First, the operation of the real-time networking system usingHSACPis illustrated in Figs. 5.7a and 5.7b. Figs. 5.7a and 5.7b show the current consumption of the
sensor nodes during the setup phase and data transmission phase respectively in the network of 30 nodes usingHSAprotocol. During the setup phase, all the nodes have to turn on their radio component to send the advertisement message and listen to the assignment messages transmitted from the BS. The time interval taken to complete the setup phase is in the range of 540 ms to 700 for all the sensor nodes.
A sample of the setup phase at four different nodes is shown in Fig. 5.7a, sensor node 1, 2, 3 and 4 spend around 560 ms, 549 ms, 662 ms and 564 ms respectively to complete its setup phase with node 3 being the longest.
During the data transmission phase, the cluster member tries it best to send data before dropping the packet during the transmission time interval,TRadioOn CM
given at the end of the setup phase. Meanwhile, the maximum time interval a CH turns on its radio isTRadioOn CH that includes time for receiving data packets from all the members, processing data and transmitting the compressed packet to the BS: TRadioOn CH =TDataRx+TDataAgg+TDataT x with the values given in Table 5.2.
Fig. 5.7b shows the data transmission phase of a cluster containing of 12 member nodes. In this figure, only 3 member nodes in this cluster and its CH are shown.
The cluster member nodes 1, 2 and 3 turn their radio on for around 11 ms, 11 ms and 4 ms respectively, meanwhile the time interval of turning on radio component of the CH is around 267 ms for data receiving, processing and transferring to the BS.
In the next experiment, the lifetime of the network is studied and compared with other conventional protocols such as LEACH-C and FCMCP presented in Chapter 4.
(a) Current consumption of the sensor nodes during thesetup phase.
(b) Current consumption of the cluster members and CH dur- ing the data transmission phase.
Figure 5.7: Measurement of the sensor nodes current consumption in different phases.
The comparison of the lifetime amongst the networks using a LEACH-like protocol, FCMCP and HSACP is shown in Fig. 5.8. As mentioned in Chapter 4, tPf irst is defined as the time until the first node in the network using protocol P runs out of energy and tPlast is the time until the number of alive nodes is smaller than the number of clusters which is three in this case.
Figure 5.8: Comparison of the network lifetime with LEACH and FCMCP
It is observed in Fig. 5.8 that the network with LEACH-like protocol has tLEACHf irst −C = 12.72 hours, meanwhile in the FCMCP network, tF CM CPf irst = 16.76 hours, and in the HSACP network tF CM CPf irst = 18.56 hours that is the longest period of time. As mentioned above, LEACH chooses the CHs in a random way.
Although, it attempts to rotate theCHrole evenly among the sensor nodes by using a predefined probability of beingCH, it is not guaranteed the equal distribution of the nodes into clusters as well as some nodes can be selected as the CH for more times. Hence, the energy capacity is drained faster. Meanwhile, FCMCP assists the network to be organized in a better way, sensor nodes are evenly allocated into clusters. Therefore, the traffic load at the CHs can be balanced amongst the clusters and fast depletion of energy at the CH can be avoided. However, the CH is only chosen among the nodes within a cluster that may not be the best in the network in term of energy distribution. When HSACP is applied, the network formation and cluster head selection are considered at the same time in order to further optimize the network organization. The best CHs in terms of energy efficiency can be selected, whereas clusters can be formed optimally. The efficiency
ofHSACP can be seen clearer by comparing the time tPlast. Fig. 5.8 shows that the time tHSACPlast = 35.17 hours that is much longer than that of FCMCP, tF CM CPlast = 21.54 hours and that of LEACH, tLEACHlast −C = 20.68 hours. At the beginning, all the nodes have the same energy level. However, during the network operation, at various time instance, each node has a different residual energy. LEACH-C and FCMCP algorithms focus only on obtaining uniform distribution of the sensor nodes into different clusters; the selection of the CHs for energy efficient operation is not optimized. It is therefore, the network lifetime may not have been maximized.
Meanwhile, HSACP involves a better way of optimizing the network control, thus, better performance presented by longer network lifetime is achieved.