Performance analysis of path selection routing protocol for UVANET based on geographical multicast routing algorithm

Abstract

Urban Vehicular Ad hoc Networks (UVANETs) are the most important form of mobile ad hoc networks (MANETs). It has been broadly used in intelligent transportation system (ITS) to supply safety and comfort services to the passengers and drivers and also provide security/privacy using vehicle-to-vehicle and vehicle-to-road side unit (RSU) communications. In UVANETs, this research which maintains the security of the sensitive information shared between vehicles and end-users and tested on different types of attacks and provide possible solutions. And different emerging open challenges along with the security threats are identified and detected and the power of each node are also managed. This research proposed geographical multi cast network mode to analysis the performance of path selection routing protocols (PSRP) for UVANET because geographical multicast network mode uses central coordinator or it constructs virtual central coordinates. Due to these benefits geographical multicast routing algorithm (GMRA) is used to handle security issue, fault detection technique and power management. To simulate the proposed scheme, for performance analysis network simulator-3(NS-3) version-3.23 is used. Simulation results show that, the PDR for five CBR connections, GMRA is 4% and 7% better than GPSR and AODV routing protocols, respectively. Similarly one can observe the same advantages of GMRA at 10 CBR and 15 number of CBR connections. The packet loss of GMRA is enhanced by 10% and 7% than GPSR and AODV routing protocols respectively. The delay of GMRA for the three CBR connections are lower than GPSR and AODV routing protocols.

PUBLIC INTEREST STATEMENT

Over the past few years, the technology of vehicular ad hoc networks (VANETs) has been a significant topic of research. The idea of creating a network of cars for a particular need gave rise to the ad-hoc nature of VANET networks. VANETs are widely recognized as dependable networks used by cars in urban settings for communication. Security for VANET should be prioritized on a par with security for other computing networks. All apps created for vehicular networks must be safeguarded from malicious manipulation because to the very sensitive nature of the information broadcast via VANET. Thus, creating a dynamic routing protocol that can assist in spreading information from one node (vehicle) to another is one of the key issues in the design of vehicular ad-hoc networks. This research presented Path Selection Routing Protocol for UVANET based on geographical multicast routing algorithm to overcome this problem and analyze UVANET performance.

Keywords:

• GMRA
• PSRP
• UVANET
• V2V

REVIEWING EDITORS:

• Qingsong Ai
• Wuhan University of Technology
• China
• Wuhan; Tao Peng
• Zhejiang University
• China
• Hangzhou

SUBJECTS:

• Engineering & technology
• technology
• electrical & electronic engineering
• electromagnetics & communication

1. Introduction

In the recent years, communication technology leads to the generation of new type of networks, to deploy in various environments (Bhoi et al., 2017). Urban vehicular ad hoc network (UVANET) is one part of such type of network, where the vehicles on the roads communicate to provide safety and comfort services for the drivers and passengers (Burger & Lauer, 2018). It has a hybrid network architecture, high speed node movement characteristics, and it can provide many applications like safety(traffic signal warning system, stop sign warning system, intelligent traffic flow control) and non-safety (file transfer, mailing, internet access and simple message transfer) and services like security and TCP/IP (Guan Lim et al., 2017). Nowadays, vehicles are implementing the communication technology using the wireless modules in the cars to communicate with other vehicles and stations. This helps the drivers and passengers to get many types of services, application, driver support, infotainment (it is a types of media which provides the combination of information and entertainment), and urban sensing (Wang et al., 2018). Government, private companies and academia are working to deploy and develop VANET all over the world. The communication between the vehicles said to be vehicular communication. Basically, the objective of deploying VANET is to decrease the level of accidents and has a great impact on passenger safety and the drivers can drive smoothly in the urban areas. When the vehicle population is increasing day-to-day, then the rate of accidents also increases, so it is necessary to communicate for the vehicles (Pal et al., 2019).

The main communication models in VANETs are classified into inter-vehicular communication, V2I communication and inter-roadside communication (Khan et al., 2021). In inter vehicular communication or V2V communication, the vehicle can communicate with each other to exchange the traffic-related information within the wireless range. For instance, when an incident occurs on the road, the vehicle can immediately send the traffic information to the other vehicles nearby, suggesting them to avoid that area that if a vehicle breaks down, immediately, the vehicle begins the information dissemination process using the multi cast communication mode. If one vehicle near to the other vehicle, in this way when the communication between them is broken down, then the message is re-transmit. In such way, vehicles are notified and can take alternative routes, avoiding traffic congestion problem. The V2I mode of communication describes the communication between vehicles and fixed infrastructure. In the V2I communication, the vehicle can exchange the safety information with the infrastructure such as RSUs which are deployed on the road. The V2I communication objective is to avoid the crashes and severe incidents, provide measures of multiple safety and precautions to the vehicles.

Many routing protocols are designed to provide faster data delivery to the destination. Mobile ad hoc network (MANET) routing protocols are used to implement VANET but it is difficult to implement in VANET using these routing protocols because of cost of product(i.e. expensive), mobility (i.e. high) and change in topology (i.e. frequent and fast) (Rizwan Ghori et al., 2018). Nowadays researchers are focusing on designing secure VANET systems to protect the genuine drivers from the malicious drivers. To give robust security against the attacks, many secure routing protocols are developed. Recently many researches and developments have been made under VANET (Thakur et al., 2017), and researchers are working on the issues like routing, broadcasting, security, traffic management, information fusion, etc. The major problems that affect the performance of path selection routing protocol (PSRP) for UVANETs are unstable routing to transfer data from source to destination. Mostly the challenge that face a routing protocol for UVANETs is not only distance but also security and power management. The fault vehicles or malicious vehicles reduce the performance of the system by affecting the QoS parameters, because routing in UVANET mainly depends on the correct information, so security is a challenging task due to lack of centralized coordinator that impede trust management for nodes i.e. no centralized secure server. Power management is the second challenge because of resource constraints. When the power of the nearest node is low ongoing process link failure occurred and then link breakages lead to frequent path failures. Thus in this research used to solve such challenges geographical multicast routing algorithm (GMRA) is proposed to enhance performance of PSRP for UVANET, this routing protocol can construct virtual central coordinator that preserve enough neighborhood information to be useful in geographic routing and do not require actual position determination. It uses polar coordinates from a center point, assign ‘virtual angle range’ to neighbors of a node, bigger radius and angles are recursively redistributed to children nodes.

The remainder of the article is structured as follows: The second section examines literature review on VANETs that has been done in order to route data on VANETs. The proposed system model for the development and implementation with the provided mathematical model is covered in detail in the third section. An extensive simulation investigation was described in Section 4. The conclusion and recommendations, as well as potential future directions for the work, are covered in Section 5.

2. Related works

Routing in ad hoc networks is an important task to send the data to the destination (Meng et al., 2015). Many routing protocols are proposed for VANET to provide efficient data delivery to the destination (Li et al., 2014; Youssef et al., 2014). Topology-based routing protocols are difficult to implement in VANET due to high dynamic activity of the vehicles. Position-based routing protocols, where path maintenance is not required, only destination position and neighbor position information is required to send the data (Li et al., 2019). So, VANET prefers GMRA, because it is not considering only distance but also consider power management and security to send the data. In the work by Al-Rabayah & Malaney (2012) proposed a hybrid of reactive routing and location-based routing protocol to reduce the routing overhead for UVANET, and to increase the scalability for UVANET. The combination of reactive routing and location-based routing used to forward the data to the destination effectively. But it lacks the number of communication gaps that generated between the junctions. Taleb et al. (2007) proposed routing protocol stability by grouping different vehicles according to their velocity vectors. The vehicles that have the same velocity gives stable communication. This decreases the link failure between the vehicles. It decreases the traffic by elongating the paths selected link times. It broadcasts only the best packets. But it lacks about the vehicles ahead connectivity. That implies to increases end-to-end delay. Gomides et al. (2020) proposed a failure detection system that detects the crash and exit faults in vehicular communication system for vehicular networks. In exit fault, the vehicles that are out of range are considered as faulty vehicles. In crash fault, the equipment which is unable to send and receive the data. In the equipment this method only considers permanent fault. This decreases the delay and message loss in the network. In the OBU incorrect data generates due to noise because it does not focus on the soft-fault detection method. Harsch et al. (2007) proposed secure position-based routing protocol that based on C2C-CC Project for UVANET. The malicious vehicles with false position information are detected by using plausibility checks and digital signature. That scheme increases the end-to-end delay because it has high computation, and it lacks the communication gaps ahead. Xiong and Li 2015) first proposed ad hoc routing protocols over a realistic vehicle mobility pattern for a city scenario, and presented a simulation study that compares a position-based routing (GSR) approach with classical ad hoc routing methods (AODV and DSR). The results of simulation also demonstrated that with respect to delivery rate and latency, Position-based routing outperforms topology-based approaches. Shastri et al. (2011) also evaluated the performance of routing protocols (AODV, DSR, FSR, and TORA) in city traffic scenarios, and found out that TORA is completely unsuitable for vehicular environment, whereas AODV and FSR showed promising results, DSR suffered from very high end-to-end delay. Naumov et al. (2006) studied the behavior of routing protocols (AODV and GPSR) in an inner city environment and on a highway segment by using realistic mobility traces obtained from a microscopic vehicular traffic simulation on real Switzerland road maps. In the investigated VANET scenarios, both exhibit serious performance problems. Position-based routing protocols methods have been proposed several times, besides conventional ad hoc routing protocols. But the comparison of between position-based and topology- based routing had been carried out in (Royer & Toh, 1999; Broch et al., 1998) and this study indicated that in highly mobile environment like VANET topology based routing is less suitable than position-based routing. As topology-based routing protocol delivery ratio is less than position based routing protocol. However, those protocols have their own drawbacks. This reason motivated us to select this algorithm and analyze its performance. Now in this work we try to solve some problems related to the routing issues, such as link breakage, fault vehicles, security and power management to enhance the performance of PSRP for UVANET geographical multi cast model used.

3. System model

This system model describes the way of the system about GMRA for UVANET which states that the way of communication between the vehicles, RSUs and V2R. As shown in Figure 1, GMRA construct central virtual coordinator to monitor each node and each RSUs. This algorithm checked the power level of each nodes and RSUs. And also detect the malicious vehicles to protect the information (data packets) from the attacker. On this system model describes about the network model, security model, and power management model. Network model presents about topology of network and communication between the nodes. Security model describes about the security mechanism used to protect the information (data packets) from the attacker. Power management model describes these management schemes deal in the management of energy resources by controlling the early depletion of the battery, adjust the transmission power to decide the proper power level of each node and incorporate low power consumption strategies into the protocols.

Figure 1. General System Model of Routing Protocol for UVANET Using GMRA.

3.1. Topology of network and communication between the nodes

For a RREQ, a node may receive multiple route reply (RREP) but a node selects a route with the shortest hop distance to the member of the multi cast group and discards the other routes as shown in Figure 2. After that, the node enables the selected next hop in its routing protocol and unicasts an activation message to that node. In routing protocol Process there are three basic terms, which are: RREQ, RREP and route error (RERR). The source node broadcasting RREQ messages to other nodes to find the destination node. As shown in Figure 2 when RREQ received by nodes, then RREP messages are sent back to the source node in unicast communication. And when a communication break occurs, then it uses RERR messages to notify. Due to this reason, a new route discovery process should begin. For detecting and monitoring connectivity with neighbors, “hello” messages are permanently used. The principal stage of this routing protocol is route discovery and the process is route discovery process (Li et al., 2010). Routing process is composed of route discovery process and route maintenance process(route maintenance (Li et al., 2010) is the mechanism by which source node is able to detect, while using a source route to destination node, if the network topology has changed such that it can no longer use its route to destination node because a link along the route no longer works). The full path is formed by storing information in intermediate nodes along the route in local routing protocol. Whenever a source node desires a route to a destination node to which it does not already have a route, it broadcasts a RREQ message to all its neighbors. The neighbors update their information (RREP) for the source and create reverse route entries for the source node in their routing protocol. In this routing protocol there are two causes that are: a neighbor node receiving a RREQ and send a RREP, if it is the destination and a neighbor node receiving a RREQ and send a RREP, if it has an unexpired route to the destination. When any one of these two cases is satisfied, the neighbor node uni casts RREP back to the source node. Along the path back to the source, intermediate nodes that receive the RREP create forward route entries for the destination node in their routing protocol. But when any one of the two cases is not satisfied, the neighbor rebroadcasts (forwards) the RREQ. If the source node does not receive RREP, then it tries again and again to find a route to the destination node till the maximum RREQ retries times.

Figure 2. Topology of PSRP for UVANET (Li et al., 2010).

As shown in Figure 2, source node ‘N0’ broadcasts a RREQ message to its neighboring nodes, i.e.‘N1’ ‘N2’ and ‘N4’, and those replayed RREP to ‘N0’ at this time ‘N0’ send data packets to the shortest one. And then by updating RREQ ‘N4’ further broadcasts to the next neighboring node ‘N7’ and it replayed RREP to‘N4’ at this time‘N4’ send data packets to‘N7’. But ‘N4’ did not broadcasts to ‘N3’ and ‘N5’ why because their power level is less than the normal threshold value or the nodes are malicious. And also by updating RREQ” N7” further broadcasts to ‘N6’, ‘N8’ and ‘N10’ and those replayed RREP to ‘N7’ at this time‘N7’ send data packets to the shortest one ‘N10’. Finally by updating RREQ ‘N10’ further broadcasts to the next neighboring node ‘N13’ or destination node and it replayed RREP to ‘N10’ at this time ‘N10’ send data packets to destination node or ‘N13’. But ‘N10’did not broadcasts to ‘N9’, ‘N11’ and ‘N12’ why because their power level is less than the normal threshold value or the nodes are malicious in the Ad hoc network. The solid (green) lines indicate the link in forward path (RREQ) and the solid (Red) lines indicate the link in reverse path (RREP) as shown in Figure 2 legend.

3.2. Flow chart of the information transferring from source to destination

Figure 3 describes the flow of information (RREQ-packet, RREP-packet, RERR-packet and data-packet) between the nodes or from source to destination. The information flow concerning with both power level and security not only shortest path. As shown in Figure 3, before transfer the data packets first check the power level of in each nodes. If the power level of each node is less than that of normal power threshold value, then the data packets are not transfer from one node to other node. But if the power level of each node is greater than or equal to that of normal power threshold value (0.05 watt), then the data packets are transfer from one node to other node destination by using GMRA. Where, the received power threshold value can be derived from the SINR model as shown in Equation (1)(1) $${\mathrm{P}}_{\text{RX}}^{\text{TH}}=10\hspace{0.17em}\mathrm{\text{log}}\hspace{0.17em}\left({10}^{\left({\mathrm{P}}_{\mathrm{N}}/10\right)}+{10}^{\left({\mathrm{P}}_{\mathrm{I}}/10\right)}{10}^{(\text{TH}/10)}\right).$$(1) (Divya et al., 2015). (1) $${\mathrm{P}}_{\text{RX}}^{\text{TH}}=10\hspace{0.17em}\mathrm{\text{log}}\hspace{0.17em}\left({10}^{\left({\mathrm{P}}_{\mathrm{N}}/10\right)}+{10}^{\left({\mathrm{P}}_{\mathrm{I}}/10\right)}{10}^{(\text{TH}/10)}\right).$$(1)

Figure 3. Flow chart of information (RREQ, RREP, RERR and Data-packet) from source to destination.

where ${\mathrm{P}}_{\mathrm{N}}$ and ${\mathrm{P}}_1$ describes the noise power and the interference power in $\text{dBm},{\zeta }^{\text{TH}}$ is the SINR threshold for a successful reception in $\text{dB}.$

The schemes of power management which deal in the energy resources management by controlling the early battery depletion, adjust power of the transmission to decide a node proper power level and incorporate that strategies of low power consumption into the protocols (Zheng & Kravets, 2003). In ad hoc wireless network, the main reasons for power management are listed as Limited reserve energy, Lack of central coordination and Optimal transmission power selection. At the same time if the neighboring node is malicious, then the data packets are not transfer to that of neighboring node. But if the neighboring node is not malicious, then the data packets are transfer to that of neighboring node by using GMRA. To detect malicious nodes there are two ways: centralized and distributed. In this aricle, we used centralized detection because GMRA has centralized coordinator. When the vehicle contain valid certificate, the vehicle is genuine but when the vehicle contain an invalid certificate, then the vehicle is a malicious. Every vehicle sets a trust value for its neighbor. The certificate is signed and provided by the trusted central authority (CA).

3.3. Mathematical description of PSRP for UVANET

The urban road network contains junctions, roads, vehicles, and RSU. The roadways are single-lane or multi-lane. The urban road network model is considered as a graph G with as junction vertex V and as road connecting the two junctions edge E, i.e. (G = (V, E) = (RSU, J)) at a particular time interval RSU and vehicle beacons. In cause of that the beacon information contains identity, location, speed, and direction. To act as decision makers, RSUs are set at the junctions. For a particular time RSUs stores the vehicles beacon information that passes through the junction and it decides in which direction the data are to be forwarded next node and also if a new vehicle arrives, then updates the data. Because RSU has computing capability. Signifies a value calculated for a path using the road information to know the ability of a road to send the data from one junction to other in a minimum time as shown in Figure 4 (Bhoi et al., 2017).

Figure 4. Graphical Expression of Mathematical System Model for UVANET (Bhoi et al., 2017).

The path value is calculated by basically three parameters that are: expected delay (t_exp) to send the message from one junction to other, shortest path, and number of communication gaps between the junctions. Communication gap between the two vehicles V_i and V_j is defined as the distance after the radio range of V_i to the position of V_j. These parameters are calculated by considering information, such as vehicle current position, speed, communication gaps, shortest paths, and expected delay to send the message from one junction to other. According to Figure 4, RSU2 finds the current positions of the vehicles for the roads J2-J1, J2-J3 and J2-J6. The vehicles current positions are calculated as shown in Equation (2)(2) $$d_{\text{cover}}=\left(t_{\text{current}}-t_{\text{lbeacon}}\right)\times v$$(2) (Bhoi et al., 2017). (2) $$d_{\text{cover}}=\left(t_{\text{current}}-t_{\text{lbeacon}}\right)\times v$$(2) where ${\mathrm{d}}_{\text{cover}}$ denotes the distance covered by a vehicle after the last beacon. ${\mathrm{t}}_{\text{current}}$ denotes the current time, ${\mathrm{t}}_{\text{lbeacon}}$ denotes the time of the last beacon is received by the RSU (RSU2), and v is the speed of the vehicle.

In this process after finding the vehicles positions, RSU2 consider only those vehicles whose positions lie between the two junctions. By using the vehicles positions and the vehicles communication range, RSU2 finds data transmission logical path ${\text{Pt}}_{\text{tr}.}$ by using the following formula i.e. $\mathit{\text{RSU}}2=J_2-J_1$ or $J_2-J_3$ or $J_2-J_6$ which implies ${\text{Pt}}_{\text{tr}.}.$ To estimate the delay from one junction to other ${\text{Pt}}_{\text{tr}}$ is generated. Assumed for all vehicles and RSUs have the same Communication range. According to Figure 4, $\mathit{\text{RSU}}2$ finds $V_{\text{next}}$ $(v_5)$ vehicle by using the positions and communication range of that vehicles.

Then this process continues until RSU3 is reached. After generating the ${\text{Pt}}_{\text{tr}}$ (RSU2 $\to \mathrm{V}5\to \mathrm{V}6\to$ RSU3) from RSU2 to RSU3, RSU2 calculates the expected delay ${\mathrm{t}}_{\hspace{0.17em}\mathrm{\text{exp}}\hspace{0.17em}}$ on path ${\text{Pt}}_{\text{tr}}$ using Equation (3)(3) $${\mathrm{t}}_{\hspace{0.17em}\mathrm{\text{exp}}\hspace{0.17em}}=\frac{(\mathrm{n}-1){\mathrm{I}}_{\mathrm{M}}}{\mathrm{r}}+\frac{\sum _{\mathrm{p}=1}^{\mathrm{n}-1-\text{cg}}{\mathrm{d}}_{\text{fp}}+\text{cg}*\text{cr}}{\mathrm{c}}+\sum _{\mathrm{j}=1}^{\mathrm{n}-1}{\mathrm{t}}_{{\text{rest}}_{\mathrm{j}}}+\sum _{\mathrm{k}=1}^{\text{cg}}{\mathrm{t}}_{{\mathrm{c}}_{\mathrm{k}}}$$(3) (Bhoi et al., 2017). (3) $${\mathrm{t}}_{\hspace{0.17em}\mathrm{\text{exp}}\hspace{0.17em}}=\frac{(\mathrm{n}-1){\mathrm{I}}_{\mathrm{M}}}{\mathrm{r}}+\frac{\sum _{\mathrm{p}=1}^{\mathrm{n}-1-\text{cg}}{\mathrm{d}}_{\text{fp}}+\text{cg}*\text{cr}}{\mathrm{c}}+\sum _{\mathrm{j}=1}^{\mathrm{n}-1}{\mathrm{t}}_{{\text{rest}}_{\mathrm{j}}}+\sum _{\mathrm{k}=1}^{\text{cg}}{\mathrm{t}}_{{\mathrm{c}}_{\mathrm{k}}}$$(3) where $\frac{(\mathrm{n}-1){\mathrm{I}}_{\mathrm{M}}}{\mathrm{r}}$ denotes the total transmission delay and $n$ denotes the number of vehicles or RSUs used to forward the data from one RSU to another. $(n-1)$ denotes the number of links between the vehicles to connect the two RSUs. $\frac{l_M}{r}$ denotes the transmission delay, where l denotes the message length and r denotes the data rate.

$\frac{{\sum }_{\mathrm{p}=1}^{\mathrm{n}-1-\text{cg}}{\mathrm{d}}_{\text{fp}}+\text{cg}*\text{cr}}{\mathrm{c}}$ denotes the total propagation delay, d_fp describes the forwarding distance between the two vehicles, where p is number of forwarding distances. c denotes propagation speed. cg denotes the number of communication gaps formed due to high mobility of vehicles and presence of soft faulty vehicles, and $\text{Cr}$ denotes the communication range, which signifies the forwarding distance established after carrying the data. $\mathit{\text{cg}}\ast \mathit{\text{cr}}$ denotes the total number of forwarding distances generated by multiplying the number of gaps with the range of the vehicle. This is done because if gap occurs, then it is assumed that the relay vehicle with the data overtakes the forward vehicle and when it reaches the forward vehicle, the forwarding distance is equal to the range $({\mathrm{d}}_{\text{fp}}=\mathit{\text{cr}}).$

${\sum }_{\mathrm{j}=1}^{\mathrm{n}-1}{\mathrm{t}}_{{\text{rest}}_{\mathrm{j}}}\hspace{1em}$ denotes the total processing and queuing delay, t_rest denotes the processing delay and queuing delay where j describes the number of rest delays.

${\sum }_{\mathrm{k}=1}^{\text{cg}}{\mathrm{t}}_{{\mathrm{c}}_{\mathrm{k}}}\hspace{1em}$ denotes the total carrying delay on the path or the total message carrying delay which signifies the delay a vehicle takes to carry the message until a new vehicle arrives and t_ck denotes the carrying delay taken by the relay vehicle until the forward vehicle is encountered. $\mathrm{k}$ denotes the number of times the data is carried when a gap is encountered In this protocol, we only consider the transmission delay, propagation delay, and carrying delay to approximate the ${\mathrm{t}}_{\hspace{0.17em}\mathrm{\text{exp}}\hspace{0.17em}}.$ When a gap appears, the forward vehicle is travelling faster than the carrying delay, which is estimated by the amount of time it takes to convey the data while going forward or by the distance it travels before it reaches the junction, until a new vehicle arrives. If the forward vehicle is moving slowly, the carrying delay is estimated by the amount of time it takes to carry the data in the forward position or to get close to the junction until a new vehicle arrives (i.e. the forward vehicle is moving quickly and slowly estimates the carrying delay). The RSU2 calculates the ratio of SPt_next to SPt_current. where SPt_next denotes the shortest path from the next junction to the destination and SPt_current denotes the shortest path from the current junction to the destination. Note that Dijkstra’s algorithm is used to calculate the shortest paths. R describes the destination closeness from RSU2. Therefore, when the value of R is less, then the destination is nearer to that the RSU. After collecting the required information, such as $t_{\text{exp}},$ SPt_next and cg, RSU2 calculates the path value (${\text{Pt}}_{\text{val}}$) for a path connected to the junction as shown in Equation (4)(4) $$P{t}_{\mathit{\text{va}}{l}_i}=w_1\ast t_{\mathrm{\text{exp}}}+w_2\ast R+w_3\ast \mathit{\text{cg}}$$(4) (Bhoi et al., 2017). (4) $$P{t}_{\mathit{\text{va}}{l}_i}=w_1\ast t_{\mathrm{\text{exp}}}+w_2\ast R+w_3\ast \mathit{\text{cg}}$$(4) let $R=\frac{\mathit{\text{SP}}{t}_{\mathit{\text{next}}}}{\mathit{\text{SP}}{t}_{\mathit{\text{current}}}}$ where, $w_1,w_2\text{and}{w}_3$ are weight of the nodes and the summation of the weights w₁, w₂and w₃ is 1. The weight of a node is the sum of the weights of the edges connected to the node or node weights are just the numbers associated with each nodes. The connections between the nodes are called edges or links. The degree centrality measure is applied to graphs where the weights in W represent similarities between the nodes. In this way, a high degree centrality value of a given node means that this node has a large number of neighbors and is closely connected to them. The degree centrality measure is applied to graphs where the weights in W represent similarities between the nodes. In this way, a high degree centrality value of a given node means that this node has a large number of neighbors and is closely connected to them. The weight of every node is calculated and then the highest weight node is determined. The weight $w_n(y)$ of an edge y is the sum of the degrees of its end nodes and $w_n(y)$ is defined as twice the degree of its unique end node.

In VANET node weight is leveled as high level, middle level, and low level nodes. The high level nodes report the information which has higher trustworthiness as compared with the other levels nodes. A node level is associated with the trust value in the entity-oriented trust model as shown in Equation (5)(5) $$w_n(y)=\{\begin{array}{cc}1,\hfill & y=H_n\hfill \\ 0.7,\hfill & y=M_n\hfill \\ 0.5,\hfill & y=L_n\hfill \end{array}$$(5) (Masood et al., 2020). (5) $$w_n(y)=\{\begin{array}{cc}1,\hfill & y=H_n\hfill \\ 0.7,\hfill & y=M_n\hfill \\ 0.5,\hfill & y=L_n\hfill \end{array}$$(5) where Hn, Mn and Ln denote high, middle and low level nodes, respectively, and also the subscript n indicates the number of link nodes (Masood et al., 2020). To send the data quickly to the next junction all the parameters used to calculate the path value should be minimum. To select the next path, RSU2 uses Equation (6)(6) $${\text{Pt}}_{\mathrm{s}}=\mathrm{\text{min}}\left({\text{Pt}}_{{\text{val}}_1},{\text{Pt}}_{{\text{val}}_2},{\text{Pt}}_{{\text{val}}_3},\dots ,{\text{Pt}}_{{\text{value}}_i}\right)$$(6) to send the data (Bhoi et al., 2017). (6) $${\text{Pt}}_{\mathrm{s}}=\mathrm{\text{min}}\left({\text{Pt}}_{{\text{val}}_1},{\text{Pt}}_{{\text{val}}_2},{\text{Pt}}_{{\text{val}}_3},\dots ,{\text{Pt}}_{{\text{value}}_i}\right)$$(6)

The two path values are calculated by RSU2 and forward the message towards the junction J3, that has a minimum path value according to Figure 4. From Figure 4, it is observed that the path J2–J3 has high vehicle density than path J2–J6. After message reaches J3, using the same process the message is forwarded to junction J7 by RSU3 and then by using multi-hop communication (if not in the range) the RSU7 forwards the data to the receiver node or the destination.

4. Results and discussion

In this section, the performance of the proposed geographical multi cast routing protocol simulated on network simulator-3 (NS-3) version 3.23 software, are presented and disused. This software working on different platforms like Linux and Windows (using Cygwin) platforms. But for this work, we select a Linux platform (i.e. Ubuntu 18.04.2.0. LTS) because it supports a number of programming tools that can be used for the simulation processes. The performance of PSRP is identified by GMRA protocol comparing with GPSR and AODV routing protocols. Thus, the performance is evaluated by end-to-end delay, packet loss, throughput, packet overhead and PDR.

4.1. Simulation setup

Network simulator-3 (NS3) version 3.23 software is installed on Ubuntu 18.04.2.0. LTS, which supports a number of programming tools that can be used for the simulation processes and we added libraries like simulation urban mobility, netanim and Gnu plot library on NS3 software. (1) By adding simulation urban mobility (SUMO) (version 0.32) library in the network simulator software (version 3.23 or NS-3.23) we get the goggle map. From, the goggle map we adjust the track, the bus and the time and we get the trace file. From the trace file the maximum number of nodes and the maximum number of time are taken and then the maximum number of nodes and the number of time added on the code to connect the link path. (2) In other side by adding netanim (version 3.108) library in the NS3 software, the animation way is obtained by adjusting TCP/IP and UDP protocols and this helps to see the movement of the nodes. (3) Finally by adding Gnu plot library in the NS3 software the graph of the result is obtained. Figure 5 indicates work flow of the software for simulation system that has been discussed before.

Figure 5. Work flow of the software for simulation system.

4.2. Simulation parameter

Table 1 shows that most of the simulation parameters with their own values.

Table 1. Simulation parameter.

Parameter	Values descriptions
Simulator	NS3 (V-3.23/SUMO-0.32)
Mac protocol	802.11p (Bhoi et al., 2017)
Number of nodes	0–110 (Silva et al., 2019)
Data rate	3Mbps (Bhoi et al., 2017)
Data type	CBR (Silva et al., 2019)
Packet size	512 Bytes (Bhoi et al., 2017)
Simulation area	1100 * 1100 (Silva et al., 2019)
Max. speed	15 m/s (Silva et al., 2019)
Transport protocol	UDP (Bhoi et al., 2017)
Transmission range	250 m (Silva et al., 2019)
Routing protocol	GMRA, GPSR and AODV
Propagation model	Two-ray ground
Simulation time	170 s

4.3. Packet delivery ratio versus CBR connections

As shown in Figure 6, as the number of CBR connections increases, then the packet transmissions are increases. So that, in the network the traffic is very high due to that packet delivery ratio PDR also increases in GMRA, GPSR and AODV protocols. The PDR increases as CBR connections increases and it depends upon the node movement and positions. But the GMRA protocol provides high PDR when we compare with GPSR and AODV protocols. According to the PDR test, when the two vehicles distance or the distance between them increase, PDR decreases or reduces. Thus Figure 6 explains the benefit of using GMRA protocol in terms of PDR with varying traffic loads. When the traffic load increases, the PDR of AODV and GPSR protocol also decreases, because of the increase in the routing and data packets due to the mobility nodes are very high that participate in the routing process, it causes channel contention and packet collision that leads to dropping in packet delivery. As one can see from Figure, GMRA protocol outperforms AODV and GPSR protocol at every traffic load. Because of this, GMRA protocol selects the routing way with the control packets reduction as well as the longest lifetime way. When the routing message rebroadcast is very less the bandwidth consumption will be small. So that when the PDR is better, then the routing protocol is more complete and correct. As a summary of Figure 6, for five CBR connections, GMRA is 4% and 7% better than GPSR and AODV routing protocols respectively. Similarly at 10 CBR connections, GMRA is better 7.5% and 6.5% than GPSR and AODV routing protocols respectively. And one can observe the same advantages of GMRA at 15 number of CBR connections.

Figure 6. Packet delivery ratio vs. no. of CBR connection.

4.4. Packet loss versus CBR connections

Figure 7 shows the packet loss for different number of CBR connections. As it as shown from the figure, the packet loss increase as CBR connections and it depends upon the node movement and positions where the GMRA protocol provides less packet loss than GPSR and AODV protocols. The packet loss at five CBR connections, the packet loss of GMRA is enhanced by 3% and 3.5% than GPSR and AODV routing protocols, respectively. And one can observe the same advantages of GMRA at 10 CBR and 15 number of CBR connections.

Figure 7. Packet loss vs. no. of CBR connection.

4.5. End-to-end delay versus CBR connections

Figure 8 illustrates the end to end delay for different number of CBR connections. As the CBR connections increase, the packet transmissions increases gradually leads to very high traffic in the network. Thus, the route discovery may take long path and the end-to-end delay will increases too. Therefore, the end-to-end delay increases as CBR connection increases and it depends upon the node movement and positions. But, the proposed GMRA protocol provides low end-to-end delay as compared with the GPSR and AODV protocols. The numerical comparison of the three protocols, let at 15 CBR connection, the delay of GMRA is at 15 CBR connections GMRA is 5.65% and 4.55% lower than GPSR and AODV routing protocols respectively.

Figure 8. End to end delay vs. no. of CBR connection.

4.6. Throughput versus number of nodes

To measure the network connection performance, throughput which is a good way because it tells that at their destination how many messages are arriving successfully. When majority of the data or the messages are delivered successfully, then the throughput is very high that will be considered. So, throughput increases as node increases and it depends upon the node movement and positions. But in some causes the network traffic throughput decreases with an increase in the number of nodes between the source and the destination because the packet error probability increases with an increase in the number of nodes. The GMRA protocol provides high throughput when compared with that of GPSR and AODV protocols because GMRA has central virtual coordinator that monitors the packet error probability. The simulation result in Figure 9, shows that the GMRA protocol gives higher throughput than the GPSR and AODV protocols. Because to select the best routing way in the GMRA protocol the node mobility that has fewer probabilities to break the link for the data transmission considered. However, the existing AODV protocols selects the shortest route and do not consider the node’s mobility rather selecting the route that may cause the frequent link breakage which affects the overall network throughput. In addition, the fewer rebroadcasts result in less degree of collision and contention, which leads the GMRA protocol to get higher throughput. In a proposed algorithm in cause of the route discovery process, it consume less time when we compared to AODV protocols and GMRA, due to the reduction of a link breakage problem. Throughput for PSRP as shown in Figure 9, GMRA protocol is better than GPSR and AODV routing protocols. But from around 7–15 nodes and from 25 to 30 nodes are not consistent or decreases because of power level. So we considered the average throughput value or the maximum threshold value.

Figure 9. Throughput vs. no. of nodes.

4.7. Packet loss rate versus number of vehicles

As shown in Figure 10, as the number of nodes increases the network connectivity increases, which leads that the packet loss rate decreases. So, the packet loss rate is decreases as the number of nodes increases for the three (i.e. GMRA, GPSR and AODV) routing protocols. The proposed GMRA is lower packet loss rate than GPSR and AODV routing protocols due to the power level and security. But, the AODV protocol is crossing smoothly the GPSR and GMRA routing protocols starting from around 75 nodes and 87 nodes, respectively. And it has low packet loss rate at higher nodes because the AODV routing protocol considers only distance but not power level to reach destination. But GMRA protocol selects the shortest pass and monitors its power level and security by increasing number of nodes and by reducing the number of hops to send the data to the destination.

Figure 10. Packet lost rate vs. no. of vehicles.

4.8. Average end-to-end delay vs. number of vehicles

Figure 11 illustrates that the proposed GMRA routing protocol is lower end-to-end delay than GPSR and AODV routing protocols. When the number of vehicle is low, the protocols is encounter network gaps in the path which increases the end-to-end delay due to increase the carrying delay. So GMRA routing protocol shows less end-to-end delay because it construct virtual central coordinator that manage the nodes in terms of power and security. When the density of the node is increased to 40, 50, 60, 70, 80 and 90, at shown in Figure 11, it is observed that no gaps are encountered by the protocols due to high availability of vehicles in the radio range. So that, all the protocols perform similar in high density scenarios by showing less end-to-end delay.

Figure 11. Average end-to-end delay vs. number of nodes.

4.9. Packet overhead versus number of vehicles

Network overhead is calculated by sending a fixed-size data transmission across the network and observing the number of extra bytes of data transmitted for the action to be completed. The overhead of a packet type is the amount of wasted bandwidth that is required to transmit the payload. The packet header is extra information put on top of the payload of the packet to ensure it gets to its destination. But there are the factors that affecting network performance like bandwidth of the transmission medium, type of network traffic and the number of transmission errors. Within the network, overhead can be happen in packets exchange process. So that when the packet overhead is very high, then the traffic is also very high which required high bandwidth. Due to that packet overhead influence the network because of the factors that affecting network performance. In ad-hoc network there is a continuous exchange of data packets, acknowledgment packet, data request packet and other packets. Hence, this work evaluated how efficient cooperative re-transmission is in avoiding excessive overhead that could be created within an ad-hoc network. As shown in Figure 12, because of GMRA control all the nodes, its packet overhead is less as compared with GPSR and AODV.

Figure 12. Packet overhead vs. no. of vehicles.

5. Conclusion

In this research, for UVANET in urban environment a PSRP based on GMRA is proposed to send the data to the destination. It uses GV2GV communication to increase the performances of the result. And also improved the power management in each node by checking the power level of each node (i.e. the power level of each node must be less than the power threshold value) and detect the impact of malicious node to secure the data in each node using GMRA. The proposed algorithm was simulated successfully in an urban sparse scenario in NS-3. The experiments demonstrate that the proposed GMRA protocol shows better performance over GPSR and AODV protocols as far as packet loss rate, PDR, end-to-end delay, throughput and packet overhead are concerned. The PDR for five CBR connections, GMRA is 4% and 7% better than GPSR and AODV routing protocols respectively. Similarly one can observe the same advantages of GMRA at 10 CBR and 15 number of CBR connections. The packet loss of GMRA is enhanced by 10% and 7% than GPSR and AODV routing protocols, respectively. And one can observe the same advantages of GMRA at 5 and 15 number of CBR connections. The delay of GMRA for the three CBR connections are lower than GPSR and AODV routing protocols. Finally, the proposed protocol also enhanced throughput, good put and packet overhead by reducing the number of hops used and by increasing number of nodes to send the data to the destination. In this research, for UVANET in urban environment a PSRP based on GMRA is proposed to send the data to the destination. And improved the power management in each node by checking the power level of each node (i.e. the power level of each node must be less than the power threshold value) and detect the impact of malicious node to secure the data in each node using GMRA. The proposed algorithm was simulated successfully in an urban sparse scenario in NS-3. The experiments demonstrate that the proposed GMRA protocol shows better performance over GPSR and AODV protocols as far as packet loss rate, good put, PDR, end-to-end delay, throughput and packet overhead are concerned. The PDR for five CBR connections, GMRA is 4% and 7% better than GPSR and AODV routing protocols respectively. Similarly one can observe the same advantages of GMRA at 10 CBR and 15 number of CBR connections. The packet loss of GMRA is enhanced by 10% and 7% than GPSR and AODV routing protocols respectively. And one can observe the same advantages of GMRA at 5 and 15 number of CBR connections. The delay of GMRA for the three CBR connections are lower than GPSR and AODV routing protocols. Finally, the proposed protocol also enhanced throughput and packet overhead by reducing the number of hops used and by increasing number of nodes to send the data to the destination.

Future study should alter the GMR algorithm to take into account many factors including speed, direction, and node density. It should also compare the GMR method in both a highway and a dense urban situation to produce simulations that are more accurate.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Notes on contributors

Fikreselam Gared Mengistu

Fikreselam Gared Mengistu received his PhD from Taiwan National University of Science and Technology (NTUST), Taiwan, in Communication Engineering. He received his MSc and BSc from Addis Ababa University and Bahir Dar University, Ethiopia, respectively. Hiruy Daricha Yihunie received her MSC from Bahir Dar University, Ethiopia and his research interests are in the areas of wireless networks and security.