**5. Routing in DRONET Layer**

The routing in DRONET layer is composed of two stages. The first stage is the Initial and search for MANET nodes to obtain positions of all nodes and possibly all MANET networks in the area. The second stage is routing itself, where the UAV communicates with MANET gateway nodes and within other Drones in order to provide a backup network for critical data transmission.

#### *5.1. Initial and Search Stage of DRONET Layer*

The main role of the Initial and search stage is to continuously search for MANET nodes in the desired area. Therefore, we assume that the operating area is known and it can be divided into multiple sub-areas. The size of sub-areas should be picked according to the UAVs antenna coverage perimeter. Another assumption is considering a local entity that is in fact the dock with operational PC and antenna. When the disruption of fixed infrastructure in the area occurs, the dock initiates the Initial and search stage until the disruption is over.

Dock continuously sends UAV's to all areas. The UAV is looking for MANET node. We assume that all UAVs are equipped with MANET IEEE 802.11 Wi-Fi communication technology and they are capable of discovering MANET nodes in the ground. When the UAV discovers the MANET node, it obtains the Topology info from each node along with GPS positions of all nodes in the network [36]. That information is sent back to the dock, where the clustering algorithm is provided. It is possible to use multiple clustering algorithms, such as simple lowest-ID or highest-connectivity (degree) algorithm [37], or more complex algorithms such as Particle Swarm Optimization (PSO) [38]. The dock returns the result of the clustering algorithm along with the cluster heads selections. The UAV then takes a position of the MANET gateway node and notifies it about cluster head election and nodes participated in its cluster. The MANET gateway node is then responsible for notification of other MANET nodes about its election.

**Figure 11.** Example of multiple encountered cluster in the area.

When multiple clusters are discovered, UAV takes the position of free cluster and dock send another UAVs takes positions of remained clusters. The multiple clusters can occur if the MANET network contains a larger number of nodes. The situation is depicted in Figure 11, where the left upper part contains MANET network that was divided into two clusters, marked by magenta and green colour. The UAV that discovered the network covers one cluster and dock in the middle of the area sends another UAV to cover the second cluster. All steps of Initial and search stage are described by the flowchart in Figure 12.

The first step includes input in the form of information about the area. Then the area is divided into N subareas based on area size and coverage perimeter of UAV's antenna. Then, all subareas are marked by associated information *SUBAREAINFO*(*N*) as "not searched". Algorithms proceed with an endless While loop, which termination is initiated by the ending of disruption scenario. One by one, UAVs are continuously sent to all subareas that are marked either as "searched" or "not searched" and at the same time as "uncovered". Only areas marked as "covered" are not searched

again. This is because even previously searched subareas can exhibit new uncovered MANET nodes or MANET subnets.

**Figure 12.** The DRONET initial and search stage flowchart.

If the MANET nodes are not discovered in the particular subarea, this area is marked as "searched" and UAV proceed to another subarea. If the MANET node is discovered, UAV request *TopologyINFO*, which includes the topology of the network and GPS positions of all nodes in the network, where discovered node participates. TopologyINFO is sent to the dock, where clustering is performed. The results of clustering are sent back to the UAV, which informs nodes about Cluster Head and participation in the cluster.

The algorithm then proceeds according to resulted clusters. If multiple clusters appear, UAV covers the randomly selected cluster and dock send another UAVs to cover remained clusters. Then, all subareas belonging to covered clusters are marked as "searched" and "covered".

#### *5.2. Routing Stage of DRONET Layer*

Routing stage in DRONET layer begins after the Initial and search stage, where at least two UAVs are connected to each other. As a communication technology, UAVs of DRONET layer in MNM should use at least IEEE 802.11 Wi-Fi. However, in terms of possible interference, it is better to use second technology such as IEEE 802.16 WiMAX. The same as in WSN layer, the UAVs, therefore, needs to implement dual protocol stack.

The earliest version of WiMAX is based on IEEE 802.16 and is optimized for fixed and roaming access. This solution was further extended to support portability and mobility based on IEEE 802.16e, also known as Mobile WiMAX. In recent years, multiple studies provide performance comparisons of routing protocols for WiMAX. Raseed et al. [39] perform a comparison of DSDV, DSR and AODV routing protocols, where table-driven protocol DSDV has the best performance in terms of the packet delivery fraction parameter. On the contrary, a AB Rahman [40] suggests, AODV outperforms DSR and DSDV routing protocols. The performance comparison in [41] shows that in a mobile environment, ZRP and AODV perform better than DSR and OLSR. Pathak et al. [42] also consider the performance of routing protocols for sending Health-care data over the WiMAX network. The results show that from studied protocols AODV, OLSR, ZRP and the LAR1, the last mentioned can offer better results in sending telemedicine data over the wireless channel with high throughput and better reproducibility.

In MNM, the reactive protocol is suggested, since periodic updates can be energy-demanding on limited UAVs resources. The deployed routing protocol also needs to implement necessary adjustments. We assume that the UAV is able to extract information about AP availability from topology obtained from a MANET gateway node. Therefore, this information should be taken into account in the routing metrics. The routing algorithm is described by the flowchart depicted in Figure 13.

The routing algorithm begins with urgent data obtained from the MANET layer. UAV than check for available UAV that has connectivity to MANET network with accessible AP. If UAV with MANET connectivity to accessible AP is not available, the Data Backup algorithm for DRONET layer is called. This algorithm store data and set timer. If the timer is zero, Data Backup returns Fail Status. If Continue status is returned, UAV check for UAV with MANET connectivity to accessible AP again. Otherwise, data are dropped. The timer is decremented and Continue status is returned. Since the DRONET layer is not connected to AP, Data Backup algorithm presented by Algorithm 1 needs to be edited. Data Backup algorithm for DRONET layer is, therefore, described by Algorithm 2.

If UAV with MANET connectivity to accessible AP is available, the urgent data are transported to particular UAV. This UAV then tries to send data to the MANET gateway node. If this node is not available, the main algorithm calls Data Backup. If the returned status is Continue, UAV check MANET gateway node again. Otherwise, data are dropped.

**Figure 13.** DRONET routing stage algorithm flowchart.

#### **6. Simulations and Results**

Based on the theoretical analysis of the MNM concept, simulations in the Matlab environment were designed to simulate the behavior and forwarding of urgent data within the model. The aim of these simulations is to point out the fact that in case of disruption scenario of the fixed infrastructure of 5G and IoT networks, the proposed concept of MNM can take over the role of critical applications and services. Simulations point out that the interconnection of several networks into a hierarchically composed multilayer model has advantages over the use of purely wireless sensor networks in the form of transmission of critical data to functional access points with higher transmission speed and lower delay. The exact description of the simulation scenarios will be provided in following subsections.

Simulations in Matlab do not account rerouting, network reconfiguration and do not use a lot of network parameters like other simulators such as Ns-3 or OPNET Modeler. However, Matlab enables us to simplify the concept of MNM network and provide proof of concept. With Matlab, it is also possible to use a variety of different mobility models for MANET networks such as Random-Way Point or social-based mobility model described in [43]. Matlab also provides possibilities to implement computationally intensive algorithms such as PSO for clustering. In future, we planed to merge Matlab implemented algorithms with OPNET Modeler to further extend simulations with all network parameters.

#### *6.1. Simulation Scenarios*

The simulations of the MNM concept were divided into three simulation scenarios, which are intended to illustrate the advantage of using multiple layers of the MNM model compared to the deployment of only WSN networks in the area affected by the disruption of fixed 5G and IoT infrastructure.

The simulation presupposes that there is an area where the fixed infrastructure has been disrupted and there is only one functional AP. Another prerequisite is the location of the WSN wireless sensors in the area intended for data collection, while the occurrence of mobile MANET devices is also considered in the same area. In the network disruption scenario, UAVs of DRONET networks are also present, arranged in the area to cover the MANET network.

#### 6.1.1. Simulation Scenario 1

This simulation scenario is intended to highlight the benefits of adding a MANET layer to a WSN network. The scenario itself consists of a 100 m × 100 m simulation area, where 400 WSN sensors are randomly distributed. The simulation assumes the use of IEEE 802.15.4 ZigBee communication technology with the RPL-Weigth routing protocol using the 6LoWPAN in WSN layer.

Based on ZigBee technology, a radio range between WSN sensors is set to 10 m with a data rate of 30 Kbps. Data rate was randomly generated on each link between WSN sensors in the range of ±50%. This range should take into account the unforeseen effects of the environment on the data rate.

Besides the WSN layer sensors, 20 MANET nodes are also randomly placed in the area. MANET nodes uses IEEE 802.11 Wi-Fi technology with the 802.11n standard for communication and IPv6 OLSR protocol. This allows the nodes to set the radio range to 40 m at a data rate of 100 Mbps.

Data rates of MANET layer were randomly generated on each MANET link from the range of ±50%. There is also one AP in the network, which has the same radio range and data rate as MANET nodes. The individual variables of the simulation scenario can be seen in Table 1.

In this scenario, one WSN sensor is chosen as a source node that attempts to send urgent data for processing to the relevant application or service on the Internet. The role of routing protocols in all simulation scenarios mentioned in this paper used by each layer of MNM is to find the optimal routing path. All routing paths are selected according to the data rates generated on each link and number of hops in therm of Dijkstra shortest path algorithm.

To illustrate the benefits over traditional WSN network and two layers of WSN and MANET networks, a single access point in the network was strategically placed sequentially in three different positions:


The positions of AP mentioned above with devices placed in the simulation area are depicted in Figure 14a for WSN scenario and in Figure 15a for WSN-MANET scenario. Optimal routing paths for both scenarios are depicted on second part of mentioned figures marked as Figure 14b and Figure 15b respectivel.

Figure 14 illustrates the network layout with WSN nodes (blue markers), WSN sensor gateways (green markers) and AP depicted by the Wi-Fi router illustration. The WSN sensor gateways are selected by the RPL-Weight protocol in the sense of so-called "sink mobility" (Section 4.1). WSN source sensor is marked with the ID number of 393 and highlighted by a blue mark with a red edge.

Figure 15 illustrates the first layer of the WSN sensors (blue marks) with WSN sensor gateways (green marks). The MANET nodes (red marks) of second layer are distributed evenly throughout the area. There is an AP depicted by Wi-Fi router illustration and the WSN source sensor depicted by a blue mark with a red edge and ID number of 393.

These simulation scenarios were run 1000 times, always with the same position distribution of WSN and MANET nodes in an effort to illustrate the difference in performance and parameters of the two networks.

**Table 1.** Scenario 1 simulation variables


**Figure 14.** (**a**) Example of simulation scenario 1 with all positions of AP for WSN network. (**b**) Example of optimal routing path from WSNT source sensor to AP deployed in Position 2 (in the middle of the simulation area).

**Figure 15.** (**a**) Example of simulation scenario 1 with all positions of AP for WSN-MANET network. (**b**) Example of optimal routing path from WSN-MANET source sensor to AP deployed in Position 2 (in the middle of the simulation area).

#### 6.1.2. Simulation Scenario 2

The second simulation scenario extends the first scenario by adding a DRONET layer to the multilayered model, which performance will be compared to the traditional WSN network depicted in Figure 14. The distribution of WSN nodes with their parameters and technologies is the same as in the case of the first simulation scenario.The number and parameters of MANET nodes remained unchanged, but some positions of MANET nodes were partially modified in order to divide the original MANET network into two subnets. This change was made to simulate the division of the MANET network into individual subnets in order to use UAVs of the DRONET network.

The UAVs uses IEEE 802.16 WiMAX technology with the AODV protocol to communicate with other UAVs and dock. Therefore, the radio range of UAVs was set to 200 m with a data rate of 250 Mbps. The number of UAVs and their distribution in the area is determined by the number of clusters and their location in the area. The PSO clustering algorithm was used in this task, which divided the individual MANET nodes into 4 clusters or logical subnets based on positions of the MANET nodes. The output of the clustering algorithm can be seen in Figure 16.

**Figure 16.** The example of the PSO clustering output.

Based on the assumptions mentioned above, the variables of the simulation scenario 2 were set according to Table 2.


**Table 2.** Scenario 2 simulation variables

As in the first simulation scenario, three different access point positions were used in the second simulation scenario as well. The specific locations coincide with the first simulation scenario and together with the locations of the WSN, MANET and UAV nodes are depicted in Figure 17. As in the case of the first simulation scenario, the second simulation scenario was run 1.000 times for all AP positions with the same initial positions of WSN, MANET and UAV.

**Figure 17.** Example of simulation scenario 2 with the positions of AP, WSN, MANET and UAV nodes.

### 6.1.3. Simulation Scenario 3

The third simulation scenario is focused on the ability of the MNM concept to collect data by MANET nodes from the WSN network. The scenario consists of 2000 WSN sensors, which are static. Mobile MANET nodes were also placed in the simulation area, but the number of those nodes will be changed from 50 to 75 and 100. The movement of nodes was based on Random-Way Point mobility model with the speed of nodes set at 5 km/h, or 1.4 ms respectively, which is the speed of human walk. Radio ranges of MANET nodes were set to 50 m. This simulation scenario aims to observe the time that MANET nodes needs to cover all WSN nodes compared to WSN gateway sensors with its radio range. If the WSN sensor or WSN gateway sensor are within radio ranges of MANET nodes, we assume that contact is negotiated and data transfer can occur. The simulation begins with the

initial position of MANET nodes that starts to move. The simulation ends when all WSN sensors or WSN gateway sensors are covered by MANET nodes at least one time. The simulations were run 1000 times with the same initial positions of WSN sensors, WSN gateway sensors and MANET nodes. This simulation with the position of WSN sensors, WSN gateway sensors and MANET nodes is depicted in Figure 18. The simulation variables are set according to Table 3.

**Figure 18.** Example of simulation scenario 3. The brown circles are illustrations of MANET node's radio ranges.

**Table 3.** Scenario 3 simulation variables


#### 6.1.4. Results of Simulation Scenario 1

In the first simulation scenario, we consider a WSN network with wireless sensors and WSN gateways according to the "sink mobility" of the RPL-Weight protocol. In all simulations, only one AP was functional and its positions were changed according to the description in Section 6.1.1. In the case of WSN-MANET simulation, the routing protocol starts at WSN sensor 393, which is the source of urgent data. This sensor searches an optimal routing path to the WSN gateway sensor, which transfers data to the MANET node if AP is not in its radio range. The MANET network, then transfers urgent data to the functional AP by optimal routing path. Amount of urgent data was set to 100 KB. The results are examined data rates, the time required to transfer the data, and the number of hops from the source node to the access point.

The first result examines the total average delivery time of 100 Kb urgent data from the source sensor to the AP. This result is depicted by the graph in Figure 19. The graph shows the total average time required to deliver data via the traditional WSN network and via WSN-MANET network within the multilayer network model. The time required to redirect data on individual devices was not considered in these simulations.

**Figure 19.** The total average delivery time of 100 Kb urgent data from the source sensor to the AP.

In the case of the WSN-MANET network, it is possible to observe a double component of time in bar graphs. Full time is composed of the time required for urgent data transmission via the WSN network and subsequently via the MANET network. With the help of this graphical representation, it can be seen that the majority of the total time required to transmit data is formed by the time required to transmit data over the WSN network. Therefore, it is possible to say that MANET network significantly speeds up data transmission.

The overall result for all AP positions shows a significant reduction of data transfer time in the case of WSN-MANET model, with the trend being more pronounced when the distance between the WSN source sensor and AP increases. Although the total data delivery time in the WSN network increases significantly with increased distance between the WSN source sensor and AP, in the WSN-MANET network this trend increases only very slowly. This is due to the fact that the delivery time of the WSN component is almost the same for each AP position since the WSN sensor gate was available for one jump in most cases and also due to MANET data rate, where transmission of 100 Kb urgent data is fast.

The second result depicted by the graph in Figure 20 expresses the average number of hops from the source WSN sensor to the AP. As with the first result, hops in the WSN-MANET network graph bar are represented as individual components of WSN and MANET networks.

**Figure 20.** The average number of hops from source WSN sensor to AP.

In contrast to the first result, increasing transmission time of WSN-MANET network through all AP positions shows insignificant trend, while the increasing trend of hops in the case of the WSN-MANET network in the case of the second result is more pronounced. However, the number of hops in the WSN-MANET network is lower in each simulation. For position 1, the number of hops decreases approximately by 42%, while for the other positions the interconnection of WSN and MANET networks reduces the number of hops by more than 50%. This reduction increases with the highest distance between the source WSN sensor and AP. The reason is the fact that in addition the highest transmission speeds, the nodes of the MANET network provide also a higher radio range.

The third result (Table 4) shows the average values of the data rate that was achieved on the individual optimal paths in the case of 100 Kb urgent data delivery. Those values were achieved by averaging the data rates achieved on the individual parts of the optimal paths.


**Table 4.** The average data rates in 100 Kb urgent data delivery from source WSN sensor to AP.

Table 4 shows that the data rates in the case of urgent data transmission via the WSN network are many times lower than in the case of urgent data transmission via the MANET network. In the WSN network, the average data rate ranges from 26 to 28 Kbps, while in the case of the WSN-MANET network data rate ranges from 45 to 82 Mbps. It is also possible to observe an increasing trend of the average data rate in the case of the WSN-MANET network. This phenomenon is caused by the fact that transmission of the data with increasing distance between the source WSN node and AP takes place in an increasing part of the MANET network. This fact is supported by the second result in Figure 20, where the number of hops in MANET networks is highest in the case of AP Position 3. A higher number of hops in the MANET part of WSN-MANET network resulted in a higher average data rate.

#### 6.1.5. Results of Simulation Scenario 2

The second simulation scenario adds UAV devices of DRONET layer to the WSN-MANET network, which complement the concept of MNM. This concept will be compared with the WSN network the same way as in the case of the first simulation scenario. In the case of the WSN-MANET-DRONET network, routing protocols use the same algorithm to find optimal paths like in scenario 1. To save energy in the DRONET layer, routing protocols deliver urgent data primarily using the MANET layer. Despite UAV accessibility, urgent data are transferred to DRONET layer only if the AP is not presented in a particular MANET subnet. The example of this scenario with the optimal routing path is depicted in Figure 21.

The routing starts at the WSN layer, where the source WSN sensor finds the most optimal path in terms of the number of hops and data rate to the nearest WSN gateway sensor, which delivers the urgent data to the nearest MANET node. If functional AP is presented in the MANET subnet, the urgent data are transferred to this point. If the AP is not presented in the MANET subnet, the data are routed to the nearest available MANET gateway. Those MANET gateways are known because of the clustering algorithm performed by DRONET layer. After transferring the data to the DRONET network, the corresponding UAV selects the routing path according to OSLR routing protocol to the UAV, which is connected to the MANET subnet with functional AP. After the urgent data are transferred back to the new MANET subnet, the routing algorithm looks for the most optimal path to the AP.

**Figure 21.** The example of optimal routing path in WSN-MANET-DRONET network. (**a**) Expample of optimal routing path through WSN and MANET layer. (**b**) Example of optimal routing path through WSN, MANET and DRONET layer.

The first result shows the total average delivery time of 100 Kb of urgent data from the source WSN sensor to the AP. However, in addition to WSN-MANET networks, these simulations also include DRONET networks. Therefore, it is possible to see a total of 3 delivery time components within AP Position 3 in the bar graph of WSN-MANET-DRONET network (Figure 22).

**Figure 22.** The total average delivery time of 100 Kb urgent data from the source sensor to the AP.

In the case of Position 1 and 2, the DRONET network was not used due to the direct connectivity of the MANET network with the AP. The DRONET layer was used in simulation with the AP on Position 3 since AP is in the separate MANET subnet. Even the farthest AP was reached quickly in terms of delivery time, despite the use of three layers. WSN-MANET-DRONET network reduces the delivery time of urgent data on Position 1 by almost 70% compared to WSN layer. In the case of Position 2, this difference has already increased to about 79% and in the third position by 90%. The time component of the DRONET network is responsible for the significant decreasing of the delivery time at Position 3 due to the highest data rate on the routing path. At this data rate, the time component of transferring 100 Kb of urgent data is very small. Despite the higher number of hops and distance

between the source WSN sensor and AP, the DRONET layer was able to reduce the time of delivery that is comparable to the results on Position 1 and 2.

The second result depicted in Figure 23 shows the average number of hops between the source WSN node and AP. This result complements the informative value of the first result, as the achieved delivery times depend on the number of hops. Based on hops, it is possible to see why the time contribution of the MANET network in Position 3 was the highest.

**Figure 23.** The average number of hops from source WSN sensor to AP.

The reason is the fact that most of the data transmission took place via the MANET network. On average, almost 6 hops within the MANET network in Position 3 compared to 1 to 3 hops within the MANET network in Position 1 and 2. In the overall comparison with the WSN network, the number of hops was reduced by 40 to 45% in the case of WSN-MANET-DRONET network. The third result presented in a tabular illustration (Table 5) shows the average data rates achieved on optimal routing paths when transmitting 100 Kb of urgent data. In the case of the WSN network, data rate ranges from approximately 27–28 Kbps, while interconnection of the WSN-MANET-DRONET network resulting in an average data rate ranges from 44 to 87 Mbps. An important factor of this increasing trend are the higher data rates of MANET and DRONET networks. This shows a great advantage over the deployment of the classic WSN network in the case of fixed infrastructure disruption.


**Table 5.** The average data rates in 100 Kb urgent data delivery from source WSN sensor to AP.

6.1.6. Results of Simulation Scenario 3

The third simulation scenario deals with data collection of MNM concept. The MANET layer, which is hierarchically located above the WSN layer, is responsible for data collection in case of disruption of the fixed infrastructure. When disruption of the fixed infrastructure occurs, the WSN sensors transfer urgent data to the nearest WSN gateway sensor, which tries to locate MANET nodes if AP is unavailable. If the MANET node is not close to the WSN gateway sensor, the gateway stores the urgent data in its cache and waits for contact with MANET node. Therefore, it is crucial to compare the times needed to collect data by MANET nodes from WSN nodes and WSN gateway sensors. Those results are depicted in Figure 24.

**Figure 24.** The average time needed to cover all WSN sensors and WSN gateway sensors.

Since the number of WSN gateway sensors is lower than the number of WSN sensors, results show that time needed to cover the last WSN sensor by MANET nodes is higher compared to the time needed to cover the last WSN gateway sensor. It can be also observed that with an increasing number of nodes, the time required to cover all nodes is lower. With twice the number of nodes, the time is even more than twice as low. This is due to the fact that the highest percentage of WSN nodes and gateways are covered with the same radio range of MANET nodes during the initial node distribution. More importantly, the time required to collect data from each WSN gateway sensor is approximately 30% lower than the time required to collect data from each WSN sensor.

A further reduction in the number of sensor gates would contribute to even less time, but a low number of gates would also increase energy costs and traffic at these nodes, as a "sink mobility" model is considered, where all communication converges to a single point. In the end, this result proves that the system of the gateway in the WSN network is useful for data collection in terms of time.

### **7. Brief Discussion about Future Steps of Proposed MNM Model**

In these sections we will discuss the our future steps of the research.

#### *7.1. Energy Consumption of Proposed MNM Model*

A lot of the research activity today is focused on research into energy consumption in the routing process for either wireless networks or multilayered networks such as [44–46]. The energy is also considered in terms of UAV management [47,48]. In our paper, energy is also considered in different areas. For example, in a DRONET network, the presence of a central point (dock) is considered, which is responsible for UAV management, charging of drained UAVs, and performing energy-intensive operations, such as clustering. Energy is also considered in interlayer communication (Section 3.5), where using of light protocols such as UDP, CoAP or EXI is advised.

In the WSN network, the sink routing model is considered, so it is possible to assume, those nodes near the sink node connected to the MANET layer will be asked to forward packets more frequently. To address this problem and also lower the traffic load and energy consumption, the WSN sink node will forward only urgent data. In the DRONET layer the reactive protocol is suggested since periodic updates can be energy-demanding on limited UAVs resources. Also, urgent data are transferred to the DRONET layer only if the AP is not presented in a particular MANET subnet.

Although energy consumption in MNM is considered, it is nevertheless not evaluated in this paper. Energy consumption in MNM is a however important issue and its evaluation is part of future research, which needs to address multiple areas from network design to management of nodes and routing protocols.

#### *7.2. Security Aspects of Proposed MNM Model*

Security is very important and broadly discussed term in 5G networks. The goals of security solutions are to provide privacy, authentication, integrity, non-repudiation, and confidentiality [49–51]. Base on heterogeneity of the Internet of Thing-based systems, the proposed systems will need support different solutions and algorithms in the sense of security, privacy, secure transmission of information over the networks, interoperability and data management [52]. Based on the characteristics of networks, we have to take into account different items from the security solutions point of view. In MNM model, the term security gains importance, because this model integrates different types of wireless networks with specific technical challenges to attack vulnerabilities.

In MNM, public safety is a very important part of the security solution. Our solution enables to increase public safety by possibilities to transport emergency information between users without any infrastructure in a disaster and crisis. We must note that our solution does not record any sensitive information about users, we only deal with technology point of view.

A second look for security is a network and information security. Each network (MANET, DRONET, WSN, and Sensor) of the MNM model have varied security challenges. In the field of security, we will solve the problem of secure and robust transmission between different a source and destination nodes. Secure routing in sensor networks is a very hard problem due to inherent properties in comparison with MANET, DRONET and different types of wireless networks. In the field of security, our research activities in MNM model will be focused on:


Due to the nature of the proposed solution of the MNM, we will implement the game theory, namely cooperative, non-cooperative and evolution games to the process of finding reliable and secure communication paths between mobile terminals to transport IoT data. The game theory also gives us the possibilities to select reliable communication paths between terminals regarding the actual situation in the networks, and we will eliminate the different malicious nodes located in the networks.

Another solution is implementation trust-based routing algorithms to transport of the IoT data between isolated islands of the terminals with limited connectivity. We are working on the blockchain routing algorithm to provide a secure path selection between the source and destination nodes. The main idea of all security solutions is to provide secure communication between mobile terminals to provide robust, reliable and secure communication between mobile terminals to transmission IoT data. Another idea of the MNM is to provide a secure public safety network to communication between different terminals without any infrastructure.

#### **8. Conclusions**

In this paper, the new multilayered network model (MNM) for the disrupted infrastructure of the 5G mobile network was introduced. The main goal of this paper is to present possibilities and ideas, which describes how multilayered network models can be built and which technologies and routing protocols is possible to implement.

Therefore, the MNM concept is composed of three independent layers of networks, which are capable of collaboration if disruption of fixed infrastructure occurs. The whole model is able to perform data collection at WSN layer using sensors, which mimic the IoT behaviour by sending those data to the Cloud. If disruption scenario occurs, only urgent data are allowed to pass into higher layers through the introduced system of WSN gateways. The disrupted part of the network can be bypassed with MANET nodes of MANET layer, which offers longer radio ranges and higher data rates and thus faster delivery. If MANET subnetworks are unable to deliver urgent data, it is possible to use backbone DRONET network, which offers even longer radio ranges and higher data rates. The UAVs of DRONET are able to discover MANET nodes and perform clustering mechanism to effectively cover MANET subnetworks with UAVs. Along with MNM, recommendations for use of possible wireless technologies with routing protocols were provided. In addition to these recommendations, the exception mechanism for urgent data delivery in routing algorithms was introduced to all layers.

In order to show that the concept is capable of providing the intended functionalities and also highlighting the differences between typical WSN network and MNM, simplified Matlab simulations were provided. More complex simulations in simulators, such as OPNET Modeler or Ns-3, are in the process of preparation and will be included in future papers.

The MNM model provides new possibilities to use wireless networks without any infrastructure as a public safety network. This model should be used not only during an emergency and crises to transport IoT data between different terminals and sensors. MNM model enables to increase the mobility of the mobile terminals, design new services and applications as well.

Future research also includes implementation of IPv6 routing to Adaptive Routing protocol for CR-MANET (AR-CRM) in order to provide methods for spectrum sensing and intelligent method for channel management, which can result in lower interference between MANET nodes. Another step is to provide research of Fuzzy logic inside AR-CRM and spectrum sensing methods in order to manage MANET channels according ZigBee channels in WSN network. A detailed study of critical areas such as access control, both network and information security as well as evaluation of energy consumption in each layer by presenting a multilayered network model will be the subject of future studies and publications. The proposed MNM concept needs improvements, especially in the security area, which is mandatory for future networks. Improvements are also needed in terms of UAV management in DRONET network, where the complex algorithm needs to be established based on energy consumption constrains, security and privacy.

**Author Contributions:** Conceptualization, D.H. and J.P.; methodology, L.D.; data analysis, L.O.; validation, D.H, J.P. and L.D.; formal analysis, L.O.; investigation, D.H.; resources, J.P.; data curation, L.O.; writing—original draft preparation, D.H and J.P.; writing—review and editing, L.D.; visualization, J.P.; project administration, J.P.; funding acquisition, J.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** This work has been performed in the framework of the ministry of education of Slovak republic under research VEGA 1/0492/18 and APVV-17-0208.

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

### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
