Introduction

Over the past decades, the idea of the Internet of Things (IoT) has become increasingly common in both academia and industry because of its ability to communicate and interact with thousands of small and self-configuring intelligent objects [1]. These objects range from measuring instruments, actuators, and sensors to different kinds of intelligent products, including smart meters, medical devices, wearable devices, smart home devices, and smart city devices.

RPL is a routing protocol for low-power and lossy networks and IoT networks. The IoT network consists of a large number of restricted nodes, which may include energy limitations, storage and processing limitations, low data rate, high reliability, low energy consumption, frame size limitations, low communication range, and a constantly changing network topology [2, 3].

A typical RPL in the LLN network consists of a few to thousands of resource-restricted sensor nodes with some routing capabilities connected to the Internet via a special node called a sink (root) node that does not have these restrictions on its own. A typical RPL network topology is illustrated in Fig. 1.

Fig. 1
figure 1

The RPL network topology

Full size image

Figure 2 shows the energy consumption of each part of the sensor node [4]. Based on these various operating modes, we can calculate the energy consumption of each node. There are many limitations and drawbacks of the RPL protocol, such as a single routing path, with the absence of load balancing that causes exhaustion of the energy of parents with overload, resulting in network breakdowns and problems of unreliability, and there are no guidelines for how metrics should be combined under the metric composition definition for the objective function.

Fig. 2
figure 2

Power consumption analysis of each part of the sensor node [4]

Full size image

In the OFs of the RPL protocol, by considering the cost of its respective paths, the cost of routing a particular route is determined, so a path with a higher hop count would appear to be more expensive than another path with a limited hop count, whereas the communal of the first path may have higher quality.

In the original RPL protocol with OF0, when making routing decisions, this can be frustrating, because there would be a better probability of choosing the route with the smallest hop count, but it could have one or more individual paths of very poor quality. In MRHOF, the rank calculation depends only on one (ETX metric), and the route with a smaller ETX is better, which causes many problems, because only one metric (ETX) is not sufficient to make a route decision, neglecting other important metrics such as the load on the node, delay, and node residual energy [5,6,7,8,9].

The remainder of this paper is organized as follows. First, “Related works" provides an overview of the RPL-related work that attempts to enhance the RPL objective function and discusses the drawbacks of these studies. “Proposed model” describes the proposed model to improve the RPL protocol. In “Simulation results evaluation”, the simulation results are evaluated with three network scenarios and the results of the proposed protocol are presented and compared with the RPL MRHOF in terms of PDR, packet loss ratio, and total power consumption. Finally, “Conclusion” concludes the paper.

Related Works

In Ref. [10], the authors defined path availability as a new routing metric that can describe the quality and quantity of diverse network resources, and then used this measure to quantify the node’s battery lifetime, is based on a demand-driven model. Smart packets make routing decisions in this algorithm, they may also construct partial broadcast query and await-reply cycles to aid in the identification of neighbors and routes. The simulation results indicate how smart packets may use this new metric to construct more energy-efficient pathways, extending the network’s operational lifetime. They are looking at new ways to use cognitive packet-based networks, such as using path availability to simulate extra network resources to aid in the creation of higher quality packet routes.

In Ref. [11], the researcher presents a novel energy-efficient routing method for mobile ad hoc networks. To construct a resilient routing protocol, the proposal uses the concept of learning from a prior study on cognitive packet networks (CPN). Smart packets are used to seek for routes, and they take use of unicasts and broadcasts. Unicasts are favored, because they have a smaller total overhead. Smart packets use a composite objective function that takes into account both the energy stored in the nodes and the route delay to learn how to make excellent unicast routing decisions. The overall outcome is a dynamic path discovery that strikes a balance between low-latency routes and optimal network resource use, extending the network’s lifetime.

In Ref. [12], an energy aware routing protocol (EARP) is proposed by the authors, which combines the network’s total consumed power and the flows’ declared QoS. The algorithm employs a source routing, which is based on the Cognitive Packet Network’s features, and runs autonomously at each network input node based on smart packets that acquire important information throughout the network using reinforcement learning at each intermediary node. EARP tries to keep the “damaged” to the packets’ major QoS metric to a value that is below an acceptable upper bound while attempting to minimize the power consumption of every flowing in a network. The simulation result shows that EARP utilizes less power and maintained the specified QoS level when compared to a purely QoS-driven method.

In Ref. [13], the authors proposed additive and lexical composition methods that merge multiple metrics for routing to maximize different performance aspects. They point out that a loop-free routing protocol must be preserved by the combined metric’s monotonicity property. However, when using lexical composition, this restriction is not necessary. This research proposes a combined packet forwarding indicator (PFI) and hop count (HC) metrics to establish shorter pathways that avoid malicious or selfish behavior of nodes. The results of the simulation show that the lexical mixture of multiple metrics offers better identification and accurate behaving badly of node route selection while presenting equivalent latency compared to the hop count metric only. The research also stated that combining hop count (HC) and residual energy (RE) metrics in either a lexical or an additive method result in better energy distribution between nodes compared to hop count alone. While this study offers a strong proposal for the distribution of energy loads between nodes, by combining the hop count and the RE metric, the impact of this mixture on the efficiency of the network is not clarified, which is a critical performance parameter. If the analysis uses an aggregated RE metric value, the local optimum value is unclear.

In Ref. [14], the authors are interested in the RPL issue, which depends only on one metric, and pointed out the problem of unbalanced traffic and, consequently, the unequal distribution of power consumption among the RPL network nodes. The study pointed out that the use of the expected transmission count (ETX) in the RPL network as a single metric would lead to excessive use of these routes, particularly those with high packet delivery rates. This excessive use of excellent-quality paths eventually contributes to network partitioning and reduces the total lifetime of the network. On the other hand, the performance of the route could be negatively influenced if energy was chosen as the sole metric of routing metric. This study proposed a new weighted energy-oriented hybrid metric that, in addition to the ETX, considers the residual energy of a node balancing node energy consumption while providing highly reliable routes. The results show that the suggested strategy, which increases the lifetime of the network by up to 12 percent, somewhat balances the energy usage. A major issue with this study is that in the simulation experiments, only up to six nodes were used, which may be insufficient to draw the conclusions stated. In addition, the developers did not include information on the influence of these metrics on network stability.

In Ref. [15], the authors proposed an RPL network energy-efficient and reliable composite metric that considers the reliability indicated by that of the energy efficiency and ETX metric that balances power consumption between nodes and improves the network lifetime. The suggested new routing metric is called the lifetime and latency aggregate-able metric (L2AM). A node running L2AM first computes the transmitting power of a node’s relation and residual power using an exponential equation to produce, which is called the primary metric. To obtain the total metric cost, the ETX metric is multiplied by the primary metric, which needs to be minimized when selecting the parent node of choice. In terms of the remaining capacity and network lifetime, the proposed metric is compared with the ETX RPL for evaluation purposes. The results showed that L2AM outperformed the ETX of RPL by 56% in terms of the life of the network. Although the research notes that the network’s lifetime benefit is achieved without affecting the network’s performance, the study does not show any reliable analysis results or explain how this conclusion is drawn by the writers. In addition, for assessment purposes, researchers are using their own simulator program, which might lack functionality comparison with well-known simulation tools such as the Cooja simulator. The studies set the trickle timer interval for emitting DIOs at one h. It seems that in their simulations, the authors implemented only one interval, which is a confused deviation from the standard trickle routing protocol operation.

In Ref. [16], as RPL focuses on one metric, such as hop count or approximate transmission cost, the study discusses the long single-hop problem introduced in large-scale networks. The authors found that the number of hops tends to have a greater effect on the measured rank than on the transmission efficiency, because the ETX metric adds the ETX values of the nodes along a routing path. When RPL relies on only one metric in large-scale networks, such as hop count or estimated transmission cost, the authors examined the long single-hop problem introduced. Because the ETX adds the ETX amounts of the nodes along the routing path, instead of the transmission efficiency, the hop count appears to have a greater effect on the measured rank. Therefore, as this passes across fewer nodes and decays a comparatively smaller cumulative ETX, a node would appear to choose a path with a limited hop count. Therefore, even though such a route has constituent links with very poor quality, the calculated ETX rank for a route with fewer hops tends to be lower. In a wide network, a long single-hop route with low-quality transmission can restrict the entire network and adversely affect network reliability. To address this issue, we propose mixing the hop count and ETX metrics to provide a combined metric called PER-HOP ETX. Based on the cumulative amount of ETX, the rank is determined along a path separated by the hop count on that path. Using the Cooja simulator tool, the new metric is computed and compared with the MRHOF and OF0 objective functions. The findings suggest that in dense networks, PER-HOP ETX enhances PDR while decreasing power consumption and latency. In this paper, a major problem with the suggested metric is that the combined metric's monotonicity property is not met, so that the network could be in danger of looping.

The advantage of the proposed method is explained; however, there is no reason to justify why the fuzzy method is much more stable. For the slightly improved delay, there is also the lack of justification. In Ref. [17], three routing metrics, namely ETX, delay, and energy, were combined with the authors using a two-stage fuzzy system. First, the delay and ETX are mixed to determine QoS. The energy and calculated QoS values were combined in the second step. The suggested fuzzy-based method was then tested against the ETX of the original protocol using a true network of 28 sensor nodes. The two protocols are contrasted in terms of PDR, power consumption, and network stability or churn (number of parent changes). It is stated that the fuzzy-based approach outperforms ETX by up to 20% in terms of packet loss ratio and marginally improves end-to-end latency marginally. In addition, it is shown that the proposed solution produces a more stable route topology with an average change per hour of 6.63 parents, compared to the ETX, with an average change per hour of 43.522 parents compared to ETX.

In Ref. [18], the authors implemented a new RPL OF called the CA-OF. In addition to the ETX, this OF implies a new routing metric for buffer occupancy (BO). By choosing parents around less congested routes, the newly suggested OF aims to increase network traffic reliability. Depending on the traffic intensity, the resulting additive metric applies adaptive weights to the ETX and BO metrics. Therefore, the proposed OF depends on the ETX metric for selecting a candidate parent with low traffic. As soon as the network appears to be congested, the ETX routing metric is refused and only the BO for the next hop selection process is considered by the OF. According to the authors, the CA-OF can achieve better network performance in terms of power consumption, reliability, and throughput.

In Ref. [19], the authors suggested an improved RPL (I-RPL), where the OF for parent node selection relies on an LCI. The LCI index, for example, link quality and node energy, was deduced from various metrics. Therefore, the next hop with the greatest LCI index was chosen for the preferred parent selection. However, if multipath routing is required to alleviate congested nodes, I-RPL also stores a possible parent node that can be used for forwarding traffic. In terms of load balancing, end-to-end delay, and packet delivery, the proposed scheme achieved promising performance.

In Ref. [20], a new OF for efficient routing (OF-ER) was introduced by the authors based on the CER metric used for the best parent node selection. A collection of weighted metrics, namely link quality ETX, packet loss due to queue occupation, node's remaining time to live, delay and bottlenecked nodes observed, is combined with the CER metric. The study stated that not only will OF-ER minimize power consumption and queue loss, but it also extends the expected lifetime of the node. Unfortunately, the researchers did not report the metric effect on traffic overload implemented or its ability to support network scaling.

In the paper [21], by integrating multiple metrics of fuzzy logic, an improved (OF), OFRRT-FUZZY, is proposed. The suggested OFRRT-FUZZY takes link and node metrics into consideration. Received Signal Strength Index (RSSI), Residual Power (RE), and Throughput (TH) are the parameters. The OFRRT protocol was developed with fuzzy logic to improve the efficiency of standard OFs and choose the most effective path to the sink node. The fuzzy inference process (FIP) is used in this proposed protocol, which is defined as “a process of mapping from a given input to an output using fuzzy set theory.” The suggested solution uses three input linguistic variables to calculate a single output linguistic variable: (RE), (TH), and (RSSI). It includes four stages in the process: (1) crisp input fuzzification, (2) rule evaluation, (3) rule output aggregation, and (4) defuzzification. The OFRRT-FUZZY protocol has the better performance in terms of energy consumption and PDR, but the performance of the protocol has not been tested in large-scale networks that contain more than 50 nodes, and did not test the protocol performance with random deployment of the nodes.

In the paper [22], using the additive composition approach, the protocol proposes a new flexible Objective Function dependent on Consumed power, ETX, and Forwarding delay (OF-ECF). This method allows for the development of a new composite metric that nodes use to choose the right parent. Instead of many metric choices, it returns a single definitive point. The OF-ECF protocol has a good performance in terms of PDR, but the results show that this protocol consumes more energy than OF0 and MRHOF protocols, We can conclude that the protocol is not suitable for the Low-Power and Lossy Network environments.

The paper [23] proposes a new method for evaluating RPL efficiency. The OF and the trickle algorithm are the two key components, the RPL-FL means RPL based on the flexible trickle algorithm, and the RPL-EC means RPL-based combined ETX and power consumption. They introduced a new RPL objective function combination called OFEC in their paper, “objective function based combined metric using fuzzy logic method”. They used the hop count to route nodes to the root after combining two key metrics: power consumption and ETX. The method is divided into four steps: first, the fuzzification process, which determines the membership degree of input parameters for fuzzy sets; second, the fuzzy intervention process, which measures the output based on merged inputs; third, the aggregation process, which unifies the outputs; and finally, the defuzzification process, which transforms the fuzzy outputs into a single defined value. Table 1 shows the summary of the RPL protocols.

Table 1 The network parameters
Full size table

In paper [24], the authors suggested a dynamic adaptive threshold routing algorithm (DATRA), which is used to handle the power, ignoring the issue in the Dragonfly topology, resulting in a lower power-delay product than UGAL-LVC-H. This method employs an adaptive threshold mechanism as well as a self-adjusting technique. These two elements are crucial in the development of a power-aware self-adaptive routing algorithm. DATRA also achieved greater efficiency in terms of energy consumption and latency while preserving load-balancing capabilities.

Proposed Model

In RPL, the OFs that are used to select the forwarding node (parent node) may suffer from the problem of long hops when the network becomes complex or heavy because of the large number of nodes when the size of the network increases [25], and because of the single metric RPL objective function, such as hop count [26] or expected transmission count (ETX) [27]. In addition, the flocking effect can be a result of the parent selection scheme in RPL, which refers to the incidence of attracting nodes and continuous switching from one parent to another; consequently, this flocking phenomenon would have a major effect on QoS provisioning, which essentially restricts the services of the IoT application. Because selecting parents in the objective function zero (OF0) does not take link quality into consideration and only tries to find a path with minimum hop count, this metric is not efficient enough, because it does not take into consider the long hop that negatively affects energy consumption and QoS, while MRHOF only takes the ETX metric without considering other metrics, which negatively affects the network performance. Because the objective function, routing parameters, constraints, and local policies to be used in the parent selection and route establishment process can be freely chosen in the RPL, to provide high flexibility, we designed a new objective function that combines three metrics: node residual energy, load metric, and ETX. Every node when it selects the parent and constructs the DODAG from the candidate parent nodes, the node has the minimum ETX, minimum load metric, and maximum node residual energy.

Calculate Residual Energy

As a representation of the network lifetime, we used the residual energy metric (RE). Therefore, when constructing a DODAG and selecting a parent, every node should not select a parent that has low residual energy to prevent choosing nodes that have low energy. Based on its various operating modes, the energy consumption of each node is calculated. These modes are usually listen mode, which includes (listen + receive (RX) + idle modes), transmission mode (TX mode), processing mode (CPU mode), and low-power mode (sleep mode). The current energy consumption can be calculated using Eq. (1) [28]

$$ E_{{{\text{con}}}} \left( a \right) = {\text{ }}P_{{{\text{les}}}} \times T_{{{\text{les}}}} + P_{{{\text{Tx}}}} \times T_{{{\text{Tx}}}} + P_{{{\text{CPU}}}} \times T_{{{\text{CPU}}}} + P_{{{\text{sleep}}}} \times T_{{{\text{sleep}}}} , $$
(1)

where Econ is the energy consumed by node (a), Tmode is the time duration at which the node spends in each operating mode (i.e., Tles, TTx, TCPU, Tsleep), and Pmode represents the power consumed in the corresponding mode at a given time (i.e., Ples, PTx, PCPU, Psleep).

Then, we can calculate the residual energy (RE) as the difference between the maximum node energy (Emax) and the current node energy consumed (Econ). Equation (2) calculates RE

$$ {\text{RE}}\left( a \right) = {\text{ }}E_{{{\text{Max}}}} \left( a \right){\text{ }} - {\text{ }}E_{{{\text{con}}}} \left( a \right), $$
(2)

where Emax(a) represent the max energy of the node (a).

Calculate Load Metric

Network data traffic is the quantity of data transmission over a specific amount of time across the network. A load-balancing method was proposed by the authors in Ref. [29]. The load metric can be used to balance the data traffic in a network. Depending on the number of children present throughout the parent node, the load is calculated. The DODAG nodes broadcast all the participant nodes with a DIO message. The participant node or sender node computes for each preferred parent the number of children. Finally, based on the cumulative number of children present in the link, DODAG generates the rank. The participant node selects the parent from the candidate parent list based on the load metric. The load metric can be calculated using Eqs. (3) and (4), respectively

$$ L(Px) = \sum\limits_{{N = 1}}^{n} {Nt(N)} . $$
(3)

Calculating node traffic based on the total number of children

$$ Nt = \mathop \sum \limits_{{i = 1}}^{n} CN\left( i \right), $$
(4)

where L(Px) represents the load on parent x, NT represents the node traffic, and CN represents the number of children.

Calculate the Proposed Objective Function

The objective function (OF) determines how to construct the DODAG and selects the parents in the RPL network to optimize the path. The proposed objective function composite three metrics to increase energy efficiency. This objective function focuses on issues such as data traffic in multipoint-to-point communications. A bottleneck occurs near the sink node. Becoming a chosen parent, for more children means unbalanced load, high congestion, more loss packet, and more overhead, thereby wasting its own energy even faster than other preferred parents. A load metric has been proposed to solve this problem by providing each chosen parent with the number of children they have. On the basis of that, in the rank measurement, we take into consideration the number of children.

Every node when it selects the parent and constructs the DODAG from the candidate parent nodes, the node has the minimum ETX, minimum load metric, and maximum node residual energy (max RE + min ETX + min Load).

Simulation Results Evaluation

To test and implement the proposed protocol in terms of total energy consumption, packet delivery ratio, and packet loss ratio calculation, and compare its performance with MRHOF using the ETX metric, we used the Cooja simulator/Contiki 3.

Simulation and Network Setup

Contiki is a portable operating system, an open source and lightweight. It is dedicated to WSNs and is widely used in IoT networks. The Cooja simulator in Contiki is a network simulator that allows developers to create a powerful simulation environment that enables them to use completely emulated hardware devices on different scale networks to run their applications [28].

The smart home is the case study of our proposed; first, we designed a 50-node network in our experiment and used a random topology to distribute the nodes in a 100 × 100 m2, and then repeated the same simulations using a 55-node random topology. We fixed the range of transmission to 50 m, the range of interference to 100 m, and the simulation results were obtained after 1200 s. Table 1 lists the network parameters used in the experiment. In this study, several metrics were calculated, such as the total power consumption, packet delivery ratio (PDR), and packet loss ratio.

Simulation Results

In this paper, three scenarios are implemented with different network size to investigate the performance of the proposed protocol and compare the simulation results with the original RPL protocol (MRHOF).

First Scenario Network with 50 Nodes

Figure 3 shows the random distribution of 50 nodes. The original RPL protocol (MRHOF) and the proposed protocol are implemented on this network.

Fig. 3
figure 3

Randomly dissemination of 50 nodes in 100 m × 100 m for the first scenario

Full size image

The power consumption was measured in four modes (LPM + CPU + radio listen + radio transmit) for each node.

Figure 4a shows that the total power consumed by all the nodes in the MRHOF network is 116.914 mW, whereas in the proposed protocol, the total power consumed is 92.659 mW, as shown in Fig. 4b, which means that the total power consumption is decreased by 24.255 mW.

Fig. 4
figure 4

a Total power consumption of the MRHOF for all the nodes in the first scenario. b Total power consumption of the proposed protocol for all the nodes in the first scenario

Full size image

Second Scenario Network with 55 Nodes

Figure 5 shows the random distribution of 55 nodes, and the MRHOF protocol and proposed protocol were implemented on this network.

Fig. 5
figure 5

Randomly dissemination of 55 nodes in 100 m × 100 m for the second scenario

Full size image

Figure 6a shows the total power consumption for all the nodes in the MRHOF network is about 178.092 mw, and in other hand, the total power consumption is about 109.512 mw in the proposed protocol, as shown in Fig. 6b; this means that the power consumption is decreased by 68.58 mw.

Fig. 6
figure 6

a Total power consumption of the MRHOF protocol for all the nodes in the second scenario. b Total power consumption of the proposed protocol for all the nodes in the second scenario

Full size image

Third Scenario Network with 60 Nodes

Figure 7 shows the random distribution of 60 nodes.

Fig. 7
figure 7

Randomly dissemination of 60 nodes in 100 m × 100 m for the third scenario

Full size image

Figure 8a shows the total power consumption for all the nodes in the MRHOF protocol network is about 264.969 mw, while in the proposed protocol, the total power consumption is about 132.809 mw, as shown in Fig. 8b; this means that we decreased about 132.16 mw of power consumption.

Fig. 8
figure 8

a Total power consumption of the MRHOF protocol for all the nodes in the third scenario. b Total power consumption of the proposed protocol for all the nodes in the third scenario

Full size image

Table 2 shows a summary of the power consumption in the three networks (50, 55, and 60 nodes), and shows the difference in the total energy consumption in all modes between the MRHOF protocol and the proposed protocol. In the original RPL protocol with the MRHOF objective function, the table shows that the average CPU power consumption is larger than the proposed protocol CPU power consumption, because, in the MRHOF network, the collision is very high, which leads to the transmission of packets, which in turn requires more processing, which in turn leads to an increase in CPU energy consumption. In the sleeping mode (LPM) power consumption, we find that it is roughly the same for both the MRHOF and the proposed protocol.

Table 2 Summary of average power consumption
Full size table

In the listening and transmission modes, we succeeded in reducing the average listening mode and the average transmitting mode power consumption in our proposed protocol because of the selection of high-quality routes based on the proposed objective function that balances the load, which demonstrated its effectiveness in the successful selection of paths and increased the efficiency of the network.

Packet Delivery Ratio (PDR)

Figure 9 shows the packet delivery ratio with the number of nodes. The packet send ratio in both the MRHOF protocol and the proposed protocol is approximately 20 packets per node in all scenarios. In the first scenario (the network with 50 nodes), we found that the packet delivery ratio in the MRHOF was approximately 0.901, but in the proposed protocol, the packet delivery ratio was approximately 0.97145. In the second scenario (network with 55 nodes), we found that the packet delivery ratio in MRHOF was approximately 0.6898, but in the proposed protocol, the packet delivery ratio was approximately 0.95185. In the third scenario (network with 60 nodes), we found that the packet delivery ratio in MRHOF was approximately 0.53985, but in the proposed protocol, the packet delivery ratio was approximately 0.94575. This indicates that the proposed protocol succeeded in increasing the packet delivery rate in all scenarios.

Fig. 9
figure 9

Packet delivery ratio (PDR)

Full size image

Packet Loss Ratio

Figure 10 shows the packet loss ratio with the number of nodes. In all scenarios, the packet send ratio in both the MRHOF and the proposed protocol is approximately 20. In the first scenario (the network with 50 nodes), we found that the packet loss ratio in the MRHOF protocol was approximately 0.02655, but in the proposed protocol, the packet loss ratio was approximately 0.001. In the second scenario (network with 55 nodes), we found that the packet loss ratio in MRHOF was approximately 0.12685, but in the proposed protocol, the packet loss ratio was approximately 0.00185. In the third scenario (network with 60 nodes), we found that the packet loss ratio in the MRHOF protocol was approximately 0.2322, but in the proposed protocol, the packet loss ratio was approximately 0.02205. This means that the proposed protocol decreased the packet loss ratio.

Fig. 10
figure 10

Packet loss ratio (PLR)

Full size image

Total Power Consumption

Figure 11 shows the total power consumption with respect to the number of nodes. In the first scenario (the network with 50 nodes), we found that the total power consumption in the MRHOF protocol was approximately 116.914 mW, but in the proposed protocol, the total power consumption was approximately 92.659 mW. In the second scenario (network with 55 nodes), we found that the total power consumption in the MRHOF protocol was approximately 178.092 mW, but in the proposed protocol, the total power consumption was approximately 109.512 mW. In the third scenario (network with 60 nodes), we found that the total power consumption in the MRHOF protocol was approximately 264.969 mW, but in the proposed protocol, the total power consumption was approximately 132.16 mw. This means that the proposed protocol decreased the total power consumption of the networks.

Fig. 11
figure 11

Total power consumption comparison between MRHOF protocol and proposed protocol with different number of nodes

Full size image

Conclusion

In this paper, by changing the method of calculating the rank when constructing DODAG and selecting parents, we propose the development of the MRHOF. Thus, we proposed a method that considers three metrics instead of using only one metric in the rank calculations by the objective function. We built a medium-density network for a smart home environment and used the Contiki 3/Cooja simulator to execute our work. In all scenarios (50, 55, and 60 nodes), we found that in terms of packet delivery ratio, the PDR improved by 0.07045, 0.26205, and 0.4059, respectively. In terms of the packet loss ratio, the proposed protocol succeeded in reducing the packet loss ratio by 0.02555, 0.125, and 0.21015 in all scenarios. In terms of power consumption, the proposed protocol succeeded in reducing the total power consumption by 24.255 mw, 68.58 mw, and 132.16 mw, in all scenarios, respectively. Therefore, we conclude that an objective function that uses only one metric to build the DODAG and select the parent node is not sufficient and the objective function that takes more than a single metric into account is more accurate and efficient.