06-08-2014, 12:19 PM
An Inter-Domain Routing for Heterogeneous Mobile Ad Hoc Networks Using Packet Conversion
[attachment=66639]
ABSTRACT
In ad hoc networks, a diversity of routing protocols exists. Each network selects a routing protocol suitable for its own environment such as a vehicular ad hoc network (VANET), a wireless mesh network, or a mobile ad hoc network (MANET) that consists of pedestrians. Generally, since there is no interoperability between different routing protocols in ad hoc networks, the communication between different networks is impossible. Therefore, in this paper, we propose Ad hoc Traversal Routing (ATR) to provide interoperability between different networks. Because of ATR, any two nodes in different networks can seamlessly communicate with each other. ATR connects two different networks to each other by converting control messages from one network to another network and adding the node address of different networks into the routing table for routing protocols. In addition, we conduct simulation experiments to evaluate the performance of ATR in heterogeneous wireless network environment that consists of a vehicle ad hoc network, a wireless mesh networks, and a mobile ad hoc network.
INTRODUCTION
For ad hoc networks, a diversity of routing protocols exists. However, since a suitable routing protocol is different in its own environment such as a vehicular ad hoc network (VANET), a wireless mesh network, or a mobile ad hoc network (MANET), there is no all-around routing protocol working effectively in any network environments. In urban area, many types of ad hoc networks such as VANETs and MANETs that consist of pedestrians exist. Since many routing protocols for vehicle-to-vehicle, vehicle-to-roadside, and person-to-person communication have been proposed, it is easy to realize the communication in each environment. However, for practical use of ad hoc networks, the communication among nodes in heterogeneous ad hoc networks should be realized.
For instance, one of the applications is a person-to-vehicle communication. In this case, different routing protocols are performed for the networks that persons and vehicles belong to, and the inter-domain routing protocol is required to realize the person-to-vehicle communication. In the Internet, a Border Gateway Protocol (BGP) is utilized for the communication between heterogeneous network systems. However, in ad hoc network environment, each network has very high dynamic network topology nature. In addition, nodes as well as networks are newly appeared or disappeared in some cases. Therefore, it is difficult to introduce BGP to communicate
between different ad hoc networks in heterogeneous ad hoc network environment.
Ad hoc Traversal Routing (ATR) is used to provide the interoperability between different networks as a gateway protocol in heterogeneous ad hoc network environment. ATR connects two different networks to each other by converting control messages from one network to another network and adding the node address of different networks into the routing table for routing protocols. For inter-domain routing protocol in mobile ad hoc networks, Cluster-based Inter-Domain Routing (CIDR) and Inter-Domain Routing for MANETs (IDRM) have been proposed.
However, intrarouting protocol in each network is specified and it is difficult
to select the local routing protocol suitable for each local network environment. On the contrary, in the proposed scheme, it is possible to select the suitable routing protocol for each local network environment such as AODV and DSDV. In such
heterogeneous mobile ad hoc network environment, ATR can provide the interoperability between different networks.
AD HOC TRAVERSAL ROUTING
As shown in Figure 1, nodes A and B are in Network 1 and perform routing protocol R1, while nodes C and D are in Network 2 and perform routing protocol R2. In this heterogeneous mobile ad hoc network environment, node A cannot communicate with node D because different routing protocols are performed in nodes A and D, and any packets cannot be forwarded from node B to node C. Therefore, we propose Ad hoc Traversal Routing (ATR) that provides the interoperability between heterogeneous mobile ad hoc networks. In case of Figure 1, ATR that is installed on nodes B and C mediates the connection between nodes B and C. In each node, ATR communicates with a routing protocol such as AODV or OLSR used in the node. We define such a routing protocol as a local routing protocol. The mechanism that ATR
communicates with a local routing protocol in each node is different according to the type of local routing protocols like reactive or proactive. In case that a local routing protocol is reactive, ATR uses the packet conversion mechanism, while in case that a local routing protocol is proactive, ATR uses the address sharing mechanism. In addition, in case that a local routing protocol is a location-based routing protocol, ATR uses the mechanism to convert from location information to hop number. Hereinafter, we call nodes that ATR is installed as ATR nodes.
Mechanism for Reactive Routing Protocol
In reactive routing protocols, route request messages are used to construct the route between a source node and a destination node and communicate with each other. In heterogeneous mobile ad hoc network environment, in case that the source node and the destination node are in different networks, route request messages sent by the source node are not received by the nodes in the network that the destination node
belongs to. As a result, the route between the source node and the destination node in different networks are not constructed. Therefore, in the proposed scheme, ATR node converts from the route request messages to new route request messages for the different network to forward route request messages to the neighboring ATR node in the different network. Figures 2 and 3 show the procedure to handle route request
messages in ATR node
ATR Route Repair
The route is repaired by the local routing protocol in each network hen the route break occurs. Therefore, ATR nodes repair the route between ATR nodes. If the route between ATR nodes breaks, the ATR node broadcasts ATR RREQ message with TTL = 2 to the neighboring ATR nodes to discover new route. When ATR node that has the route to the destination ATR node receives ATR RREQ message, the ATR node sends ATR RREP message to the source ATR node. In case that ATR nodes cannot repair the route, the ATR node sends ATR RERR (ATR Route ERRor) message with the destination ATR node back to the source ATR node. In this time, ATR nodes that receive the ATR RERR message and have the route entry to the destination ATR node delete the route entry in the routing table and then forward it to the source ATR node. In case that the local routing protocol has the function of the control message that is the same as ATR RERR, ATR nodes convert from ATR RERR messages to the corresponding control messages and then deletes the route entry in the routing table of the local routing protocol.
AD HOC NETWORK ROUTING PROTOCOLS
The behaviors of AODV, DSDV and GPSR that we use in simulation evaluation. The ad hoc routing protocols are classified into a proactive type and a reactive type. AODV and GPSR are a reactive routing protocol, while DSDV is a proactive routing protocol.
SIMULATION EVALUATION
Simulation Experiment I
The experiments are conducted to evaluate the performance of the proposed scheme in heterogeneous mobile ad hoc network environment where there are the communications for pedestrian-to-vehicle, vehicle-to-roadside, and pedestrian-toroadside. In the simulation experiments, the evaluation criteria are the delivered data packet ratio, the overhead, and the delay time.
1) Simulation Environment:
Tables I and II show the simulation environment and network environment. Figure 6 shows the simulation field map. Fixed nodes are deployed at the 200 meter interval in 10 × 10 grid structure, while the other nodes are randomly deployed. Pedestrian1 and Pedestrian2 nodes randomly move around in Area 1 and Area 2, while Vehicle nodes randomly move on the road. Although Pedestrian1 and Pedestrian2 nodes use AODV[7], the simulator handles two different types of routing protocols and Pedestrian1 node cannot directly communicate with Pedestrian2 node. In each communication pattern, data packets are forwarded from a source node to a destination, and then they are sent back from the destination node to the source node. Simulation re
Simulation Results
) Delivered Data Packet Ratio:
Figures 7 and 8 show the data packet delivered ratio of one way and two way communication from the source node to the destination node, respectively. x and y axis denote the ratio of ATR nodes and the data packet delivered ratio. As shown in these figures, Pedestrian1 nodes and fixed nodes provide the stable communication regardless of the ratio of ATR nodes because Pedestrian1 nodes become the high density in the limited area and fixed nodes do not move. On the other hand, Pedestrian2 nodes and vehicle nodes provide the unstable communication because in both cases the node density is not so high and nodes’ movable area becomes larger. As a result, the data packet delivered ratio of Pedestrian2 node to vehicle node and fixed node becomes higher than that of Perdestrian1 to vehicle node and fixe node. However, along with the increase of the ATR node’s ratio, the packet delivered ratio becomes lower due to the overhead of control packets sent by ATR nodes. In this simulation environment, since there is the mesh network consisting of fixed nodes, it is expected that the proposed scheme provides high data packet delivered ratio in any case by forwarding data packets over the mesh network. However, since DSDV protocol is performed in the mesh network, the data packets cannot be forwarded in case that the address of the destination node is not included in the routing table. Therefore, since ATR nodes in the mesh network cannot obtain the address of the destination node in many cases, the proposed scheme cannot utilize the mesh network effectively to forward data packets. The other reason why the data packet delivered ratio becomes lower is that in the mesh network, ATR nodes are randomly deployed. It is expected that the data packet delivered ratio is improved by considering the deployment of the ATR nodes in the mesh network.
Conclusion
In order to evaluate the performance of the proposed scheme, we conducted the simulation experiments in the heterogeneous ad hoc network environment consisting of vehicle nodes, pedestrian nodes, and fixed nodes and confirmed that any two nodes in different networks can communicate with each other through ATR nodes with respect to the data packet delivered ratio and delay time. In the future work, in the proposed scheme, although all ATR nodes share the node address information in all networks with each other, many nodes do not become the destination node that provides any services in the network. As a result, the high overhead is incurred to exchange the node address information among ATR nodes. Therefore, only the node address information of the destination nodes that provide services has to be efficiently shared among ATR nodes in different networks. We are planning to introduce the service dissemination mechanism for heterogeneous mobile ad hoc network environment.