Obstacle and mobility aware optimal routing for manet

Abstract

In Mobile Adhoc Network (MANET), obstacles in terrain and nodes mobility are main constrains due which performance degrades. To mitigate these effects, the routing protocol should be mobility and obstacles aware. In this work, we proposed a novel Obstacle and Mobility Aware Routing (OMAR) protocol. In OMAR, for obstacles avoidance, DeCasteljau Algorithm based on Bezier curve is been used. And to reduce effects of mobility and energy shortage of a node, an Energy based Mobility Index (EMI) routing scheme is been developed. The path possesses high EMI is selected as route. The performance of proposed Mobility and Obstacle Aware Routing protocol is evaluated using NS2 simulator. Simulation results show that the proposed algorithm reduces energy consumption, overhead, delay and increases data delivery in the network.

Keywords

Obstacle awareness mobility awareness link duration Bezier curve obstacle

1. Introduction

One of the most important issue in the current MANET technology is the mobility of the nodes which reduces the performance [1]. To enhance the performance of MANET further, a number of routing protocols are proposed [2 –4].However mobility issue are not addressed effectively. Again authors have assumed plain terrain for simulation and have not considered terrain with obstacles. Hence, in this work, we have proposed a Mobility and Obstacle Aware Routing algorithm that reduces effects of mobility and incorporate the presence of irregular shape obstacles in terrain.

In this paper significant contribution are listed below.

To avoid obstacles, DeCasteljau Algorithm based on Bezier curve is used. In this work, we have used the ‘trace of Bezier curve’ as a reference to find the Guiding Node_list(This list contains the nodes close to the Bezier trace),which will guide data to destination avoiding fixed immobile obstacles.

An Energy based Mobility Index (EMI) routing scheme is developed to reduce the effects of mobility and energy shortage in node. Nodes with high EMI are selected for constructing path to avoid obstacles. The algorithm increases stability of the path.

2. Related works

There are a number of mobility models and routing protocols available in the literature for efficient routing in MANETs. The section summaries the related works already proposed by different authors on Mobility models in plain terrain and obstacle terrain.

In [5], authors have proposed an Environment-Aware Mobility model. The node connectivity is broken if any obstacles are found in the network. In this work, Hotspot model and Route model are used to observe the node mobility in a specific environment. It nice combination of available mobility model with terrain environment. However, here no mobility aware routing is proposed to reduce mobility effect.

The paper [6] initially proposes a unique Semi-Markov smooth mobility model to copy the behavior of the airliners within the sky. Probability density function of the relative speed of two nodes is obtained depending on the mobility model. Here motion consist of three moving phase i.e Speed Up phase, Middle Smooth phase and Slow Down phase. Then the expectation of the link lifetime is obtained. However, the routing protocol is developed neglecting the presence of obstacles on the route.

In [7] authors have addressed Link availability-Based Routing Protocol (LBRP) that considers dynamic topology and repeated failure link. In this scheme, the nodes are moved based on random mobility Model and it accurately guesses the availability of link throughout within a limited period of time. This protocol takes into account regular topology change and related link failure during design. However, the protocol has not incorporated the presence of obstacles on the route.

DSR Protocol was proposed in paper [8] and as per the author DSR protocol search a route when a source node wants to send packet to a destination. There are mainly two activities present in DSR i.e. route discovery phase and route maintenance phase. In DSR source routing technique is used. Here the complete route address is stored in the route cache of a node. And this complete address guides data to the destination. Route error occurs when route breaks and that initiate the route maintenance process. So that expired routes are removed from route cache and a new route discovery is started.

The paper [9] addresses a Mobility prediction method used for improving the network stability and link connection interruptions. In this approach, genetic algorithmic rule is employed to form a history primarily based mobility prediction. However, the method has not incorporated awareness of obstacles.

In [10] authors have proposed a adaptive learning an automation mobility prediction method that predicts the future mobility based on history of node mobility in plain terrain. However, authors have not addressed any obstacle aware scheme. The paper [11] proposes Residual Link Lifetime (RLL) protocol which represents the dependence between neighboring links and the joint probability distribution of two adjacent links. However but it has not considered the presence obstacle in real terrain. In [12] authors have discussed mobility prediction methodology that estimates the soundness of paths therefore helps to boost routing by reducing the overhead and therefore the range of connection interruptions. However, the paper has not considered the presence of obstacles in terrain. The paper [13] proposes a multipath routing scheme which adapts variation of dynamic network situations that improves the network efficiency. It also does not consider obstacles in terrain.

In [14], authors have proposed a dynamic and autonomic detection of obstacles technique applying the enhanced cartography optimized link state routing. This method accurately specifies the obstacle space with high coverage and efficient exactness ratios. It renders an adequate precision ratio sufficient to efficiently avoid link burst links induced by the blocking obstacle. However the technique does not include any mobility aware method.

Authors in [15] have discussed a Link-state QoS routing protocol to determine stable paths among all nodes in a MANET. The node stability function is used for selecting the route from source to destination based on node mobility degree. However, it lacks the obstacle awareness of path in the network.

The paper [16] focuses the significance of the physical layer and its effect on performance in MANET. This was proved by simulating numerous MANET environment using Network Simulator-2 (NS-2) with increased capability by adding propagation loss models like modified Two-Ray Ground model, ITU Line of Sight Non line of Sight model combined path loss and shadowing model.However, the authors have not proposed any obstacle avoidance method. In [17] authors have proposed an obstacle-avoiding Connectivity Restoration strategy which utilizes mobility technique for avoiding any curved obstacle. However, authors have not addressed any mobility scheme. The paper [18], addresses a practical Mobility model with different shape obstacle limitation based on Bezier curves (RMBC) that characterizes realistic motion and computes smooth route among the obstacles. However, the authors have not provided any mobility prediction in the network. The path planning techniques [19] using Ant Colony Optimization (ACO) method to discover optimal path in the network environment. In this scheme, Rational Bezier curves helps to detect the obstacles. However, it lacks the mobility awareness.

In [20], authors have given a sound and reliable route choice technique supported the nodes’ mobility in mobile ad hoc networks. In this approach, a path is selected for information routing depending on a node’s speed, direction and pause time. However, authors in this work have considered plain terrain and moreover they have not considered residual energy available in any node.

3. Obstacle and mobility awareness routing protocol in MANET (OMAR)

3.1. Obstacle aware routing

In the proposed work MOAR, for routing in presence of obstacles, DeCasteljau Algorithm is used based on Bezier curve concept [18]. In this scheme, a Realistic Mobility model based on DeCasteljau Algorithm is used that incorporates node movement with collision-free and obstacles.

To avoid the obstacles we create a Guiding Node_list including nodes which are close to Bezier curve path trace. Bezier curve path passes through several nodes that constitutes a complete route to destination avoiding obstacles.

3.2. Energy based mobility aware optimal routing

In Energy based Mobility aware optimal routing we have used the concept of mobility prediction and the residual energy of a node before selecting for a final path. First an Energy based mobility_index(EMI) scheme is developed. An EMI index point is calculated for each node in the Guiding Node_list. Development of Energy based mobility_index (EMI) scheme is discussed asbelow.

Here a parameter called Energy based mobility_index(EMI) for each mobile nodes in ‘Guiding node list’ is proposed. The Energy based mobility_index is the determined depending on points assigned to pause time, direction, speed, and residual energy in specified range. The Table 1 shows various ranges of pause time, speed, direction and residual energy and their corresponding index points. It is calculated relating to its neighbor nodes available in the Guiding node_list. A Guiding list node having high EMI_index indicates it has high pause time, low velocity, suitable direction and high residual energy. A minimum threshold value of Energy based mobility_index i.e. EMI_TH is fixed for Guiding node_list. If EMI of a node is bigger than or up to the threshold value EMI_TH, then it’ll be chosen as an intermediate node the route to bypass the obstacles. Otherwise, the node will not be chosen for route formation.

Table 1
Lookup index table for assigning points

Index vaue for velocity Index value for direction Index value for pause time Index value for residual energy

Velocity in m/s Index value Direction θ In Degree Index value Pause time In second Index value Residual energy in joule Index value

V = V_min 1 30 ≤ θ < 45 and 315 < θ < 330 1 T_p = T_max 1 E = E_T 1

V_min < V0 . 1V_max 0.9 45 ≤ θ < 60 and 300 < θ < 315 0.9 0 . 9T_max < T_p < T_max 0.9 E_T < E < 0 . 9E_T 0.9

0 . 1V_min < V < 0 . 2V_max 0.8 60 ≤ θ < 75 and 285 < θ < 300 0.8 0 . 8T_max < _p < 0 . 9T_max 0.8 0 . 9E_T < E < 0 . 8E_T 0.8

0 . 2V_min < V < 0 . 3V_max 0.7 75 ≤ θ < 90 and 270 < θ < 285 0.7 0 . 7T_max < T_p < . 8T_max 0.7 0 . 8E_T < E < 0 . 7E_T 0.7

0 . 3V_min < V < 0 . 4V_max 0.6 90 ≤ θ < 105 and 255 < θ < 270 0.6 0 . 6T_max < _p < 0 . 7T_max 0.6 0 . 7E_T < E < 0 . 6E_T 0.6

0 . 4V_min < V < 0 . 6V_max 0.5 105 ≤ θ < 120 and 240 < θ < 255 0.5 0 . 4T_max < T_p < . 6T_max 0.5 0 . 6E_T < E < 0 . 4E_T 0.5

0 . 6V_min < V < 0 . 7V_max 0.4 120 ≤ θ < 135 and 225 < θ < 240 0.4 0 . 3T_max < _p < 0 . 4T_max 0.4 0 . 4E_T < E < 0 . 3E_T 0.4

0 . 7V_min < V < 0 . 8V_max 0.3 135 ≤ θ < 150 and 210 < θ < 225 0.3 0 . 2T_max < _p < 0 . 3T_max 0.3 0 . 3E_T < E < 0 . 2E_T 0.3

0 . 8V_min < V < 0 . 9V_max 0.2 150 ≤ θ < 165 and 195 < θ < 210 0.2 0 . 1T_max < _p < 0 . 2T_max 0.2 0 . 2E_T < E < 0 . 1E_T 0.2

0 . 9V_min < V < V_max 0.1 165 ≤ θ < 180 and 180 < θ < 195 0.1 T_min < T_p < 0 . 1T_max 0.1 0 . 1E_T < E < 0 . 1E_TH 0.1

V_min 0.0 θ = 180 0.0 T_p = T_min 0.0 E = E_TH 0.0

Index vaue for velocity	Index value for direction	Index value for pause time	Index value for residual energy
V = V_min	1	30 ≤ θ < 45 and 315 < θ < 330	1	T_p = T_max	1	E = E_T	1
V_min < V0 . 1V_max	0.9	45 ≤ θ < 60 and 300 < θ < 315	0.9	0 . 9T_max < T_p < T_max	0.9	E_T < E < 0 . 9E_T	0.9
0 . 1V_min < V < 0 . 2V_max	0.8	60 ≤ θ < 75 and 285 < θ < 300	0.8	0 . 8T_max < _p < 0 . 9T_max	0.8	0 . 9E_T < E < 0 . 8E_T	0.8
0 . 2V_min < V < 0 . 3V_max	0.7	75 ≤ θ < 90 and 270 < θ < 285	0.7	0 . 7T_max < T_p < . 8T_max	0.7	0 . 8E_T < E < 0 . 7E_T	0.7
0 . 3V_min < V < 0 . 4V_max	0.6	90 ≤ θ < 105 and 255 < θ < 270	0.6	0 . 6T_max < _p < 0 . 7T_max	0.6	0 . 7E_T < E < 0 . 6E_T	0.6
0 . 4V_min < V < 0 . 6V_max	0.5	105 ≤ θ < 120 and 240 < θ < 255	0.5	0 . 4T_max < T_p < . 6T_max	0.5	0 . 6E_T < E < 0 . 4E_T	0.5
0 . 6V_min < V < 0 . 7V_max	0.4	120 ≤ θ < 135 and 225 < θ < 240	0.4	0 . 3T_max < _p < 0 . 4T_max	0.4	0 . 4E_T < E < 0 . 3E_T	0.4
0 . 7V_min < V < 0 . 8V_max	0.3	135 ≤ θ < 150 and 210 < θ < 225	0.3	0 . 2T_max < _p < 0 . 3T_max	0.3	0 . 3E_T < E < 0 . 2E_T	0.3
0 . 8V_min < V < 0 . 9V_max	0.2	150 ≤ θ < 165 and 195 < θ < 210	0.2	0 . 1T_max < _p < 0 . 2T_max	0.2	0 . 2E_T < E < 0 . 1E_T	0.2
0 . 9V_min < V < V_max	0.1	165 ≤ θ < 180 and 180 < θ < 195	0.1	T_min < T_p < 0 . 1T_max	0.1	0 . 1E_T < E < 0 . 1E_TH	0.1
V_min	0.0	θ = 180	0.0	T_p = T_min	0.0	E = E_TH	0.0

An Energy based mobility_index measuring model for a node in Guiding node_list is developed. In this work free-space propagation model is incorporated [19]. Hence, the link time duration for which two nodes will stay connected is calculated for a given direction, transmission range, speed and residual energy.

Assume EMI(a,b,t) be the Energy based mobility_index of a node ‘b’ with respect to another neighboring node ‘a’, present in ‘Guiding node_list’ at any instant of time t, EMI^s (b, t) is the Energy based mobility_index of node ‘b’ calculated by the node ‘a’ that is neighbor to node ‘b’ and EMI^O (b, t) be the EMI of node ‘b’ determined by the common neighbors of nodes ‘a’ and ‘b’. Neighbor nodes are identified by reciprocating modified hello signal at regular interval in the network. A neighbor node n is decided as a common neighbor for nodes ‘a’ and ‘b’ when, the hello signals they reciprocate carry the details of node ‘n’.

Let us assume two weights of EMI^s (b, t) and EMI^O (b, t) as ‘α’ and ‘β’. Hence, EMI (a, b, t) shown in Equation (1) [18]. $EMI (a, b, t) = α {EMI}^{S} (b, t) + β {EMI}^{O} (b, t)$ (1)

Where [{EMI (a, b, t) EMI^s (b, t) {0, 1} and α, β, {0, 1} α + β = 1]

Energy based mobility_index of node ‘b’ with respect to a node ‘a’ obeys the rule i.e. EMI (a, b, t) ≠ EMI (b, a, t)

Nodes in Guiding node_list, move randomly in the area within allowed distance from Bezier trace and pause for a random span of time. After each pause node change it direction with an angle θ (0 ≤ θ ≤ 2π), here angle θ is measured with reference to the shortest distance between ‘a’ and ‘b’. Always there is a chance of node ‘b’ that move away from the transmission range of node ‘a’ or residual energy of node ‘b’ falls below minimum threshold value. So, the link time duration between two neighboring nodes depends on coverage area, mobility and residual energy of these nodes. The link time duration is also called as link expiration time [22]. Assume total time needed to do q number of observations is T. Hence T is written as shown inEquation (2) [20]. $T = \sum_{k = 1}^{q} tk$ (2) where t_k is the kth time interval and 1 ≤ k ≤ q1. Let us assume total q numbers of observations are done. For m observations, node is observed in motion and found h observations in halt state, so q = m+h. Let us take observations alternatively. Hence, T is written as shown in Equation (3) [20]. $T = T_{m} + T_{h}$ (3)

Where $Th = \sum_{z = 1}^{h} tz$ and $Tm = \sum_{l = 1}^{m} tl$

Here, t_z (1 ≤ z ≤ h) t_z is the zth time duration of halt state and t_l (1 ≤ l ≤ m) is the lth time duration of motion state. So, total halt time is T_h and the total motion time is T_m obtained during h and m observations respectively. Assume {x_q (t_l) , y_a (t_l)} be the starting coordinate for ath node in first observation. Let ${x_{a}^{'} (t_{l + 1}), y_{a}^{'} (t_{l + 1})}$ is the coordinate at time t_l+1, then $x_{a}^{'} (t_{l + 1})$ and $y_{a}^{'} (t_{l + 1})$ can be determined by $x_{a}^{'} (t_{l + 1}) = x (t) + v (a, t) cos θ (a, t)$ and $y_{a}^{'} (t_{l + 1}) = y_{a} (t_{l}) + v (a, t_{l}) sin θ (a, t_{l})$ , respectively. Here, v (a, t_l) is the speed of the node at the lth observation, θ (a, t_l) is the angle with reference to the x-axis and E (a, t) is the residual energy at lth observation. For a duration T_m, if v (a, T_m) be the average speed of the node, then it can be expressed as shown inEquation (4) [20]. $v (a, T_{m}) = \frac{1}{m} \sum_{l = 1}^{m} v_{l} (a, t_{l})$ (4) where v_l (a, t_l) is the speed of the node at the lth observation and it is calculated as shown below $v_{l} (a, t_{l}) = \frac{D_{a}^{l}}{m}$

Where, $D_{a}^{l}$ is the total distance covered by node a in lth observation and $D_{a}^{l}$ value is computed given below $\begin{matrix} (D_{a}^{l})^{2} = {x_{a}^{'} (t + 1) - x_{a} (t_{l})}^{2} \\ + y_{a}^{'} (t_{l} + 1) - y_{a} (t_{l})}^{2} \end{matrix}$

The average angle θ (T_m) (0 ≤ θ (T_m) ≤2π) for T_m is computed by the Equation (5) given below [18]. $θ (T_{m}) = \frac{1}{m} \sum_{l = 1}^{m} θ (t_{l})$ (5)

The average residual energy E (T_m) (E_TH ≤ E (T_m) ≤ E_T) for T_m is computed by Equation (6) given below $E (T_{m}) = \frac{1}{m} \sum_{l = 1}^{m} E (t_{l})$ (6)

We know EMI of node depends on its moving angle, speed, pause time and residual energy. So, we propose an index-value-based EMI computation method. In the index value based method, each node earns an index value based on its halt time T_h (T_min ≤ T_h ≤ T_max), direction θ (a, t_l) (0 ≤ θ (a, t_l) ≤2π), speed v (a, t_l) (V_min ≤ v (a, t_l) ≤ V_max) and E (a, t_l) E (E_TH ≤ E (T_m) ≤ E_T) E (a, t_l). Let $θ_{a}^{l}$ be the earned index value by node a for its direction on lth observation, v^l be the earned index value by node a for its speed on lth observation, $t_{a}^{z}$ be the earned index value by node a for its pause time on zth observation and $E_{a}^{l}$ be the earned index value by node a for its residual energy on lth observation. The node a earned index value as given in look up index table shown in Table 1. So, the Energy based mobility_index, ${EMI}_{a}^{l} (v, θ, E)$ , of node ‘a’ on lth observation for speed, angle and residual energy are computed as expressed in Equation (7) [20]. ${EMI}_{a}^{l} (v, θ, E) = \frac{1}{3} (v^{l} + θ a^{l} + {Ea}^{l})$ (7)

The average EMI_a^′ of node ‘a’ for l observation is computed as expressed in Equation (8) [18]. ${EMI}_{a}^{l} = \frac{1}{m} \sum_{l = 1}^{m} {EMI}_{a}^{l} (v, θ, E)$ (8)

Similarly at zth observation, the average Energy based mobility_index, ${EMI}_{a}^{''}$ , for halt time earned by node a is expressed in Equation (9) [18]. ${EMI}_{a}^{''} = \frac{1}{h} \sum_{z = 1}^{h} (t_{a}^{z} + E_{a}^{z})$ (9)

Where $E_{a}^{z}$ be the earned index value by node a for its residual energy on zth observation.

Hence, for the duration T, the average Energy based mobility_index, EMI_a (0 ≤ EMI ≤ 1) is computed by the Equation (10) given below [18]. ${EMI}_{a} = \frac{1}{2} ({EMI}_{a}^{'} + {EMI}_{a}^{''})$ (10)

As we know the stability of any path depends on mobility and residual energy of the node present in the route. Thus, it is required to take node’s direction, speed and residual energy in consideration while selecting a route. Assume arbitrarily two neighboring nodes a and b. Node a is having a speed of v (a, t_l), angle θ (a, t_l) with residual energy E (a, t_l), at the lth instant of time. Let {x (a, t_l) , y (a, t_l)} be the coordinate of node a and {x (a, t_l) , y (a, t_l)} be coordinate of nodes b. Hence, at the lth instant the relative speed v (a, b, t_l) between a and b is expressed by Equation (11) [20]. $v (a, b, t) = \sqrt{(α^{2}} + β^{2})$ (11) where α = v (a, t_l) cos θ (a, t_l) - v (b, t_l) cos θ (b, t_l) and β = v (a, t_l) sin θ (a, t_l) - v (b, t_l) sin θ (b, t_l)

Relative velocity can be measured when both nodes are in the coverage area of each other. Assume θ (a, b, t_l) = |θ (a, t_l) - θ (b, t_l) | is the relative angle on lth observation. Let E (a, b, t_l) = |E (a, t_l) - E (b, t_l) | is the relative residual energy on lth observation hence, the earned EMI value among a and b for relative angle and relative velocity is calculated as shown in Equation (12). $\begin{matrix} {EMI}_{ab}^{'} (v (a, b, t 1), θ (a, b, t_{1}) E (a, b, t_{1})) \\ = \frac{1}{3} (v_{ab}^{l} + θ_{ab}^{l} + E_{ab}^{l}) \end{matrix}$ (12)

Here $V_{ab}^{l}$ is the earned index value for relative velocity, $E_{ab}^{l}$ is the earned index value for residual energy and $θ_{ab}^{l}$ is the earned index value for relative angle on lth observation. So average Energy based mobility_index ( ${EMI}_{ab}^{'}$ ) for l observations is computed as shown below in Equation (13). $\begin{matrix} {EMI}_{ab}^{l} = \frac{1}{m} \sum_{l = 1}^{m} {EMI}_{ab}^{'} (v (a, b, t_{1}), \\ θ (a, b, t_{1}) E (a, b, t_{1})) \end{matrix}$ (13) ${EMI}_{ab}^{''} = \frac{1}{h} \sum_{z = 1}^{h} ({tab}^{z} + E_{ab}^{z})$ (14) where $t_{ab}^{z}$ is the acquired EMI index point and $E_{ab}^{z}$ is the acquired EMI index point at the zth observation for pause time and residual energy respectively. Hence, the relative Energy based mobility_index, EMI_ab (0 ≤ EMI_ab ≤ 1), between two nodes is determined as given inEquation (15). ${EMI}_{ab} = \frac{1}{2} ({EMI}_{ab}^{'} + {EMI}_{ab}^{''})$ (15)

We can also determine the relative mobility of any two nodes a and b present in Guiding Node_list. Figure 1 indicates that the starting points of nodes a and b are A¹ and B¹. Assume $D_{ab}^{0}$ is the distance among locations A¹ and B¹ at start time t₀. Assume the node a has shifted to point A² and node b has shifted to point B² at time t_l. Assume these nodes makes angles θ (a, t_l) and θ (b, t_l) in reference $D_{ab}^{0}$ with travelling distances of $D_{a}^{l}$ and $D_{b}^{l}$ , from their starting points. To make the calculation easy we have taken that the initial locations of nodes a and b and the locations at time t_l, lie on the circumference of a circle and all these four points make a quadrilateral area. When relative angle is known, we can calculate d₁ and d₂ using Ptolemy’s theorem as shown in the Equation (16) below [18].

Fig.1

Distance among nodes a and b in instant of time.

$D_{a}^{l} D_{b}^{l} + D_{ab}^{0} D_{ab}^{l} = d_{1} d_{2}$ (16)

Where $\begin{matrix} d_{1} = \\ \sqrt{{{(D_{ab}^{0})}^{2} + {(D_{b}^{l})}^{2} - 2 D_{b}^{l} D_{ab}^{0} \cos θ (b, t_{l})}} \end{matrix}$ $\begin{matrix} d_{2} = \\ \sqrt{{{(D_{ab}^{0})}^{2} + {(D_{b}^{l})}^{2} - 2 D_{b}^{l} D_{ab}^{0} \cos θ (a, t_{l})}} \end{matrix}$

Alternatively d₁ and d₂ are also computed without knowing the relative angles as follows:

$\begin{matrix} d_{1} = \\ \sqrt{\frac{(D_{ab}^{0} \times D_{b}^{l} + D_{ab}^{l} \times D_{a}^{l}) (D_{ab}^{0} \times D_{ab}^{l} + D_{a}^{l} \times D_{b}^{l})}{(D_{ab}^{0} \times D_{a}^{l} + D_{ab}^{l} \times D_{b}^{l})}} \end{matrix}$ $\begin{matrix} d_{1} = \\ \sqrt{\frac{(D_{ab}^{0} \times D_{ab}^{l} + D_{a}^{l} \times D_{b}^{l}) (D_{ab}^{0} \times D_{a}^{l} + D_{ab}^{l} \times D_{b}^{l})}{(D_{ab}^{0} \times D_{b}^{l} + D_{ab}^{l} \times D_{a}^{l})}} \end{matrix}$

So when relative earned index value of pause time, velocity, relative angle and residual energy of two neighboring nodes are obtained we can determine Energy based mobility_index and average relative acquired points, i.e. Energy based mobility_index using Equations (12) and (15), respectively.

Hence, EMI_ab is calculated by node a for node b, is written in Equation (17) as below [18]. ${EMI}^{S} (b, t) = {EMI}_{ab}$ (17)

The value of EMI^O (b, t) is Energy based mobility_index for node b, calculated by the common neighboring of the nodes b and a. At initial time slot EMI^O (b, t) =0. This assumption is done because node b just enters transmission range of node a and also neighbors of a. If there are n numbers of common neighbors, ${EMI}_{ab}^{O}$ can be expressed as Equation (18) [18]. ${EMI}^{O} (b, t) = \frac{1}{n} \sum_{l = 1}^{n} {EMI}_{ab}$ (18)

In this way EMI of all the nodes available in Guiding Node list are calculated. Then some alternative route bypassing obstacles are formed taking connected nodes from Guiding Node_list Formation of alternative routes using Guiding Node_list along Bezier trace is shown in the Fig. 2.

Fig.2

Formation of alternative routes using Guiding Node_list of nodes along Bezier trace.

Now EMI^O of all nodes present in a route are calculated. If EMI^O of a node is found below EMI_TH,then the route is removed from list. If all the nodes of a route posses EMI° ≥ EMI_TH then, its total EMI_T is calculated. Then route that posses maximum EMI_T is selected for data transmission

4. Simulation and performance evaluation

The simulation of the proposed Obstacle Aware Model is validated using the network simulator NS2. The most important advantage of using NS2 is the simple scalability factor using the front end Object Oriented Tool Command Language (OTCL) when compared to the back end C++ programming. The performance of the proposed OMAR considering parameters such as Packet Received Ratio, Average Delay, Control Overhead and Energy Consumption are analysed and compared with the existing AODV protocol in both plain terrain with obstacle without having avoidance model (AODV-OT). The parameters values for which simulation has been done are depicted in Table 2 [21]. Here, for each parameter five seeds of simulations are done and the average is taken to plot the graph.

Table 2
Simulation Parameters

Parameter Value

Channel Wireless Channel

Simulation span 500s

Number of nodes 50

Number of Sources 5,10,1520,25

Pause Time(s) 100 s

Max. Speed 30 m/s

MAC protocol 802.11

Routing Protocol AODV

Traffic type CBR

Initial Node Energy(J) 100 Joules

Packet Size(Bytes) 512

Terrain Area 1000m×1000m

Transmission range 250m

Network interface Wireless Phy

Mobility Model used Random Way Point

Parameter	Value
Channel	Wireless Channel
Simulation span	500s
Number of nodes	50
Number of Sources	5,10,1520,25
Pause Time(s)	100 s
Max. Speed	30 m/s
MAC protocol	802.11
Routing Protocol	AODV
Traffic type	CBR
Initial Node Energy(J)	100 Joules
Packet Size(Bytes)	512
Terrain Area	1000m×1000m
Transmission range	250m
Network interface	Wireless Phy
Mobility Model used	Random Way Point

4.1. Performance analysis in turn of varying mobility (varying speed)

In this case, Mobility of nodes is varied varying speed(m/s) while pause time is fixed at 100 sec with ten sources and performance of OMAR is analysed and compared with AODV and AODV-OT.

Simulation outcomes are listed in Table 3. Figure 4(a) shows that the PDR (%) of AODV is 95.02% at speed ‘5 m/s’ and decreases slowly to 95.01% at a speed 30 m/s. This is because at speed ‘5 m/s’, the mobility is low, so link break is less making PDR (%) high. PDR of AODV-OT is less than AODV and is 84.49%. This is because of more link breakage in presence of obstacles in realistic terrain. Same trend is observed for all other speeds. PDR of OMAR is more than AODV and is 96.56%. This is because of reduction of link breakages due to implementation of mobility aware optimal routing path. The Packet Delivery Ratio of OMAR is increased by 1.15% in comparison to AODV.

Table 3
Simulation results with varying mobility

Speed(m/s) Packet Delivery ratio(%) Average Delay(ms) Energy consumed(j) Routing Overhead

AODV AODV_OT OMAR AODV AODV_OT OMAR AODV AODV_OT OMAR AODV AODV_OT OMAR

5 95.02 84.49 96.56 34.9 47.49 32.34 4.48 7.34 3.65 0.45 1.15 0.41

10 95.29 74.76 96.08 42.8 54.76 40.38 7.21 12.72 5.93 0.58 1.33 0.51

15 96.08 72.07 97.39 55.4 65.07 51.67 9.48 16.38 8.30 0.61 1.49 0.55

20 96.32 68.43 98.09 60.1 68.43 58.48 10.12 19.42 9.28 0.75 1.58 0.68

25 94.89 65.10 95.14 78.5 94.75 69.33 11.34 20.82 9.84 0.89 1.89 0.78

30 95.01 60.33 95.95 91.8 102.34 86.19 14.14 22.88 12.93 1.01 2.14 0.98

Speed(m/s)	Packet Delivery ratio(%)	Average Delay(ms)	Energy consumed(j)	Routing Overhead
5	95.02	84.49	96.56	34.9	47.49	32.34	4.48	7.34	3.65	0.45	1.15	0.41
10	95.29	74.76	96.08	42.8	54.76	40.38	7.21	12.72	5.93	0.58	1.33	0.51
15	96.08	72.07	97.39	55.4	65.07	51.67	9.48	16.38	8.30	0.61	1.49	0.55
20	96.32	68.43	98.09	60.1	68.43	58.48	10.12	19.42	9.28	0.75	1.58	0.68
25	94.89	65.10	95.14	78.5	94.75	69.33	11.34	20.82	9.84	0.89	1.89	0.78
30	95.01	60.33	95.95	91.8	102.34	86.19	14.14	22.88	12.93	1.01	2.14	0.98

Table 4

Simulation results with varying traffic

S-D Pair	Packet Delivery ratio (%)			Average Delay (ms)			Energy consumed (j)			Routing Overhead
S-D Pair	AODV	AODV_OT	OMAR	AODV	AODV_OT	OMAR	AODV	AODV_OT	OMAR	AODV	AODV_OT	OMAR
5	95.49	85.29	97.91	43.02	59.49	39.70	3.49	7.34	2.65	0.39	0.89	0.32
10	95.29	74.62	97.07	42.8	64.76	37.38	7.21	14.72	5.93	0.58	1.33	0.51
15	93.19	70.58	95.58	84.07	109.07	78.69	11.07	19.58	9.63	1.42	2.21	1.34
20	92.18	65.10	94.16	100.3	119.43	90.72	14.43	24.19	12.19	2.02	3.76	1.94
25	81.75	60.47	85.47	400.2	550.57	350.8	18.75	29.63	15.59	2.26	4.06	2.01

Fig.3

shows the flow chart of OMAR routing.

Fig.4

(a) PDR for 50 nodes with 10 sources, (b) Avg Delay 50 nodes with 10 sources, (c) EC for 50 nodes with 10 sources, (d) Routing Overhead for 50 nodes with 10 sources, (e) PDR for 50 nodes with Speed 10 m/s, (f) Avg. Delay for 50 with Speed 10 m/s, (h) Control Overhead for 50 nodes with Speed 10 m/s.

Figure 4(b) shows the variation of Average Delay with respect to speed. Here ‘Average Delay’ is 34.9 ms for AODV at a speed of ‘5 m/s’, it increases as speed increases (Mobility increases) because link breakage also increases. ‘Average Delay’ of AODV-OT is 47.49 ms at a speed of ‘5 m/s’. The increase in delay is due to increase in link break because of presence of obstacles. ‘Average Delay’ of OMAR is 32.34 ms at at speed ‘5 m/s’which is less than AODV. The Average Delay of OMAR is decreased by 6.9 % than AODV Fig. 4(c) signifies the average Energy Consumed in AODV at the speed ‘5 m/s’is 4.48 Joules. Average Energy Consumed in AODV-OT at the speed ‘5 m/s’ is 7.34 Joules.

The increase in energy consumption is due to presence of obstacles which causes repeated link breaks and due to that again route search process is initiated making more energy consumption increases.

In case of OMAR average energy consumption at the speed ‘5 m/s’is 3.65 Joules. The average Energy Consumption of OMAR is decreased by 2.04% than AODV.

Figure 4(d) shows that the normalised Routing Overhead of AODV at the speed ‘5 m/s’is 0.45. Routing Overhead of AODV-OT at the speed ‘5 m/s’is 1.15. The increase in Routing Overhead is due to the presence of obstacles. In case of OMAR Routing Overhead at the speed of ‘5 m/s’is 0.41. The Routing Overhead of OMAR is decreased by 8.86% than AODV. Here in varying mobility scenario the performance of OMAR is better than AODV in plain terrain. This is because of use of Energy based mobility aware routing algorithm.

4.2. Performance analysis in turn of varying network traffic

In this case, source-destination pair is varied, pause time is fixed at 100 sec and speed is at 10 m/s. Simulation carried for MANET with 50 nodes. Here performance of OMAR is analysed and compared with AODV and AODV-OT.

Figure 4(e) indicates the variation of PDR% at different network traffic i.e. Source-Destination Pair (SD Pair). It is observed from Fig. 4(e) that the PDR (%) of AODV is 94.49 for five SD Pairs and decrease slowly to 81.75 for twenty five SD Pairs. The decrease of PDR (%) is due to increase of network traffic (Increase of SD Pair). As network traffic increases network congestion also increases and hence PDR (%) decreases. PDR of AODV-OT is less than AODV and is 85.29%.This is because of more link break in presence of obstacles in realistic terrain.

Same trend is observed for all SD Pairs. PDR of OMAR is more than AODV and is 96.91%.This is because of reduction of link breakage due to implementation of mobility aware optimal routing path. Here the effect of presence of obstacles is overcome by using DeCasteljau Algorithm based on Bezier curve. The Packet Delivery Ratio of OMAR is increased by 2.7 % in comparison to AODV for different cases of network traffic.

It is observed from Fig. 4(f) that the Average Delay of AODV is 43.02 ms for five SD Pairs and increases slowly to 400.2 ms for twenty five SD Pairs. The increase of PDR (%) is due to increase of network traffic because it is known that as network traffic increases network congestion also increases and hence PDR (%) decreases. Average Delay of AODV-OT is more than AODV and is 59.49 ms. This is because of more link breakage in presence of obstacles in realistic terrain. Same trend is observed for all SD Pairs.

Average Delay of OMAR is less than AODV and is 39.70 ms. The Average Delay of OMAR is decreased.

by 10.9 % in comparison to AODV for different cases of network traffic

It is observed from Fig. 4(g) that the average Energy consumed of AODV is 3.49 joules for five SD Pairs and increases slowly to 18.75 joules for twenty five SD Pairs. The increase of Energy consumption is due to increase of network traffic because it is known that as network traffic increases network congestion also increases, more link breaks and hence Energy consumption increases. Average Energy consumption of AODV-OT is more than AODV and is 7.34 joules. This is because of frequent link breaks in presence of obstacles in realistic terrain. Same trend is observed for all SD Pairs. Average Energy consumption of OMAR is less than AODV and is 2.65 joules. The Average Energy consumption of OMAR is decreased by 16.3 % in comparison to AODV for all cases of network traffic.

The performance of OMAR is better than AODV in plane terrain, this is because of effect of presence of obstacle is avoided by using the OMAR algorithm. The mobility effect and chance of dropping of a node due to shortage of energy is compensated by using Energy based mobility aware routing.

6. Conclusion

The proposed Obstacle and Mobility Aware Optimal Routing (OMAR) for MANET performs satisfactorily in presence of obstacles in a MANET terrain thereby improving the performance of the MANET. The performances of AODV and AODV-OT are compared with the proposed Obstacle and Mobility Aware Optimal Routing (OMAR) in obstacle environment. It is concluded from the simulation that the OMAR yields good enhancement in Packet Received Ratio, Average Delay, Energy Consumption and Control Overhead metrics in varying mobility and traffic scenario.

References

D.B.

Johnson and

D.A.

Maitz , Dynamic. Source Routing in Ad Hoc Wireless Networks, Mobile Computing-The Kluwer International Series in Engineering and Computer Science, Springer US 353 (1996), 153–181.

E.M.

Royer and

C.K.

Toh , A review of current routing protocols for ad hoc mobile wireless networks, IEEE Personal Communications 6(2) (1999), 46–55.

E.M.

Royer ,

P.M.

Melliar-Smith and

L.E.

Moser , An analysis of the optimum node density for ad hoc mobile networks, In Proceedings of IEEE International Conference on Communications, 3 (2001), 857–861.

Chlamtac ,

Conti and

J.J.N.

Liu , Mobile ad hoc networking: Imperatives and challenges, Ad hoc Networks 1(1) (2003), 13–64.

Lu ,

Manson and

Belis , Mobility modeling in mobile ad hoc networks with environment-aware, Journal of Networks 1(1) (2006), 54–63.

Lei ,

Wang ,

Zhou ,

Chen and

Cai , Link availability estimation based reliable routing for aeronautical ad hoc networks, Ad Hoc Networks 20 (2014), 53–63.

Han ,

Bai ,

Gong and

Wu , Link availability prediction-based reliable routing for mobile ad hoc networks, IET Communications 5(16) (2011), 2291–2300.

D.B.

Johnson and

D.A.

Maltz , Dynamic Source Routing in Ad Hoc Wireless Networks, Mobile Computing,

Imielin-ski and

Korth , Ed. Kluwer Academic Publishers 5 (1996), 153–181.

Suraj ,

Tapaswi ,

Yousef ,

K.K.

Pattanaik and

Cole , Mobility prediction in mobile ad hoc networks using a lightweight genetic algorithm, Wireless Networks 22(6) (2016), 1797–1806.

10.

J.A.

Torkestani , Mobility prediction in mobile wireless networks, Journal of Network and Computer Applications 35(5) (2012), 1633–1645.

11.

Papageorgiou ,

Birkos ,

Dagiuklas and

Kotsopou-los , An obstacle-aware human mobility model for ad hoc networks, In MASCOTS'09. IEEE International Symposium on Modeling, Analysis & Simulation of Computer and Telecommunication Systems 1 (2009), 1–9.

12.

Boschi ,

Crescenzi ,

Di and

Vocca , A Mobility Model Simulation Environment for Mobile Wireless Ad-hoc Networks, Proceedings of the 1st international conference on Simulation tools and techniques for communications, networks and systems & workshops (2008), 20–31.

13.

Kim , Adaptive MANET multipath routing algorithm based on the simulated annealing approach, The Scientific World Journal 2014, Article ID 872526 (2014), 1–8.

14.

Belghith ,

Belhassen ,

Dhraief ,

N.E.

Dougui and

Mathkour , Autonomic obstacle detection and avoidance in MANETs Driven by Cartography Enhanced OLSR, Mobile Information Systems 2015, Article ID 820401(2015), 1-20.

15.

Moussaoui ,

Semchedine and

Boukerram , A linkstate QoS routing protocol based on link stability for Mobile Ad hoc Networks, Journal of Network and Computer Applications 39 (2014), 117–125.

16.

Amjad ,

Ali ,

Jabbar ,

Hussain ,

Rho and

Kim , Impact of dynamic path loss models in an urban obstacle aware ad hoc network environment, Journal of Sensors 2015, Article ID 286270(2015), 1-8.

17.

Mi ,

Yang and

J.Y.

Yang , Restoring connectivity of mobile robotic sensor networks while avoiding obstacles, IEEE Sensors Journal 15(8) (2015), 4640–4650.

18.

Wang ,

Wang and

Zhang , A realistic mobility model with irregular obstacle constraints for mobile ad hoc networks, Wireless Networks 23 (2017), 1–20.

19.

J.M.M.

Alves , Path Planning and Collision. Avoidance Algorithms for Small RPAS, Instituto Superior Tecnico, Universidade de Lisboa, Portugal (2017), 1–10.

20.

Sarkar and

Dutta , Mobility-aware route selection technique for mobile ad hoc networks, IET Wireless Sensor Systems 7(3) (2017), 55–64.

21.

E.M.

Royer and

C.E.

Perkin , An Implementation Study of AODV Routing Protocol, Proceeding of IEEE wireless Communication and Networking Conference (WCNC 2000) 3 (2000), 1003–1008.