Figure 2. Dividing the cluster into four virtuall cells by cluster manager
nnonci E nnoncr E onc N nonc N i E r E CeM V M
λ β α
of a node is the number of neighboringordinary nodes which is in its transmission radio range in thesame cell and
is the number of all ordinary nodes in thecell.
is remaining energy of non-neighboring ordinarynode and
is its initial energy. Parameters
determine the weight of each ratio so that sum of them is 1.
The third parameter indicates the amount of remainingenergy of non-neighboring nodes should be more. As can beinferred, the possibility of failing non-neighboring nodes dueto the high remaining energy of these nodes is low.Therefore, the energy consumed by these nodes and the cellmanager to the fault management in the cell decreases.Cluster manager is connected only to cluster head andcell managers. Cell managers are connected only to clusterhead, cluster manager and nodes within the cell in proposedprotocol. We assume that these connections are single hop.
Energy Consumption Model
In DCMC, energy model is obtained from  that useboth of the open space (energy dissipation
) and multi path(energy dissipation
) channels by taking amount thedistance between the transmitter and receiver. So energyconsumption for transmitting a packet of l bits in distance
is given by (3).
0d d ,4d mpleleclE 0d d ,2d fsleleclE )d ,l(
is the distance threshold value which is obtainedby (4),
is required energy for activating the electroniccircuits.
are required energy for amplification of transmitted signals to transmit a one bit in open space andmulti path models, respectively.
Energy consumption to receive a packet of
bits iscalculated according to (5).
Fault Detection and Recovery
Managers of the cells, cluster manager and cluster headin the cluster cooperate with each other to detect the faultynode in their own cluster. Each of these nodes cooperate indetect the failure of different types of nodes in differentforms. These operations discuss in below.
Fault Detection and Recovery in the Ordinary Nodes
Initially, the managers of all cells are in sleep mode.When cluster head does not receive the data in response thesent
packet, it will immediately inform the clustermanager. Then cluster manager send an
message tomanager of the cell that desired node is located on it. Thenthe cell manager sends a Query message to the desired nodeand requests from it to responds to cell manager and clustermanager. Finally, the majority voting is done between cellmanager, cluster manager and cluster head to determine thestatus of desired node. The desired node is not used in thenext, if it is detected failed. Fig. 3 shows these operations.
Figure 3. Operations of fault detection and recovery in the ordinary nodes
Fault Detection and Recovery in the Cluster Head
For this purpose, cluster manager sends the Querymessage to cluster head and gets its response messageperiodically. Cluster manager reports to the cell managerswhen it does not receive any response from the cluster head.Cell managers also repeat this operation. Then the majorityvoting is done between cell managers and cluster manager todetermine the status of desired cluster head.
(IJCSIS) International Journal of Computer Science and Information Security,Vol. 9, No. 11, November 2011160http://sites.google.com/site/ijcsis/ISSN 1947-5500