Paper

1
Paper 9 Adaptive Protocols for Information
Dissemination in Wireless Sensor
NetworksbyWendi Heinzelman, Joanna Kulik
Hari BalakrishnanMIT, 1999.
2
What is this paper about?

• Sensor Protocols for Information via Negotiation
  (SPIN)
• Energy efficient protocol for communicating
  information in a wireless network
• - Reasons for SPINs development
• - Explanation of SPINs working
• - Comparison of SPIN with other Dissemination
  algorithms using an experimental setup

3
What we expect in a wireless sensor network

• Improved sensing accuracy by providing
  distributed sensing
• Aggregated data to provide a rich,
  multi-dimensional view of the environment
• Shifting focus to critical events sensed by other
  sensors in the network
• High fault tolerance

4
The obstacles we face!!!

• Our resources are limited!!!
• Energy
• Computation power
• Communication

5
What sparked off SPIN?

• Study of conventional protocols led to SPINs
  development protocols were characterized as
  Classic Flooding
• Start with source node
• Send data to all its neighbors
• On receiving, each node sends a copy to its
  neighbors

6
So, whats the problem?

• Implosion
• sending data to neighbor without checking
  whether it has a copy of that data
• Overlap
• nodes gathering overlapping sensor data and
  sending multiple copies to other nodes
• Resource Blindness
• nodes are not resource-aware and do not modify
  activities based on their resource levels

7
OK.how do we take care of them?

• The SPIN protocol incorporates -
• Negotiation
• SPIN nodes use meta-data (data descriptors) to
  determine whether neighboring nodes need that
  piece of data only useful information is sent!!
  
• takes care of implosion and overlap
• Resource - adaptation
• Sensor nodes poll their resource managers (to
  check their energy levels) before plunging into
  data transmissions
• takes care of resource blindness

8
SPIN Sensor Protocol for Information via
Negotiation

• Elements of the SPIN protocol -
• Meta-Data
• Format for meta-data is application specific. Ex
  for a camera, (x,y,Ø) can be the meta data
• SPIN Messages
• - ADV new data advertisement
• - REQ request for data
• - DATA data message
• SPIN Resource management
• SPIN nodes poll their resource managers to
  decide their capability of participation in data
  transmissions

9
SPIN Implementations

• 2 SPIN protocols -
• SPIN - 1
• 3-stage handshake protocol (ADV-REQ-DATA)
• SPIN - 2
• An extension of the SPIN-1 protocol - Low-energy
  threshold added

10
Some Data Dissemination Algorithms

• Classic Flooding
• every node sends data to the other except the
  node from which it got that data does not solve
  implosion or overlap
• Gossiping
• uses randomization forwards data to only 1
  node (could be sender!!) avoids implosion, but
  not overlap
• Ideal Dissemination
• uses shortest path algorithms, no energy is
  wasted A state of utopia!!!

11
A comparison of these algorithms

• A sensor network simulator ns was developed
  for the comparison. Tests were conducted under 2
  conditions -
• Unlimited energy simulations
• Limited energy simulations

12
Case 1 Unlimited Energy Simulations

• SPIN-12 worked identically
• Gossiping takes the longest for data acquisition
  SPIN-1 appears to be worse than Flooding, but
  convergence time negligible for longer
  simulations.
• Gossiping is the worst in terms of total amount
  of energy dissipated. SPIN-1 better than Flooding
  by a factor of 3.5

13
Case 2 Limited Energy Simulations

• Total energy was limited to 1.6 Joules
• SPIN-2 was able to distribute 73 of total data
  as compared to ideal (around 85)
• Flooding uses up all its energy very quickly.
  Rest remain operational for slightly longer.

14
Some Applications

• SPIN wasn't really used!!
• NNTP
• - news servers form neighborhoods
• - use names and time-stamps as meta-data to
  negotiate data dissemination
• Gossiping Broadcast Algorithms
• - epidemic algorithms
• - maintain database consistency
• - fault-tolerance

15
A quick review

• Introduction to the SPIN protocol
• SPIN uses negotiation and resource -adaptation
• SPIN-1 3 Stage handshake protocol
• SPIN-2 same as SPIN-1 with low level energy
  threshold
• By using meta-data, SPIN solves the implosion and
  overlap problem
• SPIN outperforms flooding in terms of energy
  consumption and convergence time