Convergecasting in Wireless Sensor Networks

1
Convergecasting in Wireless Sensor Networks

• Valliappan Annamalai

2
Convergecasting in WSN

• WSN are mainly used for monitoring
• Monitoring involves data collection and request
  dissemination.
• Convergecasting Process of data collection from
  all or a set of sensors in the network towards
  the base station (Many to one communication)
• Energy and latency minimization is a requirement
  in WSNs
• Why Energy?
• Energy constrained nature of
  sensors
• Why latency?
• Requirements set forth by
  monitoring applications.

3
Convergecasting

• Route construction plays a major role during
  convergecasting.
• Criterion for route construction
• Energy consumption
• Latency incurred
• Choice of MAC layer Since traffic is many to
  one.

4
Choice of MAC layer

• Traffic is event triggered. So traffic of two
  nodes close together will be dependent in nature.
• And since traffic in many-to-one, high
  probability of collision.

BS
1
2
Collision
Collision
Coverage Area or Sensing Range
6
5
3
4
5
Collisions

• Results in packet loss.
• Need reliability, use retransmissions.
• Retransmission increases energy consumption and
  latency.
• Avoided by using a contention based or contention
  free MAC protocol.

6
MAC protocol

• Contention based (MACA, MACAW, etc)
• Acquire channel mostly using control packets.
• Good for independent traffic patterns.
• Disadvantage Additional energy consumption for
  control packets.
• Contention free (TDMA, CDMA, FDMA, etc)
• Additional cost associated with channel
  allocation.
• For networks with dependent traffic patterns.
• Can be used if network topology is static or
  changes rarely.
• Disadvantage Channel allocation overhead.

7
Energy

• Total energy consumed for gathering data from all
  the nodes in the network.
• Energy consumed at a node is used for
• Running the transceiver circuitry for
  transmitting a bit (Etrx)
• Amplifying a bit of data to be transmitted
  (Eamp).
• It depends on the transmission distance.

P1 and P2 are paths BS Base Station
BS
BS
Eamp 4nj
Eamp 4nj
P2
P2
2 hops
Eamp 4nj
P1
P1
Eamp 5nj
3 hops
Eamp 5nj
n
n
Energy consumed for running transceiver To
transmit k data bits from n to BS
Amplification energy consumed for transmitting k
data bits from n to BS
P1 5 Etrx k P2 3 Etrx k
P1 (4nj 4nj 5nj) k 13 k nj P2 (4nj
5nj ) k 10 k nj
8
Energy

• Transceiver startup time.
• Frequently switching the transceiver on leads to
  higher energy wastage.
• Aggregation reduces packet header overhead.

BS Base Station
BS
Aggregation reduces transmitter startup
energy wastage
Slots allocated to children should reduce
cumulative startup time of parents receiver
1
2
Time-slot 3
4
5
3
Time-slot 1
Time-slot 2
9
Latency

• Time taken to gather data at the base station
• Latency No. of time-slots x Length of one-slot
• Balanced tree helps in reducing total number of
  time-slots and length of time-slots

Unbalanced Tree
Balanced Tree
BS Base Station
BS
BS
t 4
1
2
t 2
t 3
t 1
1
2
4
5
3
4
5
3
t 1
t 2
t 3
t 1
t 2
t 1
Number of slots 4 Length of each slot 4
packets Latency 16 units
Number of slots 3 Length of each slot 3
packets Latency 9 units
10
Summary

• Energy and latency minimized by avoiding
  collisions.
• Energy consumption can also be minimized by
• Reducing the number of hops.
• Choosing path that minimizes amplification
  energy.
• Reducing energy wastage due to transceiver
  startup time by performing data aggregation.
• Latency minimization by building a balanced
  routing tree.

11
Our Work

• Algorithm CTCCAA3
• Construct a tree and allocate channels.
• Channels are a combination of time-slots t and
  CDMA codes c (ltt,cgt).
• Improvement to CTCCAA4
• Constructs the tree with ßrule.

12
Assumptions

• Etrx lt Eamp.
• One transceiver per node.
• Nodes have maximum transmission range (MEamp).
• Clock synchronization mechanism exists.

13
CTCCAA

• Builds the tree and allocates channel for the
  nodes.
• Allocates channel for two different convergecast
  patterns
• Synchronous Used for realtime data. Enables
  aggregation. Therefore parent transmits after it
  receives from children (parent time-slot gt child
  time-slots)
• Asynchronous Used for non-realtime data. Enables
  aggregation only if data does not depend in time.

14
Synchronous Convergecast

• Data collection starts from leaf nodes.
• Each parent waits for data from its children
  before sending its data
• Reordering based on timestamp is not necessary at
  base station.

BS
BS
lt4, 1gt
lt3, 1gt
2
2
1
1
3
3
lt3, 1gt
lt1, 1gt
5
5
lt2, 1gt
4
4
Network
Convergecast Tree
Note Weights indicate the amplification energy
expended to transmit a data bit over that link
15
Asynchronous Convergecast

• Data collection takes place at independent and
  not interfering parts of the
• network
• Reordering necessary at base station.
• Latency will be low.

BS
BS
lt1, 1gt
lt2, 1gt
2
2
1
1
lt2, 1gt
3
3
5
5
lt1, 1gt
lt3, 1gt
4
4
Network
Convergecast Tree
Note Weights indicate the amplification energy
expended to transmit a data bit over that link
16
Channel allocation Criterion 1

• Each node has one transceiver
• Therefore a parent with two children cannot
  receive from both of them at the same time
  using two different codes.
• Therefore children transmit at different time
  instants.

ParentX ParentY
ParentX ParentY
Not Possible
X
Y
X
Y
ltt2,c1gt
ltt1,c2gt
ltt1,c1gt
ltt1,c1gt
17
Channel allocation Criterion 2

• Avoid exposed terminal problem

Transmission range
ParentX
ParentY
Collision
Collision
If X and Y use the same channel
If X and Y transmit at different time-slots
If X and Y transmit using different CDMA codes
X
Y
18
Channel allocation Criterion 3

• Parent cannot receive the same time it is
  transmitting.
• Therefore parent time-slot ? child time-slot.

ParentParentx
ParentParentx
ParentX
Not Possible
ParentX
ltt1,c2gt
ltt2,c1gt
ltt1,c1gt
ltt1,c1gt
X
X
19
Algorithm CTCCAA

• Builds the tree and allocates the channel (is a
  tuple of time-slot t and CDMA codes c, ltt,cgt).
• Tree constructed in top down manner.
• Uses channel allocation criteria defined earlier
• Additional criterion for synchronous
  convergecasting
• child time-slot lt parent time-slot
• Since tree construction is top down it is not
  possible to allocate a valid time-slot for
  children.
• And so does channel allocation in two phases
• Phase I
• Construct tree and allocate channel in
  increasing order of time-slots
• Phase II
• Reverse mapping of time slots to enable
  synchronous convergecast.

20
CTCCAA Phase I (Tree Construction)

• Constructs the tree by reducing number of hops
  and then choosing the path that consumes minimum
  amplification energy.
• Reason Etrx lt Eamp since transmission range of
  sensors are small.
• Starts constructing the tree with Base station
  (BS) as the root node
• Maintains a possible parent and a possible child
  list.
• Possible Parent List (PPL) All nodes recently
  added to the tree
• Possible Children List (PCL) x exists y e
  PPL such that Eamp(x,y) lt MEamp
• Parent selection
• forall x e PCL parentx arg Minforall y e PPL
  Eamp(x,y)
• If forall x e PCL parentx ? null then copy PCL to
  PPL

21
Example Phase I
Weights on links indicate the amplification
energy expended to transmit a data bit
Initially current level is 0 PPL BS PCL 1,
2 Since BS is the only possible parent both
1 and 2 choose BS as their parent.
BS
2
1
3
5
4
Note Weights indicate the amplification energy
expended to transmit a data bit over that link
22
CTCCAA Phase I (Channel Allocation)

• Use a combination of CDMA codes and time-slots.
• Allocates children a time-slot that is greater
  than parent (will do reverse mapping in phase II).

23
Example Phase I
Weights on links indicate the amplification
energy expended to transmit a data bit
This example assumes channel to be divided over
time. Initially current level is 0 PPL BS PCL
1, 2
BS
lt1, 1gt
lt2, 1gt
2
1
3
5
4
24
Example Phase I
Weights indicate the amplification energy
Initially current level is 0 PPL 1, 2 PCL
3, 4, 5
BS
lt1, 1gt
lt2, 1gt
2
1
3
lt2, 1gt
lt4, 1gt
5
lt3, 1gt
4
25
CTCCAA Phase II

• Only executed for synchronous convergecast.
• Uses maximum time-slot (Maxts) allocated in the
  network
• Actual time-slot Maxts allocated time-slot.

26
Example Phase II
Maxts 4
BS
lt4, 1gt
lt2, 1gt
lt1, 1gt
2
lt3, 1gt
3
lt4, 1gt
lt1, 1gt
lt2, 1gt
lt3, 1gt
5
4
lt3, 1gt
lt2, 1gt
27
Example
Example Shows the advantage of divided channel
over time and CDMA Codes. CDMA codes help in
reducing latency by increasing time-slot reuse.
BS
lt3, 2gt
2
1
lt2, 2gt
3
lt1, 2gt
lt1, 1gt
5
4
lt2, 1gt
28
Results

• Convergecast will be preceded by broadcast in
  monitoring applications.
• Better to maintain a single tree for both
  convergecast and broadcast.
• Measured energy and latency incurred during
  convergecast over a broadcast tree and a tree
  constructed by CTCCAA.
• Similarly we measured energy and latency for
  broadcasting over both the trees.

29
Results Continued Latency
Ratio on time taken for synchronized
convergecasting using 3 CDMA codes
Tb,c Time taken for convergecasting over a
broadcast tree. Tc,c Time taken for
convergecasting over a tree constructed by
CTCCAA. Tree constructed by CTCCAA incurs lesser
latency compared to the broadcast tree.
30
Results Continued Latency
Ratio on time taken for synchronized
convergecasting using 5 CDMA codes
Tb,c Time taken for convergecasting over a
broadcast tree. Tc,c Time taken for
convergecasting over a tree constructed by
CTCCAA. Tree constructed by CTCCAA incurs lesser
latency compared to the broadcast tree.
31
Results Continued Energy
Eb,c Energy Consumed for convergecasting over a
broadcast tree. Ec,c Energy consumed for
convergecasting over a tree constructed by
CTCCAA. Tree constructed by CTCCAA incurs lesser
Energy compared to the broadcast tree.
32
Results Continued Broadcasting Latency
Ratio on time taken for broadcasting on randomly
generated graphs
Tb,b Time taken for broadcasting over a
broadcast tree. Tc,b Time taken for broadcasting
over a tree constructed by CTCCAA. Tree
constructed by CTCCAA performs as good as
the broadcast tree for broadcasting.
33
Improved CTCCAA

• Problem with CTCCAA
• Constructed tree unbalanced.
• Improved CTCCAA
• Tree Construction
• A set of nodes chooses closest neighbors as its
  children subject to ß-rule (ß specifies number
  of children)
• This process is followed iteratively until all
  the nodes in the network join the tree
• A node chooses more than ß nodes as its children
  only if they do not have any other possible
  parent.

34
Results

• Energy for Convergecast (ß 3)
• CTCCAA and improved CTCCAA consume almost same
  amount of energy for convergecast
• Improved CTCCAA gains up to 8 over Imrich87
  for network of size gt150 nodes

35
Results

• Latency for Convergecast ( ß 3)
• Tree constructed by improved CTCCAA is almost 4
  times faster than the tree constructed by CTCCAA
  and 2 times faster than broadcast tree.

36
References

1. Estrin02 Modelling Data-Centric Routing in
   Wireless Sensor Networks by B. Krishnamachari,
   D. Estrin, S. Wicker Published in IEEE Infocom
   2002.
2. Lindsey02 PEGASIS Power-Efficient Gathering
   in Sensor Information Systems by S. Lindsey C.
   S. Raghavendra Published in IEEE Aerospace
   Conference Proceedings, 2002.
3. Valli03 On Tree-Based Convergecasting in
   Wireless Sensor Networks by V. Annamalai, S. K.
   S. Gupta and L. Schwiebert Published in IEEE
   Wireless Communications and Networking
   Conference, 2003.
4. Sarma03 A low-latency and energy-efficient
   algorithm for convergecast in wireless sensor
   networks byS. Upadhyayula, V. Annamalai and S.
   K. S. Gupta Published in Proceedings of Globecom
   2003.
5. Zhang04 Reliable Bursty Convergecast in
   Multi-hop Wireless Sensor Networks by Hongwei
   Zhang, Anish Arora, Young-ri Choi, Mohamed G.
   Gouda. Technical Report OSU-CISRC-7/01-TR42, Ohio
   State University.
6. Woo03 Taming the Underlying Challenges of
   Reliable Multihop Routing in Sensor Networks by
   Alec Woo, Torence Tong and David Culler. In
   proceedings of Sensys 2003.
7. Imrich87 Tree-Based Broadcasting in Multihop
   Radio Networks by I. Chalmatac. and S. Kutten.
   IEEE Transactions on Computers Vol. C-36, No. 10,
   Oct 1987.

37
Slot Fragmentation

• Arises when the transceiver is idle state during
  its Tx or Rx slots.
• Fragment size depends on the amount of data
  transmitted or received and slot length.
• Leads to higher latency and energy wastage.

BS Base Station
BS
t 2
TFS Transmission Fragment Size RFS Reception
Fragment Size
t 3
2
1
TFS 1
TFS 0
RFS 2
RFS 4
4
5
3
t 2
t 1
t 1
TFS 2
TFS 2
TFS 2
RFS 0
RFS 0
RFS 0
Total energy wasted is 13 Eidle Time wasted is
4 units