Performance Evaluation of AODV, DSDV, and ZRP Using Vehicular Traffic Load Balancing Scheme on VANETs

Similar documents
Simulation and Analysis of Transmission Range Effect on DSR Routing Protocol in a Vanet Network with Different Speed and Node Density

Performance Evaluation of Adaptive Control Channel Interval in VANET Based on Network Simulation Model

Analyzing Routing Protocols Performance in VANET Using p and g

PERFORMANCE EVALUATION OF DSDV, AODV ROUTING PROTOCOLS IN VANET

Performance Analysis of Wireless Mobile ad Hoc Network with Varying Transmission Power

PERFORMANCE ANALYSIS OF AODV ROUTING PROTOCOL IN MANETS

A Review of Reactive, Proactive & Hybrid Routing Protocols for Mobile Ad Hoc Network

SIMULATION BASED AND ANALYSIS OF ROUTING PROTOCOLS FOR VANET USING VANETMOBISIM AND NS-2

PERFORMANCE EVALUATION OF MOBILITY AND ROUTING PROTOCOLS FOR VEHICULAR AD HOC NETWORKS USING NS-2 AND VANETMOBISIM

Routing Protocols in Mobile Ad-Hoc Network

Behaviour of Routing Protocols of Mobile Adhoc Netwok with Increasing Number of Groups using Group Mobility Model

CS5984 Mobile Computing

Simulation and Analysis of AODV and DSDV Routing Protocols in Vehicular Adhoc Networks using Random Waypoint Mobility Model

Routing Protocols in MANETs

Performance of Ad-Hoc Network Routing Protocols in Different Network Sizes

International Journal of Advance Engineering and Research Development. Improved OLSR Protocol for VANET

3. Evaluation of Selected Tree and Mesh based Routing Protocols

Performance Evaluation of MANET through NS2 Simulation

Content. 1. Introduction. 2. The Ad-hoc On-Demand Distance Vector Algorithm. 3. Simulation and Results. 4. Future Work. 5.

IJMIE Volume 2, Issue 6 ISSN:

Performance Analysis of Aodv Protocol under Black Hole Attack

Unicast Routing in Mobile Ad-Hoc Networks

Analysis of GPS and Zone Based Vehicular Routing on Urban City Roads

Anil Saini Ph.D. Research Scholar Department of Comp. Sci. & Applns, India. Keywords AODV, CBR, DSDV, DSR, MANETs, PDF, Pause Time, Speed, Throughput.

Analysis of Routing Protocols over VANET

A Comparative Analysis of Energy Preservation Performance Metric for ERAODV, RAODV, AODV and DSDV Routing Protocols in MANET

Detection and Removal of Blackhole Attack Using Handshake Mechanism in MANET and VANET

Mobility and Density Aware AODV Protocol Extension for Mobile Adhoc Networks-MADA-AODV

Study on Indoor and Outdoor environment for Mobile Ad Hoc Network: Random Way point Mobility Model and Manhattan Mobility Model

Improving Energy and Efficiency in cluster based VANETs through AODV Protocol

Performance Evaluation Of Ad-Hoc On Demand Routing Protocol (AODV) Using NS-3 Simulator

Poonam kori et al. / International Journal on Computer Science and Engineering (IJCSE)

Performance Analysis of DSDV and ZRP Protocols with Mobility Variations in MANETs

ANALYSIS OF DIFFERENT REACTIVE, PROACTIVE & HYBRID ROUTING PROTOCOLS: A REVIEW

A Priority based Congestion Prevention Technique for Vehicular Ad-Hoc Networks

Figure 1: Ad-Hoc routing protocols.

A Survey of Vehicular Ad hoc Networks Routing Protocols

Performance Comparison of Mobility Generator C4R and MOVE using Optimized Link State Routing (OLSR)

Gateway Discovery Approaches Implementation and Performance Analysis in the Integrated Mobile Ad Hoc Network (MANET)-Internet Scenario

2013, IJARCSSE All Rights Reserved Page 85

1 Multipath Node-Disjoint Routing with Backup List Based on the AODV Protocol

Impact of Hello Interval on Performance of AODV Protocol

Performance Comparison of AODV, DSR, DSDV and OLSR MANET Routing Protocols

QoS Based Evaluation of Multipath Routing Protocols in Manets

Performance Comparison of AODV, DSDV and DSR Protocols in Mobile Networks using NS-2

Effect of Variable Bit Rate Traffic Models on the Energy Consumption in MANET Routing Protocols

Routing in Ad Hoc Wireless Networks PROF. MICHAEL TSAI / DR. KATE LIN 2014/05/14

Performance Enhancement of Routing Protocols for VANET With Variable Traffic Scenario

Analysis of Routing Protocols in MANETs

Performance Comparison of MANETs Routing Protocols for Dense and Sparse Topology

Considerable Detection of Black Hole Attack and Analyzing its Performance on AODV Routing Protocol in MANET (Mobile Ad Hoc Network)

INTERNATIONAL JOURNAL OF PURE AND APPLIED RESEARCH IN ENGINEERING AND TECHNOLOGY

Impact of Node Velocity and Density on Probabilistic Flooding and its Effectiveness in MANET

Security improvements Zone Routing Protocol in Mobile Ad Hoc Network

Experiment and Evaluation of a Mobile Ad Hoc Network with AODV Routing Protocol

Performance Enhancement of AOMDV with Energy Efficient Routing Based On Random Way Point Mobility Model

A REVERSE AND ENHANCED AODV ROUTING PROTOCOL FOR MANETS

Computation of Multiple Node Disjoint Paths

A Comparative study of On-Demand Data Delivery with Tables Driven and On-Demand Protocols for Mobile Ad-Hoc Network

An Adaptive Congestion Avoidance ACA_AODV Routing Protocol

AWERProcedia Information Technology & Computer Science

International Journal of Scientific & Engineering Research, Volume 4, Issue 9, September ISSN

Reliable Routing In VANET Using Cross Layer Approach

Performance Evaluation of Routing Protocols for Mobile Ad Hoc Networks

Analysis of Transmission Range Effect on Olsr Routing Protocol Vanet Simulation With Change in Node Speed and Node Density Against Qos Performance

Query Control Mechanisms for the Zone Routing Protocol (ZRP)

QoS Routing By Ad-Hoc on Demand Vector Routing Protocol for MANET

QUANTITATIVE ANALYSIS OF VANET ROUTING PROTOCOLS IN URBAN AND HIGHWAY SCENARIOS

Analysis of Black-Hole Attack in MANET using AODV Routing Protocol

Blackhole Attack Detection in Wireless Sensor Networks Using Support Vector Machine

A Literature survey on Improving AODV protocol through cross layer design in MANET

Performance Analysis of AODV Routing Protocol with and without Malicious Attack in Mobile Adhoc Networks

Optimizing Performance of Routing against Black Hole Attack in MANET using AODV Protocol Prerana A. Chaudhari 1 Vanaraj B.

Estimate the Routing Protocols for Internet of Things

Comparative Performance Analysis of AODV,DSR,DYMO,OLSR and ZRP Routing Protocols in MANET using Random Waypoint Mobility Model

CLASSIFICATION OF ROUTING Routing. Fig.1 Types of routing

PIONEER RESEARCH & DEVELOPMENT GROUP

COMPARITIVE ANALYSIS OF ROUTING PROTOCOLS IN MOBILE ADHOC NETWORKS

Varying Overhead Ad Hoc on Demand Vector Routing in Highly Mobile Ad Hoc Network

Performance Comparison of Routing Protocols for wrecked ship scenario under Random Waypoint Mobility Model for MANET

ENERGY EFFICIENT MULTIPATH ROUTING FOR MOBILE AD HOC NETWORKS

A Survey on Wireless Routing Protocols (AODV, DSR, DSDV)

Analysis QoS Parameters for Mobile Ad-Hoc Network Routing Protocols: Under Group Mobility Model

An Efficient Zone-Based Multicast Routing Protocol for Ad Hoc Network

Power aware Multi-path Routing Protocol for MANETS

AODV-PA: AODV with Path Accumulation

Performance Analysis of Routing Protocols in Mobile Ad-hoc Network (MANET)

Performance Analysis of Proactive and Reactive Routing Protocols for QOS in MANET through OLSR & AODV

IMPACT OF MOBILITY SPEED ON PROACTIVE AND REACTIVE ROUTING PROTOCOLS IN MOBILE ADHOC NETWORKS

A COMPARISON OF IMPROVED AODV ROUTING PROTOCOL BASED ON IEEE AND IEEE

Performance Evaluation of Routing Protocols for MAC Layer Models

Analysis and Simulations of Routing Protocols with Different Load Conditions of MANETs

A Location-based Predictive Route Caching Scheme for Pure Reactive Zone-based Routing Protocol in Mobile Ad Hoc Networks Abstract Introduction

Analysis of the Zone Routing Protocol

Evaluation of Ad-hoc Routing Protocols with. Different Mobility Models for Warfield. Scenarios

Performance Evaluation of AODV DSDV and OLSR Routing Protocols with Varying FTP Connections in MANET

POLITECNICO DI TORINO Repository ISTITUZIONALE

Keywords VANET, Routing protocols, AODV, DSR, DSDV.

Routing Protocols in MANET: Comparative Study

Simulation and Comparative Analysis of AODV, DSR, DSDV and OLSR Routing Protocol in MANET Abstract Keywords:

Transcription:

Performance Evaluation of AODV, DSDV, and ZRP Using Vehicular Traffic Load Balancing Scheme on VANETs Abdulqadir Muhtadi School of Electrical Engineering Telkom University Bandung, Indonesia koorust@students.telkomuniversity.ac.id Doan Perdana School of Electrical Engineering Telkom University Bandung, Indonesia doanperdana@telkomuniversity.ac.id Rendy Munadi School of Electrical Engineering Telkom University Bandung, Indonesia rendymunadi@telkomuniversity.ac.id Abstract Network topology in VANET environment is the nodes mobility model because the nodes connectivity is directly influenced by the nodes mobility. Drivers are now starting to use navigation systems that use vehicular load balancing scheme to find the fastest route to their desired destinations. We evaluate the effect of a load-balanced mobility model to the network performance in VANET environment. Three topology-based routing protocols that are AODV, DSDV, and ZRP, will be used. Furthermore, we analyze the network performance of each routing protocol on the designed mobility models. We use VanetMobiSim to generate the vehicles movements and Network Simulator 2 to simulate the data communications. We conclude that the network performances on the mobility model with load balancing scheme tend to decrease relatively to the network without load balancing scheme. The most suitable routing protocol for the designed mobility models is DSDV. Keywords-VANETs; mobility model; vehicular load balancing; AODV; DSDV; ZRP I. INTRODUCTION Vehicular Ad Hoc Network (VANET) has recently become one of the most research topics in the area of Intelligent Transportation System (ITS) and wireless networking [1]. Vehicular Ad Hoc Networks (VANETs) are self-organizing networks that work on intervehicle communication systems (IVC) and vehicle-to-infrastructure communication system [2]. VANETs is also a subclass of MANETs. VANETs is a promising approach to intelligent transportation system (ITS) [3]. One of the main problems of roads in big cities is traffic jam. Many factors contribute to this traffic jam problems such as the number of roads is too small compared to the number of road users, high busy hour vehicular traffic density, centralized traffic density, et cetera. One of the problems mentioned is centralized traffic density. This problem is mainly caused by most of the drivers usually take the shortest route to the destination instead of taking the fastest route to the destination. They simply do not know which road that provides the fastest travel time to their desired destination. To address this problem, many researchers are now researching and developing various vehicular traffic load balancing algorithms to balance the traffic load throughout the roads and to optimize the travel time. We can predict that in the future, most or even all of the cars will be connected to a network providing navigation services based on vehicular load balancing scheme so that the provided routes for the drivers are the fastest route. This prediction is based on the following considerations: (1) the increasing number of research about vehicular traffic load balancing scheme, (2) the increasing development of navigation application, (3) the more car manufacturers that install in-dash navigation systems in their cars, and (4) the growing development of vehicular networks. With the mentioned prediction before, we can also assume that the future mobility model of the cars will be a mobility model that is based on vehicular load balancing scheme. So it is important to do a research about VANETs performance on a mobility model by using a load balancing scheme. There are various mobility models as explained in [4] and [5]. The purpose of modeling a mobility model is to make a mobility model that can simulate a real life vehicular traffic. So that researches using mobility models, like VANETs researches, will have valid results. Next, the mobility models that have been designed before will be briefly explained. Freeway was first introduced in [6] and briefly described in [5]. This model illustrates a mobility model on a straight freeway road. The road has two directions, and each direction has more than one lane. Vehicle movements in this model are restricted to their lanes, so there is no random motion since vehicles are moving in a particular direction [5]. Manhattan was briefly explained in [5]. This model uses a grid-like map consisting of horizontal and vertical lines with intersections. Each road in this model has two traffic flow directions. Each car that reaches an intersection has a probability of.5 to move forward,.25 to turn right and.25 to turn left [5]. Manhattan mobility is commonly used in simulating urban areas. In this model the velocity of the nodes is restricted to the lanes and also there is velocity dependency between two nodes [7]. Integrated introduced in [5]. This model is an integration of various mobility models, which are Freeway mobility model, Manhattan mobility model, Stop Sign mobility model, and Traffic sign mobility model. In this model, there are urban areas that are presented similar to DOI 1.513/IJSSST.a.16.3.13 13. 1 ISSN: 1473-84x online, 1473-831 print

Manhattan mobility model and rural or suburban areas presented similar to Freeway mobility model. This model also features traffic lights and stop signs like those in Traffic Light mobility model and Stop Sign mobility model. Downtown explained in [4]. This model uses a similar map that Manhattan mobility model uses. But this model adds traffic density parameter to Manhattan model. In a real town, the traffic density is not uniformly distributed. There are zones with low traffic density and also zones with high traffic density. Zones with higher traffic density are usually in the downtown. Nodes inside the downtown zone are moving slower than the nodes outside the downtown zone. Hybrist, introduced in [8], is a mobility model that combines multiple existing mobility models into one mobility model. The reason of combining multiple mobility models is that in many practical scenarios, multiple models may exist within the same road network because of the heterogenity of nodes and user [8]. Hybrist combines the mobility of nodes at a road network with a high speed restricted middle lane that is reserved for public transportation and low speed heavy traffic unrestricted lanes. This model aims for accuracy in simulating multiple mobility cases such as the Jakarta road network that features Transjakarta Bus System. II. ROUTING PROTOCOLS We use topology-based routing protocols in this research. Topology based routing protocols use links information that exists in the network to perform packet forwarding [9]. We use three different types of topology-based routing protocols, they are DSDV proactive routing protocols, AODV reactive routing protocols, and ZRP hybrid routing protocols. A. Proactive routing protocol, DSDV Destination-Sequenced Distance Vector (DSDV) nodes maintain routes to every reachable DSDV nodes in the network. DSDV routing entry contains destination nodes, route metrics, next hop, and destination-stamped sequence numbers [1]. DSDV uses sequence numbers to distinguish stale routes from fresh ones [1]. This sequence number also prevent the formation of routing loops [11]. DSDV nodes send route update packets periodically to maintain the routing table entries. This transmission is also occurred when network topology changes are detected [11]. Every route with a higher sequence number is always preferred. When there are routes with the same sequence number, the route with better route metric will be preferred. One major advantage of DSDV is DSDV has a very low route setup delay since every DSDV nodes possess route availability to all destinations [1]. However, higher node mobility and a higher number of nodes will lead to an excessive amount of control traffic due to broken links occurred in the network [1]. This excessive amount of control traffic leads to a lower network throughput performance. B. Reactive routing protocol, AODV Ad-hoc on Demand Distance Vector (AODV) routing is an example of a reactive routing protocol. Nodes using reactive routing protocol do not maintain its routing table all the time. Instead, when a node needs to communicate with another node, it will flood a route request packet into the network to obtain the path to the destination. There are three different control packets in AODV, they are Route Request (RREQ), Route Reply (RREP) and Route Error (RERR). RREQ is initiated when a node needs to communicate with another node. Every node maintains two separate counters: node sequence number and broadcast_id [12]. The RREQ packet contains various parameters: source_addr, source_sequence_#, broadcast_id, dest_addr, dest_sequence_#, and hop_cnt [12]. The source_addr and broadcast_id pair indicates a packet as an RREQ packet [12]. The broadcast_id number is always incremented whenever a source node transmit a new RREQ packet [12]. Each node can respond the RREQ packet by transmitting RREP packet or rebroadcast it to its neighboring nodes after incrementing the hop count value. It is possible if a node receives the same RREQ packet that previously seen. The same broadcast_id and source_addr indicate the same RREQ packet [12]. If a node receives the same RREQ packet, it will not rebroadcast the packet and silently ignore it. Route reply packet is used to respond an RREQ packet. This packet is can only transmitted by the nodes that have the current route to the requested destination. These nodes can be the destination itself or a node that has the current route to the requested destination. RREQ packet travels along the reverse path. This reverse path is made as the RREQ packet travels through the nodes. Unlike the RREQ packet, RREP packet is sent in a unicast fashion back to the requesting node. As the RREP packet travels toward the source node, each node along the reverse path updates its routing information for the source node and the destination node stated in the RREP packet [12]. When a link breaks, RRER packet is sent to the affected source node whenever a packet tries to use the links [1]. C. Hybrid Protocol, ZRP Zone (ZRP) is an example of hybrid proactive/reactive protocol. ZRP limits the proactive procedure only to the node s local neighborhood while the search throughout the network is performed reactively by efficiently querying selected nodes in the network, as opposed to querying all the network nodes [13]. The ZRP maintains its routing zone using its proactive component called the Intrazone (IARP). IARP proactively maintain the routes for nodes inside the source node routing zone. ZRP uses the reactive Interzone (IERP) to acquire routes for nodes beyond the routing zone of the source node. IERP uses the information provided by IARP for the complete route between source node and destination node [1]. Figure 1 shows a routing zone of node S with a radius of 2 hops. Routing zone parameter is illustrated with a dotted circle around the node S. we can see that there are nodes which are exactly hops away from the node S. As we can see in Figure 1 nodes A-F are classified as interior nodes since they are located inside routing zone of the source node. While G- K nodes are peripheral nodes whose minimum distance to the source node is hops away. Since each node maintains its DOI 1.513/IJSSST.a.16.3.13 13. 2 ISSN: 1473-84x online, 1473-831 print

own routing zone, the zones of the neighboring nodes can overlap each other [13]. To make the map we use openstreetmap.org service to obtain the desired map area. One advantage of using openstreetmap.org is that it uses meter as the distance unit, the same as the VanetMobiSim 2. uses. So there is no need to convert the obtained coordinates. The first step of the mobility modeling is to decide which area to be used. As mentioned above the map is obtained from openstreetmap.org. We choose an area that lies in S 6.957 o S 6.916 o and E 17.6164 o E 17.6264 o as shown in Figure 2. Sometimes it is hard to modify the map directly from the osm files generated by openstreetmap.org by using VanetMobiSim 2. interface so that we use User Graph module of VanetMobiSim 2. to reconstruct the desired map. The result of this User Graph module is shown in Figure 3. Figure 1. A routing zone with radius of 2 hops [13] Each ZRP node maintains routing information only for the nodes inside its routing zone. Since the update packets are propagated locally within the routing zone, the amount of update traffic can be much smaller than it is in a pure reactive routing protocols. A node learns about its zone and maintains the routing information for its routing zone using IARP. The Interzone (IERP) is responsible for reactively discovering routes beyond routing zone of the source node [13]. Knowledge of the routing zone topology allows a node to efficiently continue the propagation of a query [13]. This is achieved by a packet delivery service, called border casting that allows a node to direct a message to its peripheral nodes [13]. Bordercast Resolution Protocol is used by ZRP to direct route request packet initiated by IERP to peripheral nodes thus eliminating excessive route request packet and increase network efficiency. To do so, BRP uses routing information provided by IARP to construct the bordercast tree [1]. III. SYSTEM MODELING A. Modeling The Mobility Subsystem In this research, two similar mobility models are modeled. The only difference between the two is the route determination algorithm. One model uses Dijkstra algorithm to decide the routes while the other one uses SpeedPathSelection algorithm which uses vehicular traffic load balancing scheme to determine the routes. Dijkstra algorithm selects the shortest path to destination while SpeedPathSelection algorithm does not only consider the length of the path but also the traffic congestion level and the road speed limit so that the fastest routes are preferred [2]. These two algorithms are used so that we can observe and analyze the effect the load-balanced mobility model to the VANETs performance. To make a valid and comprehensive mobility model, a mobility simulator that can simulate a real world traffic is needed. In this research, we use VanetMobiSim 2. to generate the desired mobility models. To make a mobility model that is as real as possible we do observe the corresponding road for some traffic-related data. Figure 2. Chosen area for the mobility model Figure 3. User Graph Module map result The next step is to determine the driving behavior of the drivers for the area chosen. Things that make this behavior are vehicle acceleration, vehicle deceleration, maximum speed, and route determination method. To design this behavior we use a built-in micro-mobility model provided by the VanetMobiSim 2. named Intelligent Driver Model with Intersection Management (IDM-IM). Inside IDM-IM, there are some parameters modeling mobility of the vehicles such as maximum speed, comfortable acceleration, deceleration, jam distance, and route determination method. DOI 1.513/IJSSST.a.16.3.13 13. 3 ISSN: 1473-84x online, 1473-831 print

As stated before the route determination algorithms used in this research are Dijkstra algorithm to illustrate the current traffic condition and SpeedPathSelection algorithm to illustrate the future load-balanced traffic condition. The maximum speed of the drivers is determined by Indonesia local regulation which is 5 kph. To determine vehicle density on a particular road we need to observe average vehicle speed and vehicle arrival rate. We observe the corresponding roads in order to obtain the accurate data. Another parameter that used to determine the behavior of drivers is the road length. Road length is calculated using the intersection coordinates of the map using Pythagoras theorem as (1) where is the distance between two points, is the difference between the two points on the -axis, and is the difference between the two points on the -axis. We use three different topology-based routing protocols that are proactive DSDV protocol, reactive AODV protocol, and hybrid ZRP protocol. The parameter is shown briefly in Table 2. B. Throughput and Packet Delivery Ratio Throughput is the average rate of successful data packets received at the destination [14]. Throughput is measured in bits per second. While packet delivery ratio is a ratio of successful number of packets received at the destination and the total packets sent by the source. Packet delivery ratio value will be similar to throughput value since they both measure the packet delivery success rate. Throughput is mathematically expressed as (2) where is the measurement duration. There are three roads to observe: R.E. Martadinata Road, Ambon Road, and Aceh Road. The observation results are shown in Table 1. The parameters that we put on VanetMobiSim 2. for each road are vehicle count and maximum allowed speed which is 5 kph. TABLE I. ROAD OBSERVATION RESULT Road Road Length Traffic Density Vehicle count R.E. Martadinata 121 m 14 vehicles/km 127 vehicles Ambon 553 m 33 vehicles/km 19 vehicles Aceh 11 m 43 vehicles/km 44 vehicles The observation aims to obtain node counts to be inserted into VanetMobiSim simulation parameters. The result of the observation shows that there are 19 nodes in total. This node count is then used for the parameter value in the simulation. IV. SIMULATION AND DISCUSSION A. Simulation Parameters The simulation is conducted using Network Simulator 2.35 (NS2) software. We install NS2 inside VirtualBox which installed in Windows 7 64 bits. TABLE II. SIMULATION PARAMETERS Operating System Ubuntu 12.4 32 bit on VirtualBox on Windows 7 64 bit Network Simulator NS 2.35 Generator VanetMobiSim 2. DSDV, AODV, ZRP Micro IDM-IM Dijkstra (without load balancing Route Determination Algorithm scheme), SpeedPath (with load balancing scheme) Number of nodes 19 Wave Propagation Model Two Ray Ground Mac Type IEEE 82.11p Antenna Type Omnidirectional Packet Queue Type Droptail Simulation Duration 8 seconds Data Traffic Type CBR. 512 kbits packet size, 2 Mbps rate Let us say the mobility model using Dijkstra path searching algorithm which does not use any load balancing mechanism called as Dijkstra or the Regular, and the mobility model using SpeedPathSelection path searching algorithm which uses load balancing mechanism called as SpeedPathSelection or the Load-Balanced Mobility Model. Packet delivery ratio is expressed as (3) where is number of packets successfully received, and is number of packets transmitted. Packet Delivery Ratio Figure 4. Packet delivery ratio on Regular Packet Delivery Ratio.6.4.2.6.4.2.51194.49686.2439.36798.314.17494 Figure 5. Packet delivery ratio on Load-Balanced DOI 1.513/IJSSST.a.16.3.13 13. 4 ISSN: 1473-84x online, 1473-831 print

As we can see from Figure 6 and Figure 7 AODV has the highest throughput that are 212.71 kbps on the Regular and 152.5 kbps on the Load-Balanced. DSDV has an average throughput of 26.38 kbps on Regular and 152.59 kbps on Load- Balanced. While ZRP has the lowest average throughput that are 11.5 kbps on Regular and 72.91 kbps on Load-Balanced. AODV is a reactive protocol so that it does not need to periodically update its routing table, but it updates its routing table when the source node wants to transmit a packet which the destination address is not in the routing table yet instead. That promotes to higher network throughput since there is no additional control traffic from periodic routing update. It is stated in [15] that as the number of the node increases the AODV network throughput is also increasing. AODV has higher throughput in Dijkstra mobility model than it has in SpeedPathSelection mobility model. This is caused by the nodes in SpeedPathSelection are not concentrated in the three main roads, R.E. Martadinata Road, Aceh Road, and Ambon Road, but are distributed due to the load balancing mechanism. This causes more intervehicle distance and faster node velocity. Thus, topology changes are more likely to occur so that route discovery process need to be executed more often. Network Throughput (kbps) Figure 6. Network throughput on Regular Network Throughput (kbps) 25 2 15 1 25 2 15 1 5 5 212.718 26.368 152.594 128.252 11.5 72.91 Figure 7. Network throughput on Load-Balanced DSDV is a proactive protocol so that every node maintains fresh route information to all reachable destinations. To maintain those route information DSDV node periodically update its routing table by transmitting routing update packet which is also executed when topology changes are detected. This routing update process generates extra traffic to the network so then DSDV network has lower throughput than the network with reactive protocol. DSDV also has higher throughput in Dijkstra mobility model than it has in SpeedPathSelection mobility model. This is due to the rapid topology changes in SpeedPathSelection mobility model so that DSDV nodes have to update its routing table more frequently that promote to higher control traffic in the network. This is worsened by the periodic routing update done by each node. ZRP has a very low network throughput which is also the lowest among the three protocols used. ZRP has low throughput in a network with a large number of nodes and in a high mobility network. High mobility of the nodes causes routes to be broken so that many packets are dropped. Besides, with increasing number of nodes, the route discovery process becomes more complex and additional control traffic increases [1]. When the route discovery packet leaves its original routing zone (source node s routing zone), because of the routing zones are heavily overlap, a node can be a member of many routing zones [13]. It is very possible that the route query will be forwarded to all the network nodes, effectively flooding the network [13]. But a more disappointing result is that the IERP can result in much more traffic than the flooding itself, due to the fact that bordercasting involves sending the query along a path equal to the zone radius [13]. C. End-to-End Delay End-to-end delay is an interval between the packet transmission and the time when the packet arrives. This delay includes propagation delay, queue delay, and route discovery duration [15]. This delay value is a very crucial parameter since there will be many critical safety applications propagated among the nodes in VANETs which requires low a delay. An average end-to-end delay is expressed as (4) where is average end-to-end delay, is packet reception time, is packet transmission time, and is number of packets received. In Figure 8 and Figure 9 we can see that AODV has the highest average delay which are 416.9 ms on Regular and 374.9 ms on Load-Balanced Mobility Model. ZRP has average delays of 92.8 ms on Regular and 34.58 ms on Load-Balanced Mobility Model. While DSDV has the lowest average delay that are 13.58 ms on Regular and 19.22 ms on Load- Balanced. AODV has the highest delay. This is due to the fact that reactive protocol does not maintain fresh routing information to the nodes in the network, but it sends a route discovery packet when it needs to communicate with another node instead. This route discovery process generates additional DOI 1.513/IJSSST.a.16.3.13 13. 5 ISSN: 1473-84x online, 1473-831 print

delay in packet transmission. That is one major weakness of reactive routing protocol. End to End Delay (ms) End toend Delay (ms) 5 4 3 2 1 Figure 8. Delay on Regular 4 3 2 1 416.936 374.9822 13.58572 19.22858 34.58834 92.85796 Figure 9. Delay on Load-Balanced On the other hand, DSDV has a very low delay and also has the lowest delay among the three protocols used. DSDV is a proactive routing protocol that maintain routing information to all reachable nodes in the network in its routing table. So that DSDV node does not need to execute route discovery process before packet transmission since it has the current routing information in its routing table. ZRP has a little bit higher delay than the DSDV has since ZRP is a hybrid routing protocol that maintains current routing information to the neighbor nodes inside the routing zone. The more the communicating nodes are in the same neighborhood, the lower the delay will be since the routing information is maintained by the proactive IARP. But when the destination node is not inside the source node s routing zone, it will use its reactive part to discover the path to the destination node. By using this, the source node will execute route discovery process which will generate additional delay. D. Performance Tendency By using different mobility model, the routing protocol performances that work on the network will also be different. There is protocol whose performance tends to increase and also there is protocol whose performance tends to decrease. As we can see in Figure 1 that all the three routing protocols performances tend to decrease if they are applied in a load-balanced mobility model compared to the mobility model without load balancing scheme. DSDV average throughput decrement is 37.8% while AODV and ZRP are 28.2% and 28.16% respectively. Throughput decrement experienced by AODV and ZRP are not as high as DSDV throughput decrement is. This is due to the reactive property of AODV and ZRP so that there is less additional control traffic generated compared to the control traffic generated in DSDV. Throughput (kbps) End to End Delay (ms) 25 2 15 1 5 5 4 3 2 1 AODV DSDV ZRP Figure 1. Throughput tendencies AODV DSDV ZRP Figure 11. Delay tendencies Regular Load Balanced Regular Load Balanced As explained in Section III.B that AODV has the highest delay, ZRP has a lower delay, and DSDV has the lowest delay. But in terms of tendencies, AODV and ZRP delay performance tend to increase while DSDV delay performance tends to decrease. Delay decrement in AODV and ZRP are 9.8% and 62.7% respectively while DSDV experiences delay increment of 4.15% relative to the delay in the mobility model without load balancing mechanism. The delay decrement in AODV and ZRP network are due to the node s position in the load-balanced mobility model are not concentrated in a certain area, but are distributed instead to achieve the fastest node travel time. This distributed node position causes the route discovery packet to travel along the network faster so that it can cover the network geographically faster. ZRP has the highest delay decrement due to the fact that route discovery process in ZRP is done by per hops basis and also the route availability checking done by the discovery packet receiver is for the nodes inside the receiver s routing zone. On the other hand AODV nodes send a route request packet in a hop by hop fashion and the route availability checking by the request packet receiver is done for the neighboring nodes only. DOI 1.513/IJSSST.a.16.3.13 13. 6 ISSN: 1473-84x online, 1473-831 print

Delay increment occurred in DSDV network is caused by the rapid topology changes in the load-balanced mobility model so that routing update process is occurred more frequent. V. CONCLUSION We conclude that mobility model without load balancing scheme can be modeled using Dijkstra route determination algorithm. While mobility model with load balancing scheme can be modeled using SpeedPathSelection route determination algorithm. Mobility models affect the ad hoc network performance since mobility model is the network topology to an ad-hoc network. The three topology-based routing protocols analyzed in this research experience network performance decrement in the load-balanced mobility model relative to the network performance in the mobility model without load balancing mechanism. Throughput decrement of DSDV is 37.8%, AODV is 28.2%, while ZRP is 28.16%. From the results we can see that the most suitable routing protocol for the both mobility models modelled, mobility model with load balancing mechanism and mobility model without load balancing mechanism, is DSDV since it has a very low delay and a reliable throughput. REFERENCES [1] D. Perdana and R.F. Sari, Enhancing Channel Coordination Scheme Caused by Corrupted Nakagami Signal and s on the IEEE 169.4 Standard, Journal of Networks, Vol. 9, No. 12, December 214 [2] J. Harri, M. Fiore, F. Filali, and C. Bonnet, Vehicular Mobility Simulation with VanetMobiSim, Trans. of Society for Modeling and Simulation, September 29 [3] S. Yousefi, M.S. Mousavi, and M. Fathy, Vehicular Ad Hoc Networks (VANETs): Challenges and Perspectives, 26 6th International Conference on ITS Telecommunications Proceedings [4] F.J. Martinez, J.C. Cano, C.T. Calafate, and P. Manzoni, CityMob: a mobility model pattern generator for VANETs, 28 ICC Workshop. Proceedings [5] M. Alam, M. Sher, and S.A. Husain, Integrated (IMM) for VANETs Simulation and Its Impact. International Conference in Emerging Technologies, 29 Internation Conference on Emerging Technology proceedings, pp 452-455 [6] F. Bai, N. Sadagopan, and A. Helmy, The Important Framework for Analyzing The Impact of Mobility on Performance of Routing Protocols for Adhoc Networks, Elsevier Ad Hoc Networks vol. 1 (23), pp. 383-43, 23 [7] D. Perdana and R.F. Sari, "s Performance Analysis using Random Dijkstra Algorithm and Doppler Effect for IEEE 169.4 Standard, International Journal of Simulation, Systems, Science, and Technology, United Kingdom Simulation Society [8] V.M. Danquah and D.T. Altilar, HYBRIST A Novel Hybrid for VANET Simulation, Internation Journal of Computer Applications, Vol. 86, No. 14, January 214 [9] U. Nagaraj, M.U. Kharat, and P. Dhamal, Study of Various Routing Protocols in VANET, International Journal of Computer Science and Technology (IJCST), Vol.2, Issue 4, 211 [1] Z.Y. Noorani, Performance Analysis of DSDV, AODV, and ZRP of MANET and Enhancement in ZRP to Improve Its Throughput, International Journal of Scientific and Research Publications, Volume 3, Issue 6, 213 [11] G. He, Destination-Sequenced Distance Vector (DSDV) Protocol, Technical report, Helsinki University of Technology, Finland [12] C.E. Perkins, E.M. Royer, Ad-hoc On-Demand Distance Vector Routing, February 1999, Proc. 2nd IEEE Workshop on Mobile Computer Systems and Applications, pp. 9-1 [13] M.R. Pearlman and Z.J. Haas, Determining the Optimal Configuration for the Zone, IEEE Journal on Selected Areas in Communication, Vol. 17, No. 8, August 1999 [14] S. Kaur and S. Kaur, Analysis of Zone in MANET International Journal of Research in Engineering and Technology, Volume: 2 Issue: 9, September 213 [15] A. Singh and A.K. Verma, Simulation and Analysis of AODV, DSDV, ZRP in VANET, International Journal in Foundations of Computer Science & Technology (IJFCST), Vol.3, No.5, September 213 Abdulqadir Muhtadi was born in Bandung, Indonesia, on October 29, 1992. He is currently studying at Telkom University majoring on Telecommunication Engineering. His research interest is network engineering. His current research is about vehicular ad-hoc networks. Doan Perdana received his BSc and MSc degrees in Telecommunication Engineering, from the Institute of Technology Telkom in 24 and 212, respectively. He completed his PhD in Electrical Engineering Department, University of Indonesia. His interests include Telecommunication Systems and Computer Engineering. Rendy Munadi received his Doctor in Telecommunication Engineering from University of Indonesia. He is now a senior lecturer of Telkom University in Bandung, Indonesia. He is currently the Head of the Expertise in Networks and Multimedia of Telkom University. He has served as the program committee on several conferences and as reviewer of papers. His current research are in the area of next generation network and new generation network, Wireless Network, Wireless Sensor Network, IMS, IP/MPLS Network, Routing Management and protocols/interface Next Generation Network. DOI 1.513/IJSSST.a.16.3.13 13. 7 ISSN: 1473-84x online, 1473-831 print