Keywords

1 Introduction

In MANET nodes are self-organizing, self-describing, independent, and adaptive. Due to the lack of a base station, nodes in MANET communicate with each other by broadcasting messages to their neighbors. The MANET’s distinct properties, such as changeable node architecture, a lack of central coordination, a lack of accurate state information, node mobility, and insufficient resources such as bandwidth and battery power, make it an important area for research [7, 20,21,22]. The performance level of service offered by a network to a user during communication is referred to as QoS. In MANET, improved QoS is required to achieve increased network performance [9, 22]. We investigated on the impact of QoS in MANETs using a novel rectangular-3D position allocator on standard routing protocols like AODV and DSDV, as well as improved mobility models like modified Gauss–Markov (MGM), enhanced modified Gauss–Markov (EMGM), and random direction-3D (RD-3D). The mobility of nodes has a significant impact on network performance [23]. The performance of the network can be affected by abrupt changes in the topology of nodes. The EMGM mobility model outperforms the other mobility models in terms of PDR and delay, while the RD-3D mobility model outperforms the other mobility models in terms of throughput for both AODV and DSDV routing protocols.

The following is how the paper is laid out: Sect. 2 provides a high-level review of the mobility models, Sect. 3 examines conventional routing protocols, and Sect. 4 depicts simulation and performance analysis. The paper comes to a close with Sect. 5.

2 Mobility Models

The mobility model aids in the comprehension of node movement. To mimic more realistically, it is critical to select the proper mobility model for the environment. This paper presents a quick overview of several mobility models and their effects on routing protocols such as AODV and DSDV. It also shows how it affects MANET QoS. This study provides a clear picture of how to choose between mobility models and their settings. We have shown the RD-3D, MGM, and EMGM mobility models, as well as the unique rectangular-3D position allocator.

2.1 Rectangular-3D Position Allocator

This position allocator orients nodes in a three-dimensional plane. Initially, nodes will be placed in both the X-axis and Y-axis like grid position allocator. After the grid is full, then the value at Z will increase, and the next node will be placed in the next grid on Z-axis. With an increasing number of nodes, the length of the Z-axis increases. The node positions are calculated using the mathematical formula:

$$ X = x_{0} + {\Delta }x\,*\,\left( {m\% n_{{\text{x}}} } \right) $$
(1)
$$ Y = y_{0} + {\Delta }y\,*\,\left( {\left( {m/n_{{\text{x}}} } \right)\% n_{{\text{y}}} } \right) $$
(2)
$$ Z = z_{0} + {\Delta }z\,*\,\left( {m/\left( {n_{{{\text{x}} }} *n_{{\text{y}}} } \right)} \right) $$
(3)

where \(x_{0}\), \(y_{0}\), \(z_{0}\) represent the minimum values of X-, Y-, and Z-axes, m represents the current node, \(n_{{\text{x}}}\), \(n_{{\text{y}}}\) represent the number of nodes at X-axis and Y-axis, Δx represents the spaces between the two successive nodes at X-axis, Δy represents the spaces between the two successive nodes at Y-axis and Δz represents the spaces between the two successive nodes at Z-axis. This position allocator allocates the nodes uniformly in the simulation area. This is one of the main advantages of this allocator for static nodes. When stable nodes are uniformly distributed in a simulation area, then the performance of the network increases automatically [11].

2.2 Gauss–Markov Mobility Model

This mobility model is proposed by Liang and Haas [4]. “At a constant interval of time ‘t,’ the values of nodes with respect to speed, direction, and pitch are calculated based on the former value of pitch, direction, and speed at (t−1)th time interval. The speed, direction, and pitch values are calculated by the following equations:

$$ S_{t} = \alpha S_{t - 1} + \left( {{1} - \alpha } \right)\tilde{S} + \sqrt {\left( {1 - \alpha^{2} } \right)} \,{\text{S}}x_{t - 1} $$
(4)
$$ D_{t} = \alpha D_{t - 1} + \left( {{1} - \alpha } \right)\tilde{D} + \sqrt {\left( {1 - \alpha^{2} } \right)} Dx_{t - 1} $$
(5)
$$ P_{t} = \alpha P_{t - 1} + \left( {{1} - \alpha } \right)\tilde{P} + \sqrt {\left( {1 - \alpha^{2} } \right)} Px_{t - 1} $$
(6)

where \(S_{t}\), \(D_{t}\), and \(P_{t}\) are the new speed, direction, and pitch at time interval t, \( \tilde{S}\), \(\widetilde{D }\), and \(\tilde{P} \) are the mean speed, mean direction, and mean pitch, \(Sx_{t - 1} ,\) \(Dx_{t - 1}\), and \(Px_{t - 1}\) are random variables and α is a random variable whose value lies within the range of 0 < α < 1” [3–5, 16, 17].

2.3 Modified Gauss–Markov Mobility Model

“The original GM mobility model has been changed. The speed, pitch, and direction of each node are calculated in this model at a predetermined distance ‘d’ based on the prior value of pitch, direction, and speed at (d−1)th distance. The following mathematical formulas are used to calculate the speed, direction, and pitch values:

$$ S_{d} = \alpha S_{d - 1} + \left( {{1} - \alpha } \right)\tilde{S} + \sqrt {\left( {1 - \alpha^{2} } \right)} \,{\text{S}}x_{d - 1} $$
(7)
$$ D_{d} = \alpha D_{d - 1} + \left( {{1} - \alpha } \right)\tilde{D} + \sqrt {\left( {1 - \alpha^{2} } \right)} \,Dx_{d - 1} $$
(8)
$$ P_{d} = \alpha P_{d - 1} + \left( {{1} - \alpha } \right)\tilde{P} + \sqrt {\left( {1 - \alpha^{2} } \right)} Px_{d - 1} $$
(9)

where \(S_{d}\), \(D_{d}\), and \(P_{d}\) are the new speed, direction, and pitch at distance interval d, \(\tilde{S}\), \(\tilde{D}\), and \(\tilde{P}\) are the mean speed, mean direction, and mean pitch, \(Sx_{d - 1} ,\) \({\text{D}}x_{d - 1}\), and \(Px_{d - 1}\), are random variables and α is a random variable whose value lies within the range of 0 < α < 1. Randomness is determined by altering the value of α” [12].

2.4 Enhanced Modified Gauss–Markov Mobility Model

“It is enhanced version of the MGM mobility model. In this model, at a fixed distance ‘d,’ the speed, pitch, and direction of each node is estimated based on the previous value of pitch, direction, and speed at (d−1)th distance. After traveling for a fixed distance, here the node pauses for a fixed time called pause time. The pause time is chosen from a predefined range [Tmin, Tmax]. The speed, direction, and pitch value are calculated by the mathematical formulas as shown in Eqs. (7), (8), and (9)” [13].

2.5 Random Direction-3D Mobility Model

“This mobility model is a 3D version of the existing random direction-2D mobility model where the node moves based on the random direction d. At a specific period, time ‘t,’ the speed v(t) of a node is determined as a Gaussian distribution from the interval [minimum_speed, maximum_speed], and an angular direction d is chosen from the interval [0, 2π]. Considering the node speed, pause time, and angular direction, each node moves toward a definite direction until it arrives at the boundary of the model. When it arrives at the boundary, it pauses, selects a new direction and speed, and starts moving toward a new direction” [14].

3 Routing Protocols

We have looked at two different forms of routing protocols, one proactive (DSDV) and the other reactive (AODV). Reactive routing protocols do not have route information, but proactive routing protocols keep a routing table with the route to the destination [8,9,10].

3.1 DSDV

This protocol is a modification of the traditional Bellman–Ford routing algorithm [1]. In this protocol, a routing table is exchanged among neighbor nodes to keep track of up-to-date information about the network’s topology. Due to the usage of a sequence number, this approach primarily resolves the count to infinity problem [21].

3.2 AODV

“In this protocol, a path is discovered whenever a node wants to exchange information with another node. It provides loop-free paths and is scalable for massive networks” [2, 10, 15, 18, 19]. “When a path is unavailable, then the path discovery procedure is initiated, and the source node broadcasts RREQ packets to all the neighbor nodes. When neighbor nodes receive RREQ packets, they either transmit a RREP packet or broadcast the RREQ packet to the rest of the network” [6]. “When a node detects a path failure, it sends a RERR packet to its related neighbor nodes, informing all of the related nodes of the broken path” [2, 6, 19].

4 Performance Evaluation

4.1 Simulation Environment

Simulation is done in NS-3 [15]. The parameters and their values for these mobility models have been given in Tables 1, 2, 3, and 4, respectively.

Table 1 Simulation environment
Full size table
Table 2 Parameters for MGM mobility model
Full size table
Table 3 Parameters for EMGM mobility model
Full size table
Table 4 Parameters for RD-3D mobility model
Full size table

4.2 Experimental Results

Huge experiments on AODV and DSDV with 25, 50, 75, and 100 sets of nodes with various improved mobility models such as RD-3D, MGM, and EMGM employing Rectangular-3D position allocator are done to examine the influence of QoS in MANET.

4.2.1 Throughput

Tables 5 and 6 demonstrate the results of AODV throughput with various sets of nodes and different mobility models. Table 5 shows that the maximum throughput for AODV is 26.1673 kbps for 75 nodes using RD-3D, while the lowest is 0.2323 kbps for 25 nodes using EMGM. For all set of nodes considered, RD-3D has a higher throughput than EMGM, whereas EMGM has a lower throughput. In comparison with the EMGM and MGM mobility models, the RD-3D mobility model has a higher average throughput for AODV. Table 6 shows that the highest throughput for DSDV is 816.5877 kbps for 100 nodes when using RD-3D, while the lowest is 0.1918 kbps for 25 nodes when using EMGM. For all the nodes, RD-3D provides higher DSDV throughput, whereas MGM provides lesser throughput. In comparison with the EMGM and MGM mobility models, the RD-3D mobility model has a higher average throughput for DSDV.

Table 5 Throughput for AODV
Full size table
Table 6 Throughput for DSDV
Full size table

4.2.2 Delay

Tables 7 and 8 demonstrate the delay values for AODV and DSDV, respectively, utilizing different combinations of nodes and our suggested mobility models. Table 7 shows that the largest delay value for AODV obtained for 25 nodes is 1231.6941 s, while the shortest is 320.4229 s using RD-3D for 75 nodes. Between the three mobility models, RD-3D yields lower delay values for 75 and 100 sets of nodes, MGM delivers lower delay values for 50 sets of nodes, and EMGM delivers lower delay values for 25 sets of nodes. Furthermore, using the EMGM mobility model, the average delay for AODV is smaller, whereas using the RD-3D mobility model, the average delay is larger. Table 8 shows that the largest delay obtained for AODV using RD-3D mobility models for 25 nodes is 11944.3999 s, whereas the shortest delay obtained using EMGM for 25 nodes is 2.1999s. The average delay for DSDV is shown to be larger when using the RD-3D mobility model and lower when utilizing the EMGM mobility model.

Table 7 Delay for AODV
Full size table
Table 8 Delay for DSDV
Full size table

4.2.3 Packet Delivery Ratio

As shown in Tables 9 and 10, one of the performance indicators used to examine the influence of various mobility models using different sets of nodes on AODV and DSDV is PDR. Table 9 shows that the maximum PDR for AODV obtained using the EMGM mobility model for 50 nodes is 0.9705, whereas the lowest PDR obtained using the EMGM mobility model for 25 nodes is 0.2467. It is worth noting that EMGM delivers greater PDR values for 50, 75, and 100 nodes, whereas MGM delivers greater PDR values for 25 nodes. Furthermore, the EMGM mobility model is found to be superior to the MGM and RD-3D mobility models in terms of PDR for AODV. Table 10 shows that the greatest PDR for DSDV achieved using the EMGM mobility model for 75 nodes is 0.4592, while the lowest is 0.0245 using EMGM for 25 nodes and RD-3D for 100 nodes. It is worth noting that EMGM delivers greater PDR values for 50 and 75 nodes, whereas MGM delivers higher PDR values for 25 and 100 nodes. Furthermore, the EMGM mobility model is found to be superior to the MGM and RD-3D mobility models in terms of PDR for DSDV.

Table 9 PDR for AODV
Full size table
Table 10 PDR for DSDV
Full size table

5 Conclusion and Future Works

We conducted extensive tests on AODV and DSDV with a variety of nodes, QoS parameters, and mobility models to check their impact on QoS in MANET. In comparison with MGM and EMGM mobility models, the throughput for both AODV and DSDV is higher when employing the RD-3D mobility model, as a result, QoS support is improved. It is also analyzed that delay is better for both the routing protocols using EMGM mobility models, and it is worst using RD-3D mobility models. So, it is concluded that EMGM outperforms well to the parameter delay compared to other mobility models. It is also noticed that EMGM gives higher PDR values than the other mobility models for both AODV and DSDV routing protocols. Therefore, it is concluded that EMGM gives higher QoS support with respect to PDR and delay in comparison to MGM and RD-3D mobility models, whereas RD-3D mobility models give better QoS support to throughput. It is concluded that different mobility models give different results and have huge impact on QoS support in MANET. One of the future directions of this paper is to decrease the delay values and increase the PDR of the RD-3D mobility model by enhancing the model with better border solutions when the node reaches the boundary. Another way is to improve the MGM and EMGM mobility models so that the throughput can be increased. This paper is helpful to the researchers who are working on MANET. This research will help them to identify a suitable mobility model for enhancing their research work.