known in advance No more than one query for each event type
(Q –Queries in total) Assume access points to be the most heavily used area
of the sensor network
Comparison based on Cost
Cost External storage
(ES)
Local storage
(LS)
Data-centric storage
Cost for Storage
O(n) 0 O(n)
Cost for query 0 O(n) O(n)
Cost for Response
0 O(n) O(n)
Comparison based onHot-spot/Total Usage n - Number of nodes T - Number of Event types Q – Number Of Event types queried for Dtotal – Total number of detected events
DQ- Number of detected events for queries
DCS TYPES
Normal DCS – Query returns a separate message for each detected event
Summarized DCS – Query returns a single message regardless of the number of detected events
(usually summary is preferred)
Comparison Study – contd..
ES LS DCS
Total
Hot
spot
nDtotal nDQn q nDnDnQ qtotal
totalDqDQ qDQ
)(summarynQnDnQ total
)(2 summaryQ
Observations from the ComparisonDCS is preferable only in cases where
Sensor network is Large There are many detected events and not all
even types queried
Dtotal>>max(Dq,Q)
Simulations
To check the Robustness of GHT
To compare the Storage methods in terms of total and hot spot usage
Simulation Setup
ns-2 Node Density – 1node/256m2
Radio Range – 40 m Number of Nodes -50,100,150,200 Mobility Rate -0,0.1,1m/s Query generation Rate -2qps Event types – 20 Events detected -10/type Refresh interval -10 s
Performance metrics
Availability of data stored to Queriers
(In terms of success rate)
Loads placed on the nodes participating in GHT (hotspot usage)
Simulation Results for Robustness GHT offers perfect availability of stored
events in static case It offers high availability when nodes are
subjected to mobility and failures
Simulation Results under varying Q
Number of nodes is
constant= 10000
Simulation results under varying N
Number of Queries Q =50
Simulation Results for comparison of 3-storage methods S-DCS have low hot-spot usage under
varying “Q” S-DCS is has the lowest hot-spot usage
under varying “n”
Conclusion
Data centric storage entails naming of data and storing data at nodes within the sensor network
GHT- hashes the key (events) in to geographical co-ordinates and stores a key-value pair at the sensor node geographically nearest to the hash
GHT uses Perimeter Refresh Protocol and structured replication to enhance robustness and scalability
DCS is useful in large sensor networks and there are many detected events but not all event types are Queried
REFERENCES Deepak Ganesan, Deborah Estrin, John Heidemann,
Dimensions: why do we need a new data handling architecture for sensor networks?, ACM SIGCOMM Computer Communication Review, Volume 33 Issue 1, January 2003 Scott Shenker, Sylvia Ratnasamy, Brad Karp, Ramesh Govindan, Deborah Estrin, Data-centric storage in sensornets, ACM SIGCOMM Computer Communication Review, Volume 33 Issue 1, January 2003
Sylvia Ratnasamy, Brad Karp, Scott Shenker, Deborah Estrin, Ramesh Govindan, Li Yin, Fang Yu, Data-centric storage in sensornets with GHT, a geographic hash table, Mobile Networks and Applications, Volume 8 Issue 4, August 2003
Chalermek Intanagonwiwat, Ramesh Govindan, Deborah Estrin, John Heidemann, Fabio Silva, Directed diffusion for wireless sensor networking, IEEE/ACM Transactions on Networking (TON), Volume 11 Issue, February 2003
R. Govindan, J. M. Hellerstein, W. Hong, S. Madden, M. Franklin, S. Shenker, The Sensor Network as a Database, USC Technical Report No. 02-771, September 2002