Chapter 5
Results, Verification and Validation
This chapter details the results of the experiments that were conducted. Proof of how these re-sults were verified and validated are also provided. All the data used to create the rere-sults pre-sented in this chapter, can be found on the CD accompanying this dissertation (Appendix F). The North-West University’s Statistical Consultation Service (SCS) assisted with the statisti-cal analysis of the data presented in this chapter. Appendix E contains a letter which confirms this.
5.1
Introduction
The results of the different experiments are presented in Section 5.2. The working of the energy aware routing schemes used in the experiments are also verified as correct in Section 5.2. The accuracy of the SN voltage and current measurements are verified in Section 5.3. A statistical analysis of the results is presented in Section 5.4. This analysis shows that the results are valid. Finally, the results are discussed and the chapter is concluded.
Chapter 5 Results
5.2
Results
This section details and depicts the results of the shortest hop path routing scheme and the MTTPR scheme experiments.
5.2.1
Shortest Hop Path Experiment
The 4 possible shortest hop paths for the testbed setup used can be seen in Figure 5.1. All the experiments were started by powering the SNs in the same sequence. This explains why the path chosen by the shortest hop path routing scheme was the same in all three experiments. The experimental setup is described in Section 4.3.
Chapter 5 Results
Three experiments were conducted for a shortest hop path routing scheme using the experimental setup detailed in the previous chapter. A 3D contour plot of the average node energy consumption, for the three experiments, versus the node deployment can be seen in Figure 5.2. From this plot it is clear that the routing scheme chose the path labelled shortest hop path 2 in Figure 5.1. This is because the energy consumption peaked for the nodes along this path. It can therefore be verified that the route chosen by the routing scheme corresponds to one of the theoretically expected routes. The 3D contour plots of the individual experiments can be found in Appendix D. The average packet delivery ratio of the shortest hop path routing scheme experiments is 98.0 %.
Figure 5.2: A 3D Contour Plot of the Average Node Energy Consumption (J) of the Shortest Hop Path Routing Scheme Experiments Versus the Node Deployment (x,y)
Chapter 5 Results
5.2.2
Minimum Total Transmission Power Routing (MTTPR)
Exper-iment
The two possible MTTP paths for the testbed setup used can be seen in Figure 5.3. All the experiments were started by powering the SNs in the same sequence. This explains why the path chosen by the MTTPR scheme was the same in all three experiments.
Figure 5.3: A depiction of the Possible MTTP Paths
Three experiments were conducted for a MTTPR scheme using the experimental setup detailed in the previous chapter. A 3D contour plot of the average node energy con-sumption, for the three experiments, versus the node deployment can be seen in Figure 5.4. From this plot it is clear that the routing scheme chose the path labelled MTTP 1 in Figure 5.3. This is because the energy consumption peaked for the nodes along this
Chapter 5 Sensor Node Voltage and Current Measurement Accuracy Verification
path. It can therefore be verified that the route chosen by the routing scheme corre-sponds to one of the theoretically expected routes. The 3D contour plots of the indi-vidual experiments can be found in Appendix D. The average packet delivery ratio of the MTTPR scheme experiments is 97.6 %.
Figure 5.4: A 3D Contour Plot of the Average Node Energy Consumption (J) of the MTTPR Scheme Experiments Versus the Node Deployment (x,y)
5.3
Sensor Node Voltage and Current Measurement
Ac-curacy Verification
This section details the verification of the SN voltage measurement accuracy for an idle state as well as the SN current measurement accuracy when the SN is in an idle state, when the SN is transmitting with a minimum transmission power setting (-24.6
Chapter 5 Sensor Node Voltage and Current Measurement Accuracy Verification
dBm) and when SN is transmitting with a maximum transmission power setting (19 dBm). When a SN is in an idle state no transmissions occur, except those used for routing update and discovery packets. The supply current and voltage measurements of 10 SNs are compared to measurements taken using a Tektronix DMM4050 precision digital multimeter. Figure 5.5 depicts each SN’s supply voltage measurement as well as the Tektronix DMM4050’s measurement of each SN’s supply voltage. All the SNs are in an idle state. The supply voltage value measured by the Tektronix DMM4050 for each SN is the average of 200 measurements. The SN voltage measurement is the average value measured by each SN over a period of 5 minutes. The Tektronix DMM4050 measurements were taken over the same 5 minute interval as the SN measurements.
Figure 5.5: Plot of the Sensor Node Supply Voltage as Measured by the Sensor Node and the Tektronix DMM4050 Precision Digital Multimeter for 10 Nodes
The descriptive statistics of the voltage measurements can be seen in Table 5.1. VSN
is the SN supply voltage as measured by the SN itself. VBM is the SN supply
volt-age as measured by the Tektronix DMM4050 precision digital multimeter. The ratio VSN/VBM provides the accuracy of the SN voltage measurements. The mean value,
standard deviation and 95% confidence intervals of this ratio as presented in Table 5.1 was calculated using the data from 10 SNs.
Each SN’s supply current measurement as well as the Tektronix DMM4050’s measure-ment of each SN’s supply current is depicted in Figure 5.6. All the SNs are in an idle
Chapter 5 Sensor Node Voltage and Current Measurement Accuracy Verification
Table 5.1: Voltage Measurements Descriptive Statistics
Mean Standard Deviation -95 % Confidence 95 % Confidence VSN/
VBM 1.009017 0.003486 1.006523 1.011511
state. The supply current value measured by the Tektronix DMM4050 for each SN is the average of 200 measurements. The SN current measurement is the average value measured by each SN over a period of 5 minutes. The Tektronix DMM4050 measure-ments were taken over the same 5 minute interval as the SN measuremeasure-ments.
Figure 5.6: Plot of the Sensor Node Idle State Supply Current as Measured by the Sensor Node and the Tektronix DMM4050 Precision Digital Multimeter for 10 Nodes Descriptive statistics of the idle state current measurements can be seen in Table 5.2. ISN is the SN supply current as measured by the SN itself. IBM is the SN supply
cur-rent as measured by the Tektronix DMM4050 precision digital multimeter. The ratio ISN/IBMprovides the accuracy of the SN current measurements. The mean value,
stan-dard deviation and 95% confidence intervals of this ratio as presented in Table 5.1 was calculated using the data from 10 SNs.
Chapter 5 Sensor Node Voltage and Current Measurement Accuracy Verification
Table 5.2: Idle State Current Measurements Descriptive Statistics Mean Standard Deviation -95 % Confidence 95 % Confidence ISN/
IBM 1.033666 0.008019 1.027502 1.039830
Figure 5.7 depicts each SN’s supply current measurement as well as the Tektronix DMM4050’s measurement of each SN’s supply current. Each node transmits 200 pack-ets with a payload of 50 bytes each every 10 seconds. The transmission power set-ting of each SN is -26.4 dBm. The supply current value measured by the Tektronix DMM4050 for each SN is the average of 200 measurements. The SN current measure-ment is the average value measured by each SN over a period of 5 minutes. The Tek-tronix DMM4050 measurements were taken over the same 5 minute interval as the SN measurements.
Figure 5.7: Plot of the Sensor Node Supply Current, when Transmitting at -26.4 dBm, as Measured by the Sensor Node and the Tektronix DMM4050 Precision Digital Multi-meter for 10 Nodes
The descriptive statistics of the current measurements taken with a transmission power setting of -26.4 dBm can be seen in Table 5.3. ISN is the SN supply current as measured
by the SN itself. IBMis the SN supply current as measured by the Tektronix DMM4050
cur-Chapter 5 Sensor Node Voltage and Current Measurement Accuracy Verification
rent measurements. The mean value, standard deviation and 95% confidence intervals of this ratio as presented in Table 5.3 was calculated using the data from 10 SNs. Table 5.3: Minimum Transmission Power Current Measurements Descriptive Statistics
Mean Standard Deviation -95 % Confidence 95 % Confidence ISN/
IBM 1.017322 0.010565 1.009764 1.024880
Each SN’s supply current measurement as well as the Tektronix DMM4050’s measure-ment of each SN’s supply current is depicted in Figure 5.8. Each node transmits 200 packets with a payload of 50 bytes each every 10 seconds. The transmission power setting of each SN is 19 dBm. The supply current value measured by the Tektronix DMM4050 for each SN is the average of 200 measurements. The SN current measure-ment is the average value measured by each SN over a period of 5 minutes. The Tek-tronix DMM4050 measurements were taken over the same 5 minute interval as the SN measurements.
Figure 5.8: Plot of the Sensor Node Supply Current, when transmitting at 19 dBm, as Measured by the Sensor Node and the Tektronix DMM4050 Precision Digital Multime-ter for 10 Nodes
Descriptive statistics of the current measurements taken with a transmission power setting of 19 dBm can be seen in Table 5.4. ISN is the SN supply current as measured
Chapter 5 Validity of the Results (Statistical Analysis)
by the SN itself. IBMis the SN supply current as measured by the Tektronix DMM4050
precision digital multimeter. The ratio ISN/IBM provides the accuracy of the SN
cur-rent measurements. The mean value, standard deviation and 95% confidence intervals of this ratio as presented in Table 5.4 was calculated using the data from 10 SNs. The results presented in this section verified the precision of the SN current and voltage measurements.
Table 5.4: Maximum Transmission Power Current Measurements Descriptive Statistics Mean Standard Deviation -95 % Confidence 95 % Confidence ISN/
IBM 1.027214 0.016341 1.015524 1.038904
5.4
Validity of the Results (Statistical Analysis)
Figure 5.9 depicts the average energy consumption of the network for the shortest hop path routing scheme experiments, the MTTPR scheme experiments and the control ex-periments. An ANOVA test was performed on the total network energy consumption data. The obtained P-value of 0.000002, which is significantly less than 0.05, indicates that there is at least one statistically significant difference between the tested schemes.
Figure 5.9: A Plot of the Average Network Energy Consumption (J) Versus Routing Scheme
Chapter 5 Discussion
Levene’s test for homogeneity of variances was performed next. The results of this test are presented in Table 5.5. The obtained P-value of 0.158715 , which is greater than 0.05, indicates that the variance between the samples is homogeneous. Finally, a Tukey HSD test is performed to determine if the differences between the routing schemes and control are due to random chance or not. The Tukey test approximate probabilities can be seen in table 5.6. All the P-values are smaller than 0.05 which means that there are statistically significant differences between all the routing scheme experiments as well as the control experiments. The results are therefore statistically significant and valid.
Table 5.5: Levene’s Test for Homogeneity of Variances
MS Effect MS Error F p Total Network Energy Consumption (J) 95.19685 37.46541 2.540926 0.158715
Table 5.6: Tukey HSD Test Approximate Probabilities Routing Scheme Control Shortest Hop Path MTTP Control - 0.000227 0.000227 Shortest Hop Path 0.000227 - 0.037666
MTTP 0.000227 0.037666
-5.5
Discussion
Each SN’s minimum transmission power setting corresponds to -26.4 dBm (0.0000022908 W) and its maximum transmission power setting corresponds to 19 dBm (0.07943 W). From this it is clear that there is a significant difference between the minimum and maximum transmission power settings of the SNs. The MTTPR scheme therefore ex-pects that the power consumption of a SN with its transmission power setting set to the maximum value should be significantly more than that of a SN with its transmission power setting set to the minimum value.
The average energy consumption of the network for the shortest hop path routing scheme experiments, the MTTPR scheme experiments and the control experiments is
Chapter 5 Discussion
depicted in Figure 5.9. During the shortest hop path experiments 4 SNs, with their transmission power setting set to the maximum value, are located on the path chosen by the routing scheme. While only 2 SNs, with their transmission power setting set to the maximum value, are located on the path chosen by the MTTPR scheme. The MTTPR scheme expects that the power consumption of a node with its transmission power set to the maximum value should be 34673.48 times greater than that of a node with its transmission power set to the minimum value. The difference between the average total network energy consumption of the control experiments and that of the shortest hop path (ESHP−C) and MTTPR scheme (EMTTP−C) experiments denotes the
energy consumed as a result of the data generator generating and routing data for each case. The MTTPR scheme expects that ESHP−C should be about twice the value of
EMTTP−C because the shortest hop path experiments have 2 more maximum
transmis-sion power nodes on the path chosen by the routing scheme than that of the MTTPR scheme. However this is not the case. The results show that the ratio ESHP−C/EMTTP−C
is equal to 1.2. To explain this result a closer look needs to be taken at the energy con-sumption characteristics of SNs with different transmission power settings.
Figure 5.10 depicts the energy consumption of 10 nodes when they are in an idle state, when they are transmitting at minimum transmission power and when they are trans-mitting at maximum transmission power. The transtrans-mitting nodes transmit 200 pack-ets, with a payload of 50 bytes each, every 10 seconds. These values are collected over a period of 20 minutes. Let EMax−Idledenote the average difference in energy
consump-tion between the idle nodes and the nodes transmitting at their maximum transmission power setting. The average difference in energy consumption between the idle nodes and the nodes transmitting at their minimum transmission power setting is given by EMin−Idle. The ratio EMax−Idle/EMin−Idle is equal to 2.82. This is significantly less than
the value of 34673.48 which the MTTPR scheme expects. This is because the efficiency of a low power transceiver is significantly lower at low transmission power settings than at high transmission power settings [58].
The MTTPR scheme adds low transmission power hops to the chosen route, expecting a significant saving in power consumption. However because the efficiency of a low
Chapter 5 Conclusion
Figure 5.10: A Depiction of the energy consumption of 10 nodes when they are in an idle state, when they are transmitting at minimum transmission power and when they are transmitting at maximum transmission power
power transceiver is typically lower at low transmission power settings, the actual sav-ing in power consumption is not as great as the MTTPR scheme expects. The MTTPR scheme might (under the right circumstances) choose routes that use the same amount of energy or a greater amount of energy than a shortest hop path protocol would. It is clear that the MTTPR scheme needs to be modified to take the transmission power set-ting dependant efficiency of SN transceivers into account. The majority of simulation environments, used to evaluate the MTTPR scheme (as well as other similar routing schemes including: MTTCP routing, MTRTP routing and CMMBCR routing), also do not take this into account [29], [59], [60], [61], [27], [62].
5.6
Conclusion
This chapter presented the results of the MTTPR scheme and shortest hop path rout-ing scheme experiments. The verification and validation of these results were also de-tailed. The results emphasize the importance of taking the transmission power setting
Chapter 5 Conclusion
dependant efficiency of SN transceivers into account when designing routing schemes and simulating WSNs.
