Performance Evaluation and Validation of QCM (Query Control Mechanism) for QoS-Enabled Layered-Based Clustering for Reactive Flooding in the Internet of Things

Internet of Things (IoT) facilitates a wide range of applications through sensor-based connected devices that require bandwidth and other network resources. Enhancement of efficient utilization of a heterogeneous IoT network is an open optimization problem that is mostly suffered by network flooding. Redundant, unwanted, and flooded queries are major causes of inefficient utilization of resources. Several query control mechanisms in the literature claimed to cater to the issues related to bandwidth, cost, and Quality of Service (QoS). This research article presented a statistical performance evaluation of different query control mechanisms that addressed minimization of energy consumption, energy cost and network flooding. Specifically, it evaluated the performance measure of Query Control Mechanism (QCM) for QoS-enabled layered-based clustering for reactive flooding in the Internet of Things. By statistical means, this study inferred the significant achievement of the QCM algorithm that outperformed the prevailing algorithms, i.e., Divide-and-Conquer (DnC), Service Level Agreements (SLA), and Hybrid Energy-aware Clustering Protocol for IoT (Hy-IoT) for identification and elimination of redundant flooding queries. The inferential analysis for performance evaluation of algorithms was measured in terms of three scenarios, i.e., energy consumption, delays and throughput with different intervals of traffic, malicious mote and malicious mote with realistic condition. It is evident from the results that the QCM algorithm outperforms the existing algorithms and the statistical probability value “P” < 0.05 indicates the performance of QCM is significant at the 95% confidence interval. Hence, it could be inferred from findings that the performance of the QCM algorithm was substantial as compared to that of other algorithms.


Introduction
The Internet of Things (IoT) has become quite famous in the recent years in that many of our daily routine devices are being connected with us, covering many capabilities such as sensing, autonomy, and contextual awareness [1]. The IoT, resulting from Internet progress and the innovative evolution of smart devices, has led to the development of new computing prototypes. IoT is the next revolutionary technology that converts the present communication infrastructure into a completely  Figure 1 shows the system model with different sensor motes communicating with each other between the physical (sensor) and network layers of IoT. From the figure, it can be seen that redundant messages, unwanted queries and network flooding are the major causes of inefficient utilization of resources, thus resulting in IoT devices consuming more energy with a high computational time (i.e., delay in data transmission), which in turn affects the network QoS [4]. Moreover, solving these issues in IoT networks is demanding due to the constraint nature of the devices with limited energy. Presently, to the best of our knowledge, no mechanisms for identification of redundant queries have been developed in this domain.
We examine several approaches for tackling unwanted and redundant communication in IoT networks to enable us to understand the sequence of actions that take place when flooding happens and propose a Cluster-Based Flooding (CBF) technique. The proposed technique is an interoperable solution both for physical layer and network layer devices. CBF divides the network into different clusters; local queries information are proactively maintained by the Intralayer clustering (IALC), while Interlayer clustering (IELC) is responsible for reactively obtaining the routing queries to the destinations outside the cluster. CBF is a hybrid approach, having the potential to be more efficient than traditional schemes in term of query traffic generation.
A QoS-enabled QCM model is developed, and the results of the simulation show the superior performance against state of the art approaches in terms of traffic delay, QoS throughput, and energy consumption, under various performance metrics compared with traditional flooding and state of the art. In order to figure out real understanding of flooding in IoT networks, we provide modeling of the redundant queries which leads to flooding in Figure 2 [34]. We examine several approaches for tackling unwanted and redundant communication in IoT networks to enable us to understand the sequence of actions that take place when flooding happens and propose a Cluster-Based Flooding (CBF) technique. The proposed technique is an interoperable solution both for physical layer and network layer devices. CBF divides the network into different clusters; local queries information are proactively maintained by the Intralayer clustering (IALC), while Interlayer clustering (IELC) is responsible for reactively obtaining the routing queries to the destinations outside the cluster. CBF is a hybrid approach, having the potential to be more efficient than traditional schemes in term of query traffic generation.
A QoS-enabled QCM model is developed, and the results of the simulation show the superior performance against state of the art approaches in terms of traffic delay, QoS throughput, and energy consumption, under various performance metrics compared with traditional flooding and state of the art. In order to figure out real understanding of flooding in IoT networks, we provide modeling of the redundant queries which leads to flooding in Figure 2 [34].

Motivation
The Internet of Things has gained substantial attention over the last few years because of connecting daily things in a wide range of application and domains. Many sensors require bandwidth and network resources to give-and-take queries among heterogeneous IoT networks. The network sometimes becomes unable to handle unwanted and redundant queries generated

Motivation
The Internet of Things has gained substantial attention over the last few years because of connecting daily things in a wide range of application and domains. Many sensors require bandwidth and network resources to give-and-take queries among heterogeneous IoT networks. The network sometimes becomes unable to handle unwanted and redundant queries generated from different smart devices. In addition, the network is also not able to prioritize the important queries among flooded queries. This research developed a new query control mechanism that could manage priority queries and refrain redundant and unwanted queries. This idea was able to save time and resources of networks with an efficient query management. Further, the performance evaluation of such query control mechanisms required inferential analysis of simulated results to statistically validate the performance parameters of QCMs under discussion.

Problem Statement
Network flooding is a key questioning strategy for successful exchange of queries. However, the risk of the original flooding is prone to unwanted and redundant network queries which may lead to cause heavy network traffic. Redundant, unwanted and flooded queries are the major cause of inefficient utilization of resources. IoT devices consume more energy and high computational time as compare to wireless sensor networks [15]. More queries lead to consumption of bandwidth, increase cost, and degrade QoS. Current existing approaches focus primarily on how to speed up the basic routing for IoT devices. However, solutions for flooding are not being addressed. This research proposed a new query control mechanism and evaluated its performance by statistical means.  Table 1 presents the important illustration of symbols and abbreviations used this the methodology. The research considered the following two hypotheses for inferential analysis, Alternative hypothesis H 1 (µ 2− µ 1 > 0): There is statistical significant relationship between the proposed QCM algorithm and other existing algorithms (DnC, SLA, and Hy-IoT) for QoS-enabled layered-based clustering for reactive flooding in the Internet of Things.

Methodology
Further, the researcher performed a T-test and an ANOVA test for the above hypothesis testing. Let the Sample mean difference be where µ 1 is the sample mean of the data set of results for the first algorithm and µ 2 is the sample mean of the data set of results for the second comparable algorithm. Sample standard deviation Here, data points are x 1 , x 2, x 3, , , x N in the data sets of results of the two comparable algorithms.
Paired Sample T-test: Here, n represents the number of observations. We find the probability value (p) by observing the test statistics under the null and alternative hypothesis. This probability would help to identify the magnitude of the significance in the results for our proposed QCM algorithm.
The test calculates the probability value (P-value) based on the data sets of the results for different comparable algorithms. The standard confidence interval is 0.05 (95% confidence interval); P-values less than 0.05 are considered statistically significant. On the contrary, P-values larger than the chosen confidence interval infer that performances of comparable algorithms have no statistical significance and hence no algorithm outperformed the other algorithms in this comparison.
In addition, the authors performed an "ANOVA test" [35][36][37] to validate the performance measure of algorithms. The ANOVA test contains the following features, Mean square for samples, Similarly, the mean square for error, Now the F statistics becomes This research, by statistical means, evaluates the performance of different QoS-enabled layered-based clustering algorithms for reactive flooding in the Internet of Things with the following measures.

1.
Inferential analysis in terms of Energy Consumption a. Energy consumption with different intervals of traffic b.
Energy consumption with malicious mote c.
Energy consumption with malicious mote with a realistic condition 2. Delay with malicious mote c.

Inferential analysis in terms of Delay
Delay with malicious mote with a realistic condition 3.

a.
Throughput with different intervals of traffic b.
Throughput with malicious mote c.
Throughput with malicious mote with a realistic condition Based on our hypothesis theories stated above, we find the probability value "P" employing the statistical t-test to figure out acceptance or rejection of our Null hypothesis (or alternative hypothesis) as a metric for performance evaluation of proposed and existing algorithms.

Results
The performance estimation and evaluation of the proposed technique against up-to-date DnC, SLA [2,3] and Hy-IoT [32] methods for tracing and mitigating the unwanted and redundant reactive flooding are described in this section. Routing protocol and MDP protocol [24] are ad-hoc routing and Contiki, respectively. To obtain the appropriate results, simulation is performed 60 times based on the following three scenarios. We present here the rejection of the Null hypothesis and acceptance of the alternative hypothesis since the QCM algorithm outperforms (95% confidence interval) the existing algorithms, i.e., DnC, SLA, and Hy-IoT for QoS-enabled layered-based clustering for reactive flooding in the Internet of Things.

Case 1: Inferential analysis in terms of energy consumption
Here, in this case, we discuss the performance evaluation in terms of energy consumption with three different scenarios, i.e., different intervals of traffic, malicious mote and malicious mote with realistic conditions. Figure 3 depicts energy consumptions with respect to different scenarios, i.e., different intervals of traffic, with malicious mote and with malicious mote and with realistic conditions. The proposed QCM technique outperformed DnC, SLA, and Hy-IoT approaches in term of dropping the average consumption of energy. Because the proposed technique is capable of detecting flooder motes and detaching them from the network, this reduces the level of energy consumption that arises during redundant and unwanted flooding attacks, whereas the average energy consumption of DnC and SLA is approximately 21 and 18%, respectively, from (1 to 5) seconds of intervals, and this ratio continuously rises as the interval increases. However, in the case of the proposed mechanism, the ratio of the consumption of energy falls to 6% as compared to 13% in the existing Hy-IoT approach. consumption of energy. Because the proposed technique is capable of detecting flooder motes and detaching them from the network, this reduces the level of energy consumption that arises during redundant and unwanted flooding attacks, whereas the average energy consumption of DnC and SLA is approximately 21 and 18%, respectively, from (1 to 5) seconds of intervals, and this ratio continuously rises as the interval increases. However, in the case of the proposed mechanism, the ratio of the consumption of energy falls to 6% as compared to 13% in the existing Hy-IoT approach. We can observe that QCM achieves the lowest energy consumption as compared to other prevailing algorithms in the described scenarios. We can observe that QCM achieves the lowest energy consumption as compared to other prevailing algorithms in the described scenarios. Table 2 presents the statistical observations of data related to the inferential analysis of QCM and other existing algorithms. We can see that the statistical significant value P is less than our chosen confidence interval of 0.05 which is evidence that our proposed QCM algorithm outperforms the existing algorithms. Hence, the Null hypothesis is rejected and QCM achieves the significant prediction value in the desired confidence interval. An exact realistic analysis of QCM is conducted to find the level of mischievous motes during flooding expansion in the network. It is evident from the result that in the presence of malicious motes, the level of energy consumption increases gradually. At malicious mote 2, the levels of energy consumption are approximately (8 and 5%) for DnC and SLA approaches respectively, and at malicious mote 15, this consumption level reaches approximately (48 and 40%). Hence, by introducing QCM, this level falls to approximately (2, 4, and 20%) at malicious mote (2, 6 and 15), respectively. Table 3 presents the ANOVA test statistics of the proposed QCM algorithm compared with other algorithms. We can find here that "F statistics" values are sufficiently larger than "F critical values". In addition, the "P values" are less than 0.05, which achieves our 95% confidence interval, showing that the proposed QCM algorithm outperforms the existing algorithms evaluated through inferential analysis.

Case 2: Inferential analysis in terms of Delay
The effect of traffic delay on the number of malicious motes, the time interval and malicious motes under realistic conditions are described in this section. QCM outperformed DnC, SLA and Hy-IoT by having the least traffic delays. QCM has the ability to detect, pause and detach the flooding mote from the network, which helped in improving its performance. On the other hand, the redundant and unwanted queries were also removed by detaching the flooding motes.
Here, in this case, we discuss the performance evaluation in terms of delay with three different scenarios, i.e., different intervals of traffic, malicious mote and malicious mote with realistic conditions. Figure 4 presents the "delay" with respect to different scenarios, i.e., different intervals of traffic, with malicious mote and with malicious mote and with realistic conditions. We can observe that QCM achieves the lowest delay as compared to other prevailing algorithms in described scenarios. Table 4 presents the statistical observations of data related to the inferential analysis of QCM and other existing algorithms. We can see that the statistical significance value P is less than our chosen confidence interval 0.05 which is evidence that our proposed QCM algorithm outperforms the existing algorithms. Hence, the Null hypothesis is rejected and QCM achieves the significant prediction value in the desired confidence interval.
Sensors 2020, 20, 283 9 of 15 redundant and unwanted queries were also removed by detaching the flooding motes.
Here, in this case, we discuss the performance evaluation in terms of delay with three different scenarios, i.e., different intervals of traffic, malicious mote and malicious mote with realistic conditions. Figure 4 presents the "delay" with respect to different scenarios, i.e., different intervals of traffic, with malicious mote and with malicious mote and with realistic conditions. We can observe that QCM achieves the lowest delay as compared to other prevailing algorithms in described scenarios.  Table 4 presents the statistical observations of data related to the inferential analysis of QCM and other existing algorithms. We can see that the statistical significance value P is less than our chosen confidence interval 0.05 which is evidence that our proposed QCM algorithm outperforms   Table 5 presents the ANOVA test statistics of the proposed QCM algorithm compared with other algorithms. We can find here that "F statistics" values are sufficiently larger than "F critical values". In addition, the "P values" are less than 0.05, which achieves our 95% confidence interval, showing that the proposed QCM algorithm outperforms the existing algorithms evaluated through inferential analysis.

Case 3: Inferential analysis in terms of throughput
The proposed QCM is compared with DnC, SLA, and Hy-IoT using network throughput which we refer to as QoS. Here, QoS is measured for these four flooding mechanisms using three scenarios: time interval, increasing number of malicious motes and malicious motes with realistic network conditions.
Here, in this case, we discuss the performance evaluation in terms of throughput with three different scenarios, i.e., different intervals of traffic, malicious mote and malicious mote with realistic conditions.  Figure 5 presents the "Throughput" with respect to different scenarios, i.e., different intervals of traffic, with malicious mote and with malicious mote and with realistic conditions. We can observe that QCM achieves the highest throughput as compared to other prevailing algorithms in the described scenarios.

Case 3: Inferential analysis in terms of throughput
The proposed QCM is compared with DnC, SLA, and Hy-IoT using network throughput which we refer to as QoS. Here, QoS is measured for these four flooding mechanisms using three scenarios: time interval, increasing number of malicious motes and malicious motes with realistic network conditions.
Here, in this case, we discuss the performance evaluation in terms of throughput with three different scenarios, i.e., different intervals of traffic, malicious mote and malicious mote with realistic conditions. Figure 5 presents the "Throughput" with respect to different scenarios, i.e., different intervals of traffic, with malicious mote and with malicious mote and with realistic conditions. We can observe that QCM achieves the highest throughput as compared to other prevailing algorithms in the described scenarios.  Table 6 presents the statistical observations of data related to the inferential analysis of QCM and other existing algorithms. We can see that the statistically significant value P is less than our chosen confidence interval of 0.05 which is evidence that our proposed QCM algorithm outperforms the existing algorithms. Hence, the Null hypothesis is rejected and QCM achieves the significant prediction value in the desired confidence interval. Table 7 presents ANOVA test statistics of the proposed QCM algorithm compared with other algorithms. We can find here that "F statistics" values are sufficiently larger than "F critical values". In addition, the "P values" are less than 0.05, which achieves our 95% confidence interval, showing that the proposed QCM algorithm outperforms the existing algorithms evaluated through inferential analysis.  Table 6 presents the statistical observations of data related to the inferential analysis of QCM and other existing algorithms. We can see that the statistically significant value P is less than our chosen confidence interval of 0.05 which is evidence that our proposed QCM algorithm outperforms the existing algorithms. Hence, the Null hypothesis is rejected and QCM achieves the significant prediction value in the desired confidence interval. Table 6. Inferential analysis of the proposed QCM algorithm in terms of "Throughput" scenarios.  Table 7 presents ANOVA test statistics of the proposed QCM algorithm compared with other algorithms. We can find here that "F statistics" values are sufficiently larger than "F critical values". In addition, the "P values" are less than 0.05, which achieves our 95% confidence interval, showing that the proposed QCM algorithm outperforms the existing algorithms evaluated through inferential analysis. Table 7. ANOVA statistics of the proposed QCM algorithm in terms of "Throughput" scenarios.

Discussion (Hypothesis Testing)
This research was based on the hypothesis that the proposed QCM (Query Control Mechanism) algorithm (Khan, F. A., Noor, R. M., Mat Kiah, M. L., Noor, N. M., Altowaijri, S. M., & Rahman, A. U., 2019) outperforms the other existing algorithms, i.e., DnC, SLA, and Hy-IoT for QoS-enabled layered-based clustering for reactive flooding in the Internet of Things. The study elaborated numerous defensive techniques against unwanted and redundant routing queries which lead to heavy network traffic and flooding in IoT networks. In this study, the authors implemented the reactive part Interlayer clustering (IELC) of Cluster based flooding (CBF) and proposed a Query control mechanism (QCM) to detect and terminate the unwanted and redundant queries based on link signal strength, consistency of query packet and query limit threshold.
It is evident from the results that the proposed QCM had superior performance compared with the state of the art defensive techniques in terms of the average consumption of energy, traffic delay, and QoS which we referred to as network throughput. Thus, QCM drops the average consumption of energy to a significant rate as compared to the DnC, SLA, and Hy-IoT under varying intervals of traffic. The performance of QCM is also better regarding average consumption of energy with malicious motes against the traditional approaches by droping the consumption at different motes. Additionally, QCM also exhibits dominant performance regarding network delay by decreasing the delay as compared to the state of the art.
In the case of malicious motes, the proposed QCM drops the network delay to a significant level. Lastly, QCM enhances the amount of QoS to a greater extent as compared to Hy-IoT. The Proposed QCM technique employs the Query Limit Threshold (QLT) for detecting and terminating the redundant and unwanted query request packets, and in this way boosts the IoT network performance in terms of signal strength of query packets and improves the location consistency checking of connected motes to keep the network away from reactive flooding attacks.
This performance clearly shows the difference between our approach and the contemporary approaches. We plan to extend this work in the future by considering a discrete component circuit implementation model using Bouali's system to detect some other attacks in IoT by extending the number and types of motes in order to test the reliability of our approach in the presence of many motes. Also, we plan to include the proactive part Intralayer clustering (IALC) of the CBF, which is favorable in high priority and less delay IoT networks, i.e., smart transportation, smart health, and smart security, and to model a physical prototype for it.
The statistical tests calculated the probability value, "P-value", based on the data sets of results for different comparable algorithms. We kept the standard confidence interval as 0.05 to determine the 95% confidence interval. "P-values" less than 0.05 were considered statistically significant. On the contrary, "P-values" larger than the chosen confidence interval inferred that the performance of comparable algorithms had no statistical significance for results and hence no algorithm outperformed the other algorithms in this comparison.
This research employed statistical measures to evaluate the performance of different QoS-enabled layered-based clustering algorithms for reactive flooding in the Internet of Things with the following measures. The inferential analysis was performed in the context of Energy Consumption (with different intervals of traffic, with malicious mote and with malicious mote with realistic conditions). Similarly, Inferential analysis was performed in terms of Delay (with different intervals of traffic, with malicious mote and with malicious mote with realistic conditions). Further, the research estimated the inferential measures in the context of Throughput (with different intervals of traffic, with malicious mote and with malicious mote with realistic conditions).
Based on our hypothesis theories stated earlier, we found the probability value "P" (in all statistical evaluations) remained less than 0.05, which rejected the Null hypothesis that there was no statistical significance of results for the proposed QCM algorithm as compared to the results of other existing algorithms, i.e., DnC, SLA, and Hy-IoT for QoS-enabled layered-based clustering for reactive flooding in the Internet of Things. Further, in the context of the alternative hypothesis, the evaluation of performance measures revealed that the alternative hypothesis was accepted since the proposed QCM algorithm outperformed the other existing algorithms.

Conclusions
This research article presented a statistical performance evaluation of different query control mechanisms. The performances of such query control mechanisms rely on minimizing the energy consumption, cost and network flooding. This article simulated and evaluated the performance measure of different query control mechanisms for QoS-enabled layered-based clustering for reactive flooding in the Internet of Things. By statistical means, we infer the significant achievement of the QCM algorithm (Khan, F. A., Noor, R. M., Mat Kiah, M. L., Noor, N. M., Altowaijri, S. M., & Rahman, A. U., 2019) that outperformed the prevailing algorithms, i.e., DnC, SLA, and Hy-IoT for identification and elimination of redundant flooding queries. The inferential analysis for performance evaluation of algorithms was measured in terms of energy consumption with energy consumption, delay and throughput with different intervals of traffic, malicious mote and malicious mote with realistic conditions. It is evident from the results that the QCM algorithm outperforms the existing algorithms, depicting the statistical probability value "P" < 0.05, indicating the performance of QCM significantly achieved the 95% confidence interval. Hence, the performance of the QCM algorithm is significant as compared to the performance of other algorithms.