Presentation: Energy Efficient Communication Protocol for Wireless Microsensor Networks

1
Presentation Energy Efficient Communication
Protocol for Wireless Microsensor Networks

• Wendi Rabiner Heinzelman, Anantha Chandrakasan,
  and Hari Balakrishnan
• Massachusetts Institute of Technology

2
Underlying Model

• Hundred to thousands of nodes
• A fixed basestation
• A distant basestation (from sensor patch)
• Basestation is not energy constrained
• Sensing nodes are heterogeneous
• Sensing nodes are energy constrained
• Sensing nodes can aggregate data

3
Radio Characteristics

• 50 nano-Joules per bit to operate transmitter and
  receiver
• 0.1 nano-Joules per bit per
• Sending
• Receiving
• Operating the radio, not counting generating the
  signal, takes power equal to the
  signal amplifier transmitting 31m.

4
Direct Communication Protocol

• Givens k bits, n nodes, r meters between each
  node
• Direct communication from furthest node to
  basestation

r
r
n

r
5
Minimum-Energy Multi-Hop Routing (MTE)

• Assumption each node sends to nearest neighbor.
• Direct communication from furthest node to
  basestation

r
r
n

r
6
The Observation

• Directly sending to the basestation takes less
  global energy in certain situations
• Which for the numbers given occurs when
  exceeds 1000

7
Pattern of Sensor Node Deaths

• Direct Routing nodes further from basestation
  die first
• MTE Routing nodes closer to base station die
  first
• Static Clusters cluster heads die first.

8
The LEACH Protocol (Summary)

• Divided into rounds, made of up of turns
  consisting of
• Nodes decide to be a cluster heads this turn.
• Nodes picks the closest cluster head.
• Cluster heads broadcast transmit schedule.
• Nodes send data to their cluster head.
• Cluster head aggregate/compress data and send it
  to the basestation.

9
LEACH Comparison

• Ignoring cluster setup portion of algorithm.
• 8 times longer for first node to die.
• 3 times longer for last node to die.

10
LEACH Comparison (cont.)
11
LEACH Comparison (cont.)
12
LEACHs Death Pattern
13
Details of LEACH

• Deciding to be a cluster head uses a statistic
  method, where P is the optimal number of clusters
  (dependent on network parameters (5 in test
  cases)). Probability of being a cluster head
  

if havent been a cluster head this round
0 otherwise
14
Details of LEACH (continued)

• Each cluster head broadcasts a cluster-head-advert
  isement using CSMA MAC protocol.
• Nodes receiving the advertisement, choose a
  cluster head based on signal strength.
• Nodes (using CSMA protocol) inform chosen cluster
  head of their choice.

15
Details of LEACH (continued)

• Cluster heads generate a TDMA schedule and
  broadcast to member nodes (may also pick a
  spreading code for members).
• Nodes transmit data based on TDMA schedule.
  Radios turned off when not in use.

16
Details of LEACH (continued)

• After data has been received, cluster head
  perform signal processing/compression and send to
  basestation.
• After a certain time (determined a prior) a new
  turn begins.
• After 1/P turns, a new round begins.

17
LEACH Animated (P33)
1. Decide cluster heads
2. Broadcast advertisement
3. Nodes transmit membership
4. Heads broadcast schedule
5. Nodes transmit data
6. Heads compress data and send to basestation
7. New turn begins goto 1.
18
Future Work

• Hierarchical Clustering
• Cluster size should be variable with distance
  from basestation?
• Other thoughts?

19
Issues (things to think about)

• Does random picking of cluster heads result in
  occasional bad turns?
• Their example was hokey, what would the numbers
  be versus different network layouts?
• Their MTE routing algorithm always sent to
  nearest neighbor, how does it compare without
  this restriction?
• Is power consumption really fair (those far
  away still seem to consume more power)?