*4.2. Simulation Results*

#### 4.2.1. Network Lifetime

In this work, the network lifetime is defined as the total time that the nodes are alive, which can be considered the number of rounds. When the residual energy of a node decreases to zero, the node is considered a dead node. In our simulation, we compared three protocols to evaluate the network lifetime while varying the number of nodes in 1000 rounds, as shown in Figure 4. In Figure 4a, we consider a network of 450 nodes in 1000 rounds with a transmission range at 200 m; the results show that nodes began to run out of energy at round 300 for LEACH, DBR, EEBLC, and EECMR. The number of dead nodes of the proposed protocol was the lowest compared to the other protocols.

In Figure 4b, we varied the number of nodes in 1000 rounds with the transmission range at 200 m. The LEACH protocol had the highest number of dead nodes compared to the DBR, EEBLC, and EECMR. Since EECMR allows the nodes to form a cluster according to their depth level, nodes only communicate with the nodes at the same depth level in the same cluster. However, nodes in the LEACH protocol became cluster heads in turns according to the generated random number. The nodes may have consumed high energy as a result of increasing the number of transmissions during cluster formation. In EEBLC, the clustering allows unequal clusters in the networks in which nodes can elect to become cluster head considering the number of neighbors and its residual energy. As a result, the number of dead nodes in EEBLC is lower than that of DBR and LEACH, as in Figure 4. In the case of DBR, instead of forming into clusters, the nodes sent data to the sink via a multi-hop routing path according to their depth level. The node would send packets to the forwarder node at the upper depth level without considering the number of neighbors sending packets to the same destination. The forwarder node may have had a high load, which causes high energy consumption while sending packets.

**Figure 4.** Number of dead nodes: (**a**) Transmission range 200 m, 450 nodes in 1000 rounds; (**b**) Transmission range 200 m, nodes 50 to 450; (**c**) Transmission range 150 m, 450 nodes in 1000 rounds; (**d**) Transmission range 150 m, nodes 50 to 450.

In Figure 4c,d, we evaluate the network performance when the transmission range was 150 m. It is clear that the network performance depended on the transmission range of the node. The number of dead nodes of the three protocols was higher when the transmission range decreased. However, the network performance of clustering protocols of LEACH, EEBLC, and EECMR was similar, which was better than DBR. Therefore, the clustering protocol showed better performance than the multi-path routing protocol.

#### 4.2.2. Residual Energy at the Nodes

In Figure 5, the residual energy of the nodes is shown in different network scenarios. At each round, the nodes generated packets and then forwarded them to the sink node. In Figure 5a, the residual energy of the node decreased according to the timeline, while the number of rounds increased. It is clear that our EECMR performed better in terms of conserving energy; the residual energy in EECMR was higher than that of DBR, LEACH, and EEBLC. DBR was developed for routing protocols in underwater wireless sensor networks, so DBR performed better than LEACH. EEBLC allows nodes to become the cluster head in turns according to their location and residual energy, and the nodes can save energy when varying the number of nodes and transmission range.

In Figure 5b, we varied the number of nodes in the network, while three protocols decreased the residual energy. Despite this, the residual energy in EECMR was higher than in EEBLC, LEACH, or DBR. However, the residual energy of the nodes in LEACH declined quickly compared to EECMR. This can be explained as follows. In a dense network deployment, the nodes consume higher energy to perform more communication tasks, such as transmitting and receiving packets to the larger number of nodes. The nodes in EECMR can act as a cluster head or cluster relay; therefore, the cluster head has the information of its cluster member, which reduces the number of transmissions between nodes. As noted in the previous section, the number of dead nodes in EECMR is less because the nodes conserve energy while sending packets. EEBLC performs better than the others in the case of a low transmission range which ensures a long network lifetime. In LEACH, the node that elects to become a cluster head broadcasts information to the threedimensional area network; as a consequence, the node consumes more energy. In DBR, the node selects the upper-level depth to become its forwarder. If the number of nodes increases, the forwarder may receive more packets, and then it sends these to the upperlevel depth before reaching the sink. More transmission at the nodes and forwarders leads to high energy consumption, as shown in Figure 5.

**Figure 5.** Average residual energy of a node: (**a**) Transmission range 200 m, 450 nodes in 1000 rounds; (**b**) Transmission range 200 m, nodes 50 to 450; (**c**) Transmission range 150 m, 450 nodes in 1000 rounds; (**d**) Transmission range 150 m, nodes 50 to 450.

In Figure 5c,d, the decrease in the transmission range led to higher energy consumption in EECMR, EEBLC, LEACH, and DBR. These protocols gradually diminished the residual energy with respect to the number of rounds. When increasing the number of rounds with a low transmission range, EERBLC performs better than EECMR. When the number of nodes increases, LEACH and EEBLC perform better than EECMR in the case

of a low number of nodes; EECMR performs better than LEACH in the case of dense deployment.

In a comparison of the residual energy performance with different transmission ranges in EECMR, the residual energy for a 200 m transmission distance was higher than a 150 m transmission range. This can be explained as follows. The cluster heads in EECMR aggregate the data of all cluster members and then forward these to the cluster relay which belongs to another cluster. Since the number of data packets is re-transmitted via multihop, the amount of energy consumption increases. In addition, the change in the network topology causes the re-established cluster to consume energy-transmitting and -receiving control packets. However, the total received packets at the sink should be considered. This is presented in the next section.

#### 4.2.3. Received Packets at the Sink Node

We assume that all the sensed data of nodes will be received at the sink. In the network deployment, the network topology changes every 100 rounds, which may cause a failure to receive packets at the cluster head and cluster relay. This is because the distance from a cluster member to the cluster head is greater than the transmission range and the received packets at the cluster head or cluster member may fail as a result of a low level of received power. Therefore, the cluster must be re-established, and a new cluster head, cluster member, and cluster relay must be selected. In Figure 6, the total received packets at the sink are evaluated in four cases. In Figure 6a,c, despite different transmission ranges, the received packets in EECMR are higher than in LEACH, EEBLC, or DBR. It is noted that the received packets at the sink increase when the transmission range increases. The number of total received packets in the EECMR scheme is occasionally different, which causes high jitters. Despite this, the increase in transmission range leads to a lower residual energy at the node, as shown in Figure 5. This can be considered a pay-off with respect to the transmission range and network performance.

In Figure 6b,d, the received packets at the sink are shown according to the number of rounds. Due to the number of packets at the sensor node or the number of clusters at each round, the results in the timeline fluctuate. In total, the received packets at the sinks in EECMR are higher than other protocols, which results in high throughput and a more reliable network.

**Figure 6.** *Cont*.

**Figure 6.** Received packets at the sink: (**a**) Transmission range 200 m, 450 nodes in 1000 rounds; (**b**) Transmission range 200 m, nodes 50 to 450; (**c**) Transmission range 150 m, 450 nodes in 1000 rounds; (**d**) Transmission range 150 m, nodes 50 to 450.

#### **5. Conclusions**

In this work, we propose the energy protocol EECMR for routing data packets in UWSNs. EECMR is a depth-based clustering protocol that uses the depth level of the node to select cluster head nodes and forwarder nodes for multi-hop routing. EECMR considers the residual energy of the node which elects cluster heads in turns. The nodes can change roles as cluster head, cluster member, and cluster relay. The cluster relay node forwards data from a deeper level to the sink. With the aid of a cluster relay, the energy consumption for transmission is decreased, leading to fewer dead nodes. The simulation results showed that EECMR achieves better performance in terms of higher residual energy, longer network lifetime, and higher received packets at the sink. Although the proposed protocol can properly select the cluster head and cluster relay according to the depth level and residual energy, the high energy consumption at the cluster relay between different depth levels may result in re-clustering or frequent re-clustering. However, the clustering protocol may cause high latency due to the multi-hop routing path. The different types of sensor nodes will have different data priorities, an issue which must be addressed in future work.

In addition, several issues remain open for future work, including optimization of the network topology and implementation of the clustering protocol in firmware to support UWSN application in oceans. In our future work, we will implement the clustering protocols into firmware to investigate the practical performance.

**Author Contributions:** The main contributions of N.-T.N. and T.T.T.L. were creation of the main ideas and performance evaluation through extensive simulations. H.-H.N. contributed to the manuscript preparation and designed the theoretical analysis. M.V. served as consultant of N.-T.N. to discuss, create, and advise the main ideas and performance evaluations together. All authors have read and agreed to the published version of the manuscript.

**Funding:** The research leading to these results received funding from the Czech Ministry of Education under grant No. SP2020/65, conducted at the Technical University of Ostrava and partially by The Ministry of Education, Youth and Sports from the Large Infrastructures for Research, Experimental Development and Innovations project "e-Infrastructure CZ–LM2018140". This work was also supported by Saigon University, Vietnam, grant No. TÐ2020-17.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Data is contained within the article.

**Conflicts of Interest:** The authors declare no conflict of interest.
