Storage Management Issues for Sensor Networks

1
Storage Management Issues for Sensor Networks

• Sameer Tilak (SUNY Binghamton)
• Wendi Heinzelman (Univ. of Rochester)
• Nael Abu-Ghazaleh (SUNY Binghamton)

2
Network Partitioning
3
Problem

• For tiny sensors storage is a scarce resource
• Naïve protocols can lead to loss of information
  or inefficient storage
• Need efficient storage management protocol
• Un-partitioned network
• If delays can be tolerated instead of reporting
  to the base station, data can be aggregated and
  stored in the network to increase energy
  efficiency

4
Protocol Design Goals

• Higher storage-time
• Low average storage/sensor
• Fault-tolerance, high reliability
• Scalable
• Low protocol overhead
• High energy efficiency

5
Candidate Protocols

• Local Storage
• Deterministic Broadcast (Bcast)
• Probabilistic Broadcast (PBcast)
• Clustering

6
Local Storage
storage/sensor

• No communication
• Each sensor stores its own data locally, low
  fault-tolerance
• No spatial aggregation

7
Broadcast (Bcast)
storage/sensor

• All sensors store data
• All sensors broadcast data to neighbors, high
  fault-tolerance
• Spatial aggregation possible

8
PBcast
storage/sensor

• Sensors probabilistically send/store data, high
  fault-tolerance
• Spatial aggregation possible

9
Clustering
Round 1 (time 0)
Round 2 (time 20)

• Only CH stores data
• Rotate CH
• Distributed storage, medium fault-tolerance
  
• Spatial aggregation possible

Round 1 (time 40)
10
Protocol Comparison
Metric Local Bcast PBacst Clustering
Storage-time Low Low Medium High
Fault-tolerance Low High High Medium
Scalability High Low Medium High
Overhead Low Low Low Medium
Energy efficiency High Low Low High
11

• Clustering Scales well, low storage
• Bcast Scaling problem, high storage
• Pbcast,Local Scales well