Ant-Colony Optimization Based Energy Aware Cross Layer Routing Protocol to Improve Route Reliability in MANETs

Improved route reliability is one of the key parameter for successful dissemination of the multimedia and emergency messages in decentralized, highly dynamic Mobile Ad Hoc Networks (MANETs). In order to achieve this, an energy efficient optimized routing protocol is essential, as mobile nodes run on battery power. There is plethora of studies that emphasizes on route reliability and energy consumption by modifying the existing MANET reactive routing schemes. Most of these are based on shortest path with traditional single layer approach, allows only one or two node parameters for both relay node selection and optimal path. However, due to the nature of the MANETs, still several factors such as unnecessary flooding of control packets and intermittent connectivity among the nodes distress the battery power directly. Therefore, in this paper, an Ant colony optimization based energy aware cross layer routing protocol (AOERP) is presented. The proposed AOERP includes series of mechanisms: firstly, selection of an adaptive relay nodes based on high energy factor and neighbor node ratio at each node in the network to avoid dead nodes and secondly identifying a finest path among the selected relay nodes in the direction of destination using Ant Colony Optimization (ACO) method. These mechanisms make an attempt to improve the route reliability and in turn increases entire network lifetime to overcome the rerouting in the network by making use of multiple node parameters from lower three layers such as physical, MAC and network layers. The performance of the proposed AOERP protocol is carried out using Network Simulator Version 2.34 (NS-2.34) and compared over existing energy aware routing schemes. The simulation results show that the proposed AOERP outperforms in terms of reliability, energy factor, average end-to-end delay and percentage of dead nodes for different node density and node speeds.


Introduction
Mobile ad hoc network (MANET) is an advanced version of wireless communications, offers entertainment services like streaming of audio/video content and emergency services such as traffic/road safety information exchange among highly dynamic mobile nodes. Thus, MANETs have gained considerable attention in the recent past. However, due to undesirable characteristics of mobile nodes makes data transmission a challenging constraint in MANETs [1,2]. In addition,based on the limited transmission range of the mobile nodes, each and every node in the network communicates each other directly, if the nodes come under desirable communication range.Otherwise, demands multi-hop communication through additional cooperative relay nodes toexchange the packets. Towards this, there exists several routing schemes have been designed byconsidering position and topological information of the mobile nodes. The position/location basedrouting schemes establishes communication links through the exact location of the mobile nodesusing Global Positioning System (GPS). Wherein, if GPS fails to work, there will be no exchangeof information until replacement of GPS device is done. On the other side, topology based routingalgorithms starts exchanges the information via routing links and suffers from rerouting duringlink breakages. However, as per the literature, the drawback of rerouting can be addressed throughselection of adaptive relay node with an efficient optimization method [2]. Further, topology basedRouting schemes for MANETs are categorized into: proactive, reactive and hybrid protocols [3]. Proactive routing schemes are also called as table-driven protocols, initiates routing processirrespective of demand of the route for information exchange. The reactive/on demand routingschemes establishes links with in the network when there is a demand to exchange the informationwith the other mobile nodes. Finally, Hybrid routing protocols, the combination of both proactiveand reactive protocols [4]. Thus, routing is an essential aspect of networking of data packetsbecause the performance metrics like reliability and congestion depends upon the routinginformation.
However, due to the wireless medium in ad-hoc networks, suffers from low bandwidth and high error rates. Thus, there will be a significant effect on energy consumption of mobile nodes or even may lead to power failure of the nodes due to multi-hop communication. Further, the power failure of the mobile nodes reduces the entire network reliability due to sudden increase of dead nodes. Therefore, the effective utilization of node energy plays a crucial role in MANETs and recommends an energy aware routing scheme with reduced hops for successful transmission of the packets in MANETs [5].
Towards this, ant colony optimization based routing schemes have been designed to get better results by considering position and topological information of the mobile nodes [6]. All these schemes includes mobile agents termed as artificial ants behaves as routing packets for identification of reliable links among mobile nodes. These ant based optimized schemes are characterized by Swarm Intelligence to have adaptive routing in mobile ad-hoc networks and robust to agent failures with load balancing. In addition, swarm intelligence methods are used for monitoring mobile nodes and to address the combinatorial optimization issues [7][8][9]. Accordingly, this allows energy efficiency and thus extending the reliability of the network [10].
Most of the metaheuristic ACO routing protocols in MANETs are designed based on traditional single-layer approach that can fulfill two or three Quality of Service (QoS) metrics. Therefore, in order to support all the QoS requirements, cross layer paradigm is considered in addition to ACO in our proposed scheme to make use of lower three layers such as physical, MAC and network layers. Thus, a cross-layer routing design with ACO permits data exchange among different layers to reach desirable network performance.
Therefore, Ant colony optimization based energy aware routing protocol (AOERP) is presented for improved reliability in MANETs. AOERP exploits cross layer information to select adaptive relay nodes based on high energy factor and identifies a finest path among the selected relay nodes for networking of data packets using Ant Colony Optimization (ACO) method in the direction of destination.
The important contributions of the proposed AOERP are listed as follows: 1. Adaptive Relay Nodes (ARNs) are selected based on the calculation of Energy Factor and Neighbor Node Ratio (NNR) at each mobile node in the network. 2. Identifying the optimal path through ARNs by calculating the Pheromone value by considering Stability factor, Link Expiration Time (LET), Congestion and Hop count towards the destination to find the finest path.

Performance evaluation of proposed AOERP is compared with existing Efficient Power
Aware AODV (EPAAODV) and K means AODV ACO an Energy Aware Ant based routing algorithms [24,25] for MANETs in terms of reliability, energy factor, average end-to-end delay and percentage of dead nodes for different node density and node speeds.
The rest of the paper is separated into different sections. Section 2 illustrate the literature of existing energy aware and ant based routing protocols briefly. Section 3 discuss the proposed AOERP Then, using Network Simulator (NS)-2.34 the performance study of the AOERP is verified in Sect. 4. At the end, Sect. 5 concludes the paper.

Literature Survey
According to the existing works, recently, there has been a plethora of research on Energy aware routing protocols for networking of data packets in MANETs to overcome the drawbacks due to inherent characteristics of the mobile nodes [10][11][12][13][14][15][16][17][18][19][20]. Towards this, authors in [13] have proposed an enhanced AODV routing protocol in MANET to improve energy efficiency and network lifetime by considering average energy of network. Maleki et al. [14] have proposed a power aware on demand routing mechanism to extend the lifetime of the network. Authors have applied a greedy policy so as not to use the overused links to reduce the routing error and battery cost. Authors in [15] proposed an optimized energy aware routing (OEAR) protocol by considering node energy and number of packets buffered to identify the finest path. In addition, the OEAR protocol allows mobile nodes to select the reliable path from source to destination through on-demand routing. In [16] authors have proposed an Optimized Energy Efficient Route Assigning (OEERA) method to reduce energy consumption and enhance network lifetime in MANETs.
Deepa et al. [17] have proposed an improved AODV routing scheme known as Dynamic Energy AODV (DE-AODV) routing protocol. The proposed DE-AODV allows mobile nodes to have dynamic energy to improve overall network lifetime by reducing the energy consumption. In addition, DE-AODV routing protocol selects the finest routing path through the selection of intermediate nodes with highest energy in the direction of destination. Simulation results showed that the performance of DE-AODV is superior over existing on demand energy aware routing protocols for different QoS metrics.
. In [18] authors have proposed a new load balancing congestion control energy aware routing protocol. In which, authors have provided a solution to transfer the transmission load from low energy to high energy nodes and identifies the multiple paths in the network for efficient data transmission. Authors in [19] made an attempt to enhance the efficiency of the network by improving peer search rate and reduce traffic overhead. In this, Abkenar et al. [20] have proposed a novel Ant Colony Routing (ACR) named as Weighted Probability Ant-based Routing (WPAR). In this, authors have combined both ant and probabilistic routing, in addition selected two routing metrics like energy consumption and hop count for optimal path between source and destination to improve lifetime of the network.
De Rango et al. [21] have proposed another ant based algorithm for energy saving in MANETs with use of various optimization metrics, such as energy utilization, load balancing and delay. In [22] authors proposed an Ant Colony-based Energy Control Routing (ACECR) protocol chooses finest route using the positive feedback character of ACO. In addition to the hop count, ACECR considers minimum and average energy of the mobile node to provide energy saving and extend the network life time. Authors in [23] have proposed an ant colony based AODV routing approach to select suitable path for data transmission during Link Failure (LF). Towards this, a mechanism named prediction based mechanism is employed for selecting the stable path during route maintenance phase.
Mafirabadza, C. et al. [24] have proposed an Efficient Power Aware AODV (EPAAODV) as an enhancement to PAAODV to improve network life span and reducing delay by choosing finest path with maximum energy.
Kumar et al. [25], presented a modified AODV using ACO and K-means clustering method to suppress routing issues in MANET. Primarily, the routing of control packets is initiated same as in traditional AODV. Then, using ant based optimal path, data packets are transmitted from source to destination. The AODV-ACO protocol chooses the optimal path by selecting two critical routing factors such as power constraints and delay metrics. Based on simulation results, the performance of AODV-ACO is superior over existing AODV under low mobility and traffic scenarios.
However, from the literature it is observed that the most of the Energy aware and Ant based routing protocols designed for MANETs are based on traditional single-layer approach. All the routing protocols mentioned above do not allow to use all routing metrics at a time from lower three layers such as physical, MAC and network layers. Further, traditional single layer based routing mechanisms allows only few routing metrics such as node propagation range, travel time and neighbor node density in relay node selection. Thus, to address above mentioned issues, we propose an Ant colony optimization based energy aware cross layer routing protocol in MANETs against inherent behavior of the mobile nodes. In addition, cross layer paradigm allows routing protocols to consider multiple node parameters for relay selection and as well as optimal path identification.

Proposed work
The proposed AOERP scheme is explained in this section to enhance reliability of the network by improving energy efficiency and through optimal path from source to destination in MANETs. Further, in AOERP for effective communication among mobile nodes, each node is a equipped with IEEE 802.11p standard against Global Positioning System (GPS) for channel access dynamically.

Adaptive Relay Node (ARN) Selection in the Network
In MANETs, most of the MANET routing schemes preferred shortest path algorithms for packet transmission, however it is always not preferable for packet transmission due to different speeds, energy consumption and congestion in the node. Therefore, in the proposed AOERP scheme, at the beginning, each mobile node preserves immediate or 1-hop neighbor list through periodic exchange of HELLO packets. In addition, each and every node in the network maintain the information like, address of source, destination nodes, delay time and hop count. Along with this information, the proposed AOERP enables nodes to maintain energy factor and neighbor node ratio at each node's routing table by adding new fields to the traditional AODV Route Request (RREQ) control packet as shown in Table 1.
Therefore, in the proposed AOERP, selection of ARN is majorly done based on energy factor and Neighbor Node Ratio (NNR) at each mobile node in the direction of destination and are elaborated mathematically in 3.1.1 and 3.1.2 sub sections. Then, the nodes with maximum energy factor and minimum NNR are chosen as adaptive relay nodes.

Energy Factor of a node
In the process of identifying relay nodes, Energy factor is chosen as one of selection parameter in addition to hop count, so as to eliminate the distress caused by the limited energy sources. It is defined as the ratio of residual energy to that of initial energy of a relay node. It can be obtained using the Eq. (1) and its value always lies in 0 to 1. I i , E i are termed as energy levels before and after simulation of a node. Then, node energy factor is recorded in AOERP-RREQ packet. If the nodes energy factor is less than threshold value will be discarded from routing, else AOERP-RREQ initiates calculation of NNR.
Threshold value for energy factor consideration at each node is given by E Th ≥ Ef/2.

Calculation of Neighbor Node Ratio Value
As per the literature, the hop count alone is not always preferable for adaptive relay selection, thus the proposed AOERP calculates NNR value at each mobile node having energy factor greater than 0.5 for the transmission of the AOERP-RREQ to the destination. Mathematically, NNR value for choosing relay node is calculated using Eq. (2) as shown below Here, X i , Y i , X j and Y j represents the coordinates of the relay nodes i and j respectively.
The term Link Expiration Time (LET) is defined as the time lapse at a particular node for data exchange. Where the parameters R i and S i represents packet received and transmitted times at node i. The denominator m represents the number of nodes in the network.
[27] Finally, the node count in each path is calculated using the Eq. (5)

Routing Table Field Format of the Proposed AOERP
The functioning of the proposed AOERP is truly based on the data available in the routing table at each node. The additional fields like Energy Factor, NNR, Stability factor and Pheromone values are incorporated to the existing fields presented in the normal AODV at each node. Table 1 represents the AOERP routing table (Table 2).

Ant Colony Based Optimal path Selection
Besides the existing Ant based optimization schemes, the proposed AOERP includes more number of critical factors like delay time, stability factor, congestion and hop count along the path for the optimal route selection. The information about these metrics are updated every time in adaptive relay node's routing table. Based on the identified ARNs using Neighbor Node Ratio, the pheromone value is calculated using the formula given by Eq. (6) at all identified relay nodes. Where Dt ij is delay time of jth neighboring node of 'I, Sf j is stability factor of jth relay node, H ij is the hop count that the AOERP-RREQ control packet has traversed between i th and jth mobile nodes. In addition, whenever source node wishes to communicate with the destination, source node verifies for the link through the information available in its routing table. If identifies any path, data packets are transmitted. Else, source floods AOERP route request control packets to its immediate 1-hop neighbors. If the immediate 1 hop neighbor is not a destination, delay time calculation is initiated using the Eq. (7) and compared with threshold value (T th ) given by Eq. (8).
Where, : is wavelength in meters. T: represents the Time, P t : referred as transmitted antenna power signal in watt. Gt & Gr: known as the transmitter and receiver node gain in decibel. E d : measures the distance (m). L: represents the system loss. R: Indicates radio range (m).
If the delay time is less than the half of the radio range (R), source node records delay time in 1-hop neighbor node's pheromone table. Further, the Hop-Count field of modified AOERP-RREQ packet is incremented by 1 and the same is updated in corresponding neighboring node pheromone table. In the similar lines, adaptive relay node stability factor associated is calculated using Eq. (9).
where Dt ij is delay time of jth neighboring node of 'i' and Dt ai is an accumulated delay times of the same node 'i' in the latest period. Then the pheromone value is computed based on delay time, number of relays, stability factor and route cost metrics. The calculated pheromone value is stored in modified AOERP-RREQ control packet pheromone value field. If the immediate relay is the destination, AOERP registers an entry of AOERP_ RREQ in the routing table. It measures the pheromone value and the same is stored in pheromone fields of received RREQ_AOERP and routing table. Upon receiving multiple AOERP RREQ control packets, the procedure is continual. Then, the destination node records the pheromone values. Then AOERP RREQ control packet is sent back to the originated adaptive relay nodes using the reverse path with least pheromone value.
Route maintenance in the proposed AOERP is done by identifying delay time and stability factor performance metrics. As mentioned, for route preservation, the delay time is maintained as half of the radio range with unity stability factor so as to quickly react to link breakages and run an alternative good quality links. Thus reduces the route error rate and choses the most stable path.

AOERP-Performance Evaluation
In this section, simulation environment is defined and then the performance analysis of proposed AOERP is carried out under two different scenarios such as, variable node density and node speeds to known the dynamic behavior.
Nodes are randomly distributed in the 1500 m X 1500 m by considering random way point model. The remaining network parameters are chosen as per the Table 3. The simulation of proposed AOERP is performed over a topology 1000 × 1000 m size using Network Simulator (NS)-2.34 and includes the following scenarios.

Scenario 1
In this, the impact of variation in number of nodes from 50 to 300 for different QoS routing metrics is examined to measure the performance of AOERP. In addition, speed of the each node and packet size are kept constant at 25 m/s and 512 bytes respectively.

Scenario 2
In this, to check the dynamic behavior of the proposed AOERP is studied by varying the speed of the nodes from 5 m/s to 25 m/s for different QoS routing metrics. Here, node density and packet size are kept constant at 200 nodes and 512 bytes respectively. The performance of proposed AOERP is estimated using the following routing metrics.
• Average End-to-End (E2E) delay : It is defined as the average amount of time spent for networking of a data packet from source to destination.  • Reliability: It measures the ratio of successful reception of data packets over transmitted. It is denoted by the letter R and the value is always lies in [0, 1], therefore, R = 1 shows the best case.
• Dead nodes (%): It illustrates the ratio of number of dead nodes over node density in the network.
• Energy factor: It determines the ratio of residual energy to that of initial energy of a relay node. It can be obtained using the Eq.
(1) Figure 1 shows the performance analysis of proposed AOERP and existing EPAAODV [24] and K means AODV ACO [25] routing protocols for different node density against end to end delay metric. It is observed that the average delay for transmission of data packets to the destination is gradually increasing in all the routing schemes as number of nodes increasing. However, the proposed AOERP exhibits reduced delay over existing approaches even at higher node density. This, is because of selection of adaptive relay nodes at the initial stage of the simulation using two critical parameters such as energy factor, neighbor node ratio and ant based optimal path using stability factor, link expiration time, hop count and congestion. Further, it is evident that the delay factor is about 31% and 21% less in proposed AOERP over existing EPAAODV [24], K-means AODV ACO [25] schemes respectively. The reduced delay in AOERP indicates reduced hop count through NNR and optimal path selection using ACO. Figure 2 illustrates that the decrease in energy factor for increase in node density from 50 nodes to 300 nodes in all the three routing schemes. However, the energy factor in proposed AOERP is superior to existing approaches. Further, the energy factor in AOERP is about 10% higher against EPAAODV and 17% efficient than Sum of data packets received Sum of packets transmitted Total Number of dead nodes Node density in the network Fig. 1 End to End Delay for Number of nodes K means AODV ACO schemes at higher node density. The improvement in energy factor indicates the reduced energy consumption of the mobile nodes and improved lifetime due to selection of adaptive relay nodes, finest path using NNR and ACO methods. Figure 3 shows that the reliability performance metric is better in case of proposed AOERP for variation in number of nodes over existing variants of AODV routing protocols [24,25]. This is the measure of safe dissemination of data packets to the destination. Further, the reliability parameter is 10% and 15% higher in case of AOERP over EPAAODV and K means AODV ACO. The improvement in reliability is due to selection of adaptive relay nodes, finest path using NNR and ACO methods.Similarly, Fig. 4 shows that the variation of percentage of dead nodes for different node density in all the three routing protocols. In addition, the number of dead nodes are less in percentage against existing EPAAODV and K means AODV ACO as number of nodes varying from 50 to 300. This is due to the inclusion of energy factor metric in adaptive relay node selection and stability factor parameter in optimal path identification for data packet transmission.  On the other side, Fig. 5 represents the performance variation of proposed AOERP in terms of delay metric for different node speed against existing EPAAODV and K means AODV ACO approaches. Further, simulation results illustrates that the average end to end delay is decreasing in all the three routing schemes as the speed increases from 5m/s to 25m/s. However, AOERP shows better delay performance and is about 25%, 10% less over existing power aware, ACO approaches in MANETs.
From the Fig. 6, it is evident that energy factor is decreasing gradually for different speeds in all the three approaches. However, the energy factor is superior in case of proposed scheme over existing routing approaches. Numerically, energy factor in AOERP is more and is about 10%, 13% higher against EPAAODV, K-means AODV ACO. This is because of relay node selection and optimal path identification in the direction of destination using energy factor, ACO.Similarly, Fig. 7 represents the variation of reliability metric in case of proposed AOERP and existing schemes [24,25] for different  node speeds. Further, it is evident from the graph that the reliability metric is superior in proposed AOERP and is about 15% and 21% higher against existing approaches at higher mobility conditions. This shows, efficient performance of proposed approach for dynamic variation of node speeds varying from 5 to 25m/s. The improvement in reliability is due to selection of adaptive relay nodes, finest path using NNR and ACO methods.On the other side, Fig. 8 illustrates the variation of percentage of dead nodes for different node speeds in all the three routing protocols. In addition, the number of percentage of dead nodes are less in proposed AOERP and is about 10%, 24% over existing EPAAODV and K means AODV ACO at higher mobility scenario. The reduced number of dead nodes in the proposed scheme is due to the inclusion of energy factor metric in adaptive relay node selection and stability factor parameter in optimal path identification for data packet transmission

Conclusion
In this paper, an Ant colony optimization based energy aware cross layer routing protocol (AOERP) for MANETs is presented. The proposed AOERP aims at establishing reliable links through the selection of adaptive relay nodes using energy factor, node neighbor ratio and optimal path among selected relay nodes using ACO. The performance of AOERP has been evaluated by comparing with the existing EPAAODV and K means AODV ACO routing schemes for different node density and speeds. The simulation results shows that the proposed AOERP is superior to existing schemes against different performance metrics like, end to end delay, energy factor, reliability and percentage of dead nodes. Further, as an extension, we plan to study the proposed scheme through multiple critical factors to improve security challenges against vulnerable attacks in MANETs under different scenarios to meet the latest spreads in safety applications.
Authors Contribution Drafting and Experimentation: SS; Supervision: DVR.
Funding Not Applicable.

Declarations
Competing interests Authors declare that they have no Competing Interest.
Ethical Approval Not applicable.

Consent to Participate Not applicable.
Consent to Publish Not applicable.