Simulations

All the results in this section are based on the analyses that are validated by rigorous simulations. we performed thorough simulations to verify the effectiveness of DCM on

Castalia 3.0 simulator, which is based on OMNeT++ 4.0 platform [32]. In both analysis and simulations we used the basic frame types of IEEE 802.15.4 standard as shown in Fig.

2.5. The CSMA/CA MAC parameters were set as: length of data frame (L= 7), length of ACK frame (Lack= 2), back-off exponents (macMaxCSMABackof f= 4), and (macMinBE

= 3). All the parameter values used for both analysis and simulation are listed in Table 2.1.

All the radio parameters are taken from and CC2420 datasheet. The hardware parameter values are set according to MICAz motes. The channel bandwidth was set to 1Mb/s. We assumed that each node reports data once every 300ms(t =0.3s). All simulations were run independently with each result is averaged under 10 different seeds.

Fig. 2.8 shows the average energy consumption comparison of a node Ecomm(k) with different loss rates. Initially, we deployed 50 homogeneous nodes randomly in a sensing field of area 100×100m²with each node having communication range of 10 m. Then we varied the number of nodes from 50 to 300 with fixed field area. We also increased the communication range gradually from 10m to 50 mto see the impact. It is obvious from the figure that energy consumption is increased with the increase of communication range.

One of the reasons is that with fixed area size, the neighbor nodes in the communication range of a source node will increase, which will results in more energy consumption due to packet forwarding. It is also clear that higher the frame loss rate, more the energy will be consumed.

Fig. 2.9 shows the comparison of average energy consumption of a node with overheads caused by multi-hop communications. We analyzed the results with different packet loss rates. It is clear from the figure that a node will consume more energy with the increase of communication range. Similarly, more energy is consumed with the increase in the number of hops from source to the sink during data propagation. However,

Table 2.1 Analysis and simulation parameters.

Parameters Values

Initial energy of node Sleep power (Psleep) Idle power (Pidle)

Transmit amplifier energy(_amp) Data rate (^)

Communication range (rc) Network size (|S|)

Frame loss rate (lf) Network area (A) Simulation time (t) Frame size

Supply voltage to node (Vs)

3000Joules(MICAz mote) 0.016(mW)

12.36(mW)

100pJ/bit/m² 0.6to250Kbps 10to400m 50to300 0to50%

(100×100)m² (200 to 1000)seconds

133 bytes (Data) 19 bytes (Control)

2.7V

Figure 2.8: Average energy consumptionEcomm(k)versus different loss rate of frames (lf), here, duty cycle (α) = 1, l= 100m, and λ =1 Kbps.

Figure 2.9: Comparison of average energy consumption including multi-hop overheadsE(h)with different lf, here, α=1, |S|=100, andl=100m.

Figure 2.10: Average energy consumption comparison of a nodeEcomm(k) in DCM with the model proposed in [18]. Here α=1, lf= 0%, l=100m, λ =2Kbps.

maximum communication range must be maintained according to the threshold analysis for optimal communicating range for a sensor node [25].

Fig. 2.10 shows the comparison of DCM with the energy model proposed in [18]. The energy model in [18] considered fixed clusters with single hop transmission. For comparison, we set all the parameter values similar to the parameter values in [18]. It is shown in the figure that the single hop network is more efficient compared to the multi-hop network. The energy consumption is increased with the increase of communication range of a node. It is also shown that with fixed network field size, the energy consumption is increased with the increase in the density of the network.

Fig. 2.11 shows comparison of the average lifetime versus sleeping time schedule of a node using DCM, with the models proposed in [18], [33], [34]. We set all the parameters same as used in [18] for lifetime analysis of a node. For DCM, we kept the packet loss rate to 10% of the total packet transmitted. For lifetime analysis, we also considered the energy consumption through a sensor processing, logging, sensing, and actuation along with the energy consumption due to data communication. It is indicated in the figure that

the overall lifetime of a node will be decreased with the increase in the packet loss rate.

Therefore, if packet loss rate does not take into account, just like in previous energy models, then the measurement of the lifetime of a node will be over estimated.

Consequently, it indicates that the packet loss rate is fundamental issue to consider while precisely estimating the lifetime of the network.

Fig. 2.12 shows the average energy consumption for a packet that is successfully delivered at the sink node. Simulation was performed by sending 300 packets to the sink with total simulation time as 1000 seconds. The result is based on the number of packets delivered at the sink. It is shown that the average energy consumption per packet in both GOSSIP and Directed Diffusion [36] routing techniques is slightly decreased with the increase in the packet delivery ratio.

Fig. 2.13 shows the comparison of energy efficiency by varying the number of hops and communication range of a node. It is shown in the figure that a network will be nearly 100% efficient, if nodes are directly connected with each other. However, due to limited communication range, a node has to communicate with the sink via multiple hops.

It is shown in the figure that energy efficiency is gradually decreased with the increase of number of hops from source to the sink. Similarly, the energy efficiency can be achieved with the decrease of communication range with the condition that communication range must maintain the upper limit of optimum range. It means that multi-hop communication is less efficient in terms of energy compared to single hop when the source is in the

Figure 2.11: Lifetime (in days) versus sleeping time of a node with different energy models having AA Alkaline batteries (1500mAh), here,|S|=100, l=100m, lffor DCM = 10 %.

Figure 2.12: Successful delivery of packets versus average energy consumption per packet using GOSSIP and Directed Diffusion.

Figure 2.13: Comparison of energy efficiencyη(h)with different number of hops having different communication range (rc). Here |S|=100, lf= 0 %,L=100m,λ =1Kbps.

communication range of the sink.