You are on page 1of 14

# Modified Heterogeneous Leach

Project Guide: Prof S.Renuga Devi

called advanced node and others are normal nodes. 2 .  It can be done through: Initial setting.Heterogeneous WSN  Some nodes have given higher energy. or after re-energizing some sensors. after some operation time.

Routing  Goal: Increases network lifetime/coverage  Direct transmission to sink  Min-energy routing  LEACH – Low Energy Adaptive Clustering Hierarchy 3 .

between 0 and 1  If s<T(i). n x P nodes elected per round n n = total number of nodes 4 . s. node i becomes a cluster head in current round where: T( i ) = { where: P/ if i ? G 1 (1 .p×( r mod P )) 0 otherwise P = desired percentage of cluster heads G = set of nodes that have not been a cluster head in the last 1/P rounds  Each node is elected cluster-head once every 1/P rounds (epoch length)  On average.Selecting Cluster-heads in LEACH  Node i chooses random number.

feedback and cluster-head election becomes unreliable for a long time.In a heterogeneous condition The result of initial setting or evolution of the sensor network’s operation • “advanced” node has α times more energy than “normal” node  Once the first node dies. 5 . • advanced nodes don’t get elected as often as they should.

o Observation: n Each normal node is elected cluster-head once every epoch n Each advanced node is elected cluster-head (1+α) times every epoch  This ensures well balanced energy consumption 6 .Modification to LEACH  View network as homogeneous n Every “virtual” node is “normal” with unit energy n We have n+a * m * n = n (1+ a * m) “virtual” nodes  m = fraction of nodes that are advanced o Increase epoch length from 1/p rounds to (1+a * m)/p rounds.

Problem with this Solution  Probability of electing an advanced node is not checked by its additional energy n Advanced nodes may be underutilized n Lifetime is prolonged. but time until the first node dies can still be short! 7 .

Proposed Protocol  Idea: force each advanced node to be elected every sub-epoch of length (1+α x m)/P /(1+α) rounds  Probability of a normal node getting elected as clusterhead is Pnormal P/ Pnormal = (1 + a × m) Pnormal / T( i ) = { 1 (1 .Padvanced ×( r mod Padvanced ) 0 otherwise if i ? Gadvanced 8 .Pnormal×( r mod Pnormal )) 0 otherwise if i ? Gnormal  Probability of an advanced node getting elected as cluster-head is Padvanced  Average number of nodes elected per round = nxP P (1 + a )/ Padvanced = (1 + a × m) T(i)={ Padvanced 1 1.

Performance Measures  Stability Period n Time until death of the first node  Network lifetime n Time until death of the last alive node  Number of cluster-heads per round n Nodes which will directly send aggregated information to the sink  Number of alive nodes of each type per round  Throughput n Rate of data reporting to cluster-heads and to sink 9 .

Simulation Model  One network n 100 nodes uniformly distributed over 100mx100m     Sink is in the center of the sensor field Message size = 4000 bits Initial energy for a normal node = 0.5 Joules Energy consumed depends on radio characteristics 10 .

 Throughput is low 11 .Results for LEACH  LEACH does not take full advantage of the extra energy of advanced nodes  Normal nodes die very fast n Election process becomes unstable fast.

Results for LEACH  Advanced nodes die very slowly n They are not elected as often as they should  Throughput is low 12 .

Results for modified leach protocol  It takes full advantage of the extra energy of advanced nodes n Stable region increases by Significant percentage over LEACH n Higher throughput  Advanced nodes follow the death process of normal nodes n Smaller unstable region 13 .

Conclusion  Proposed Protocol is heterogeneous-aware n Cluster-heads elected based on node’s energy relative to that of others  It is scalable and dynamic n Even normal node can be elected n No global knowledge required at every round n No prior distribution of energy levels assumed 14 .