Gateway Forwarding Strategies in Ad hoc Networks

Similar documents
Gateway Forwarding Strategies for Ad hoc Networks

Investigating Race Conditions in Multi-Homed On Demand Ad Hoc Networks

Enhancing the Performance of Mobile Ad Hoc Networks with the Aid of Internet Gateways 1

AODV-PA: AODV with Path Accumulation

Virtual Hierarchical Architecture Integrating Mobile IPv6 and MANETs for Internet Connectivity

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

A COMPARISON OF REACTIVE ROUTING PROTOCOLS DSR, AODV AND TORA IN MANET

Routing Protocols in MANET: Comparative Study

Hybrid gateway advertisement scheme for connecting mobile ad hoc networks to the Internet

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

Mieso K. Denko* and Chen Wei

Performance Analysis of Three Routing Protocols for Varying MANET Size

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

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

Routing Protocols in MANETs

A NOVEL APPROACH OF AODV FOR STABILITY AND ENERGY EFFICIENT ROUTING FOR MANET USING IPV6

Performance of Internet Access Solutions in Mobile Ad Hoc Networks

Performance Comparison of Ad Hoc Routing Protocols over IEEE DCF and TDMA MAC Layer Protocols

Charles Perkins Nokia Research Center 2 July Mobility Support in IPv6 <draft-ietf-mobileip-ipv6-14.txt> Status of This Memo

Survey on Techniques providing Internet Connectivity to Mobile Ad Hoc Networks

ROUTE STABILITY MODEL FOR DSR IN WIRELESS ADHOC NETWORKS

Extending Global IP Connectivity for Ad Hoc Networks 1

Mobile Ad Hoc Networking Working Group

An Approach to Efficient and Reliable design in Hierarchical Mobile IPv6

A New Energy-Aware Routing Protocol for. Improving Path Stability in Ad-hoc Networks

Performance Comparison of MANETs Routing Protocols for Dense and Sparse Topology

PERFORMANCE EVALUATION OF DSR USING A NOVEL APPROACH

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

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

COMPARATIVE STUDY AND ANALYSIS OF AODTPRR WITH DSR, DSDV AND AODV FOR MOBILE AD HOC NETWORK

Computation of Multiple Node Disjoint Paths

A Graph-based Approach to Compute Multiple Paths in Mobile Ad Hoc Networks

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

Mobile & Wireless Networking. Lecture 10: Mobile Transport Layer & Ad Hoc Networks. [Schiller, Section 8.3 & Section 9] [Reader, Part 8]

Evaluation of Routing Protocols for Mobile Ad hoc Networks

A SURVEY OF ROUTING PROTOCOLS IN MOBILE AD HOC NETWORKS

Figure 1: Ad-Hoc routing protocols.

Iit Istituto di Informatica e Telematica

Expanding Ring Search for Route Discovery in LOADng Routing Protocol

A Performance Comparison of Multi-Hop Wireless Ad Hoc Network Routing Protocols

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

Analysis of TCP and UDP Traffic in MANETs. Thomas D. Dyer Rajendra V. Boppana CS Department UT San Antonio

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

Throughput Analysis of Many to One Multihop Wireless Mesh Ad hoc Network

Impact of Hello Interval on Performance of AODV Protocol

Probabilistic Mechanism to Avoid Broadcast Storm Problem in MANETS

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

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

Mobile Ad-hoc Network. WIDE project/keio University

A Performance Comparison of Multi-Hop Wireless Ad Hoc Network Routing Protocols. Broch et al Presented by Brian Card

A Novel Rebroadcast Technique for Reducing Routing Overhead In Mobile Ad Hoc Networks

PERFORMANCE ANALYSIS OF AODV ROUTING PROTOCOL IN MANETS

2013, IJARCSSE All Rights Reserved Page 85

Aanchal Walia #1, Pushparaj Pal *2

Investigation on OLSR Routing Protocol Efficiency

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

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

Simulation & Performance Analysis of Mobile Ad-Hoc Network Routing Protocol

A Routing Protocol for Utilizing Multiple Channels in Multi-Hop Wireless Networks with a Single Transceiver

Measure of Impact of Node Misbehavior in Ad Hoc Routing: A Comparative Approach

STUDY AND COMPARISION OF PROACTIVE AND REACTIVE ROUTING PROTOCOL FOR MULTICHANNEL WIRELESS AD-HOC NETWORK

MEWLANA-Mobile IP Enriched Wireless Local Area Network Architecture

Mobile SCTP for IP Mobility Support in All-IP Networks

Research Paper GNANAMANOHARAN ET AL., INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING TECHNOLOGY E-ISSN

Security Scheme for Malicious Node Detection in Mobile Ad Hoc Networks

Performance Analysis of Aodv Protocol under Black Hole Attack

Variable Length and Dynamic Addressing for Mobile Ad Hoc Networks

3. Evaluation of Selected Tree and Mesh based Routing Protocols

A Highly Effective and Efficient Route Discovery & Maintenance in DSR

Power aware Multi-path Routing Protocol for MANETS

A Performance Comparison of MDSDV with AODV and DSDV Routing Protocols

Performance Evaluation of MANET through NS2 Simulation

DYNAMIC ROUTES THROUGH VIRTUAL PATHS ROUTING FOR AD HOC NETWORKS

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

[Kamboj* et al., 5(9): September, 2016] ISSN: IC Value: 3.00 Impact Factor: 4.116

An Extensive Simulation Analysis of AODV Protocol with IEEE MAC for Chain Topology in MANET

AWERProcedia Information Technology & Computer Science

Volume 3, Issue 6, June 2015 International Journal of Advance Research in Computer Science and Management Studies

LECTURE 9. Ad hoc Networks and Routing

Efficient On-Demand Routing for Mobile Ad-Hoc Wireless Access Networks

LECTURE 8. Mobile IP

Implementing Ad Hoc to Terrestrial Network Gateways

Comparative Study of Mobility Models using MANET Routing Protocols under TCP and CBR Traffic

Performance Analysis of MANET Routing Protocols OLSR and AODV

Performance evaluation of reactive and proactive routing protocol in IEEE ad hoc network

Integrating WLANs & MANETs to the IPv6 based Internet

Comparison of proposed path selection protocols for IEEE s WLAN mesh networks

Name Resolution in Mobile Ad-hoc Networks

Study of Route Reconstruction Mechanism in DSDV Based Routing Protocols

Security Enhancement of AODV Protocol for Mobile Ad hoc Network

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

Performance Analysis and Enhancement of Routing Protocol in Manet

Simulation and Performance Analysis of Throughput and Delay on Varying Time and Number of Nodes in MANET

ENERGY-AWARE FOR DH-AODV ROUTING PROTOCOL IN WIRELESS MESH NETWORK

A Reliable Route Selection Algorithm Using Global Positioning Systems in Mobile Ad-hoc Networks

Mobile Communications Chapter 8: Network Protocols/Mobile IP

A Study of Bellman-Ford, DSR and WRP Routing Protocols with Respect to Performance Parameters for Different Number of Nodes

CMPE 257: Wireless and Mobile Networking

Performance Analysis of DSR Routing Protocol With and Without the Presence of Various Attacks in MANET

Lecture 16: Wireless Networks

Transcription:

Gateway Forwarding Strategies in Ad hoc Networks Erik Nordström Department of Information Technology Uppsala University erikn@it.uu.se Per Gunningberg Department of Information Technology Uppsala University perg@it.uu.se Christian Tschudin Computer Science Department University of Basel christian.tschudin@unibas.ch ABSTRACT This paper studies the efficient integration of gateway forwarding strategies in wireless ad hoc networks using Mobile IP. The problem to solve is the forwarding to one or more gateways in an environment where there is no hierarchical addressing. We compare the properties of two proposed forwarding strategies; traditional default routes and tunneling. We find that default route forwarding will not operate efficiently in a multi-hop environment and that it will, without modifications, operate incorrectly with multiple gateways. On the other hand, we find tunnel forwarding to be architecturally appealing with many properties that make it a suitable forwarding strategy with multiple gateways. 1. INTRODUCTION In multi-hop ad hoc networks, it is commonly suggested that Mobile IP [8] is used to provide Internet connectivity [5, 2, 10, 12]. Visiting nodes will keep their home addresses when joining the ad hoc network. Thus, the network will contain a mix of prefixes and will have flat addressing. In this setting, Mobile IP provides topologically correct addressing for nodes in the ad hoc network by hiding topologically incorrect addresses behind one or more care-of addresses. This allows nodes to keep their home network addresses when visiting a foreign network, while still enjoying full Internet connectivity. In other words, Mobile IP solves the problem of how to route to addresses that are not located at the correct place in the Internet s hierarchy. However, in the ad hoc scenario, Mobile IP only solves half of the problem - how to get traffic into the ad hoc network. Similar to how Internet routing without Mobile IP can not route to addresses in the ad hoc network, some ad hoc routing protocols have problems resolving routes to external addresses (e.g., Internet addresses), that are not part of the ad hoc network. These protocols belong to the reactive or on-demand class, e.g., AODV [9] and DSR [4]. In an ad hoc network with flat addressing, these protocols can only configure routes to nodes that are searchable within the network s boundaries. In contrast, a proactive protocol can determine which destinations are external because it maintains a complete list of nodes in the ad hoc network. In this paper we study how to integrate gateway forwarding with reactive routing protocols. More specifically, how to efficiently route packets, destined for the Internet, through the ad hoc network. We look at two proposed strategies; default routes and tunneling. We find that default routes that are adopted from traditional LAN settings need modifications to work in a multi-hop ad hoc environment. Despite these additions, default routes have problems with multiple gateways and inconsistent routing state. Establishing tunnels to the gateways, on the other hand, provides an architecturally appealing solution and works well with multiple gateways. The rest of the paper is structured as follows. In section 2 we describe our assumptions and the challenges with forwarding to a gateway using existing approaches and suggest modifications for ad hoc networks. In section 3 we introduce unidirectional tunneling to gateways. Section 4 reports on results from evaluating default routes and tunneling in simulation. In section 5 we discuss related work, while section 6 concludes the paper. 2. FORWARDING TO GATEWAYS IN AD HOC NETWORKS Suppose that a node in the ad hoc network wants to send a packet to an Internet host (Figure 1 shows an ad hoc network with flat addressing and forwarding through a gateway). First the source node needs to find out the location 130.238.12.1 HA 192.168.6.1 10.0.4.2 MN 66.35.250.151 Internet MIP Tunnel 142.67.8.245 130.238.12.5 63.3.5.23 130.238.8.1 FA 192.168.1.1 65.43.32.1 Figure 1: A wireless ad hoc scenario with random addresses and two gateways, one Mobile IP Home Agent (HA) and one Foreign Agent (FA). Mobile IP s reverse tunneling makes sure that the mobile node s (MN) return traffic is sent to the correct gateway (via its HA), although it uses a topologically incorrect home network address when sending traffic through the FA.

of the destination, i.e., whether it is currently in the ad hoc network or a host on the Internet. Thereafter it needs to find a route to a gateway. With our assumption that the ad hoc network uses flat addressing, a reactive protocol would initiate a network wide search for the destination, but there would be no reply. This could indicate that the destination can be found in the Internet. However, a more efficient approach is suggested by Broch et al. [3]. They propose that the gateway can, in response to a route request, send a proxy route reply to signal that it can take care of the forwarding to the requested destination. A proxy route reply from a gateway may be used to configure host routes for Internet destinations [2]. This would be in line with how reactive protocols work - one routing table entry is created for each destination. The downside is that routing tables will grow and timers and state must be maintained independently. In a protocol like AODV, sequence numbers for these Internet hosts must be managed by the gateway to avoid loops. 2.1 Default Routes When requests for fixed Internet destinations are common, it is likely that the resulting forwarding paths will share hops. Like in the fixed Internet this can be exploited by introducing multi-hop default routes for the benefit of route aggregation. The problem with a default route solution in a reactive routing context is that each intermediate node must ensure that the packet s destination address is not in the local ad hoc network before forwarding along the default route. This means that each intermediate node must flood the network with a fresh route request. This problem is described by Nilsson et al. in [6] as a cascading effect. Furthermore, in a Mobile IP scenario the appearance of multiple gateways cannot be prevented. At times, an ad hoc node may have connectivity to more than one gateway, for example to perform hand-over. We have found that default routes are unstable in this scenario. A control message received at an intermediate node could divert the default route in a way that is conflicting with the default route of upstream nodes, leading to stale NAT state at the gateway and interrupted TCP connections. An example of how this problem can occur is depicted in Figure 2. In this scenario, node B might still believe that it has to register at GW2 in case it is a Mobile IP foreign or home agent, while packets are forwarded to GW1. A B RREQ C GW1 GW2 Default route pointing to GW1 A B C RREP GW1 GW2 Default route pointing to GW2 A B C mismatch Figure 2: Inconsistent default routes with multiple gateways. Node A s Route Request is lost at GW2 and will make node C reconfigure its default route when processing the Route Reply from GW1. B keeps it default route pointing to GW2, which establishes a mismatch between node B s and node C s view of the forwarding paths. GW1 GW2 2.2 Improving Default Route Forwarding In ad hoc networks, default routes must be extended to support multi-hop paths and multiple gateways. To avoid cascading look-ups, [11] suggest adding host route entries for Internet destinations that point to the default route. Figure 3 (a) shows a routing table with this configuration. To support multiple gateways, the table is modified like in Figure 3 (b). An explicit mapping between the default route and the gateway is added to discern between default routes leading to different gateways. This is necessary for ad hoc nodes to, among other things, keep track of the gateway that they are currently registered at, in case they are using Mobile IP. Destination Next Hop Hop Cnt 63.3.5.23 63.3.5.23 1 66.35.250.151 default _ default 63.3.5.23 3 (a) Destination Next Hop Hop Cnt 192.168.1.1 63.3.5.23 3 63.3.5.23 63.3.5.23 1 66.35.250.151 default _ default 192.168.1.1 3 (b) Figure 3: Two different routing table configurations to the same end. The address 66.35.250.151 is a destination on the Internet. For AODV, setup (b) does not need sequence numbers for the default route entry, since it is just a mapping to the gateway route. Note that (a) may require two routing table accesses and (b) in worst case three. Route look-ups may be necessary at each intermediate node on the path to the gateway and might prove costly. Although better than Figure 3 (a), there are still other problems with the approach in Figure 3 (b). The forwarding relies on host route entries in the routing table. Therefore, whenever a route to a gateway is updated, intermediate nodes that were not part of the path before the update, must gather the mapping state of upstream nodes (i.e., the nodes that have a higher hop count to the gateway) for replicating it in their own routing tables. Otherwise, they may not be able to forward the packets of the upstream nodes. Currently, we do not know a clean way to handle this. 3. FORWARDING TO GATEWAYS USING TUNNELS Tunnels have the architecturally appealing property that they can be used by two end points to create a one hop illusion over many hops in the real network. In [5] Jönsson et al. propose tunneling as an alternative approach to default routes in ad hoc networks. We have looked at unidirectional tunneling, that tunnels traffic in one direction to a gateway acting as a Mobile IP home or foreign agent (or gateway running NAT). An encapsulated packet for an Internet destination, originated at an ad hoc node, is sent to the gateway using the gateway s explicit IP address and the IP forwarding mechanism as configured by the ad hoc routing protocol. At the gateway, the packet will exit the tunnel and is decapsulated. In the second step, the initial packet is routed towards the final destination in the global Internet. Return traffic inbound at the gateway does not need to be tunneled, since the return IP address (the ad hoc source node s home address) is routable within the ad hoc network. 3.1 Benefits

to gateways exhibit the following desirable properties: 3.1.1 Protocol transparency The tunneling concept is transparent with existing routing protocols. The minimum required modifications are extra routing table states in the source and gateway nodes which do not affect the protocol. Furthermore, there is no need for new state in intermediate nodes. This means that intermediate nodes can be unaware of gateways and tunnels, an important characteristic for legacy nodes. This also avoids cascading look-ups. 3.1.2 Route aggregation achieves route aggregation at intermediate nodes since all Internet destinations are encapsulated by gateway addresses instead of one entry for each destination which is the case for default routes. 3.1.3 Stability and reduced overhead Once a source node has configured a tunnel to a gateway, that tunnel will not be diverted to another gateway unless connectivity with the gateway is completely lost. In case the source node is running Mobile IP, it is notified and can re-register at a new gateway. 3.1.4 Multiple gateways Source nodes can maintain routes to multiple gateways for fault tolerance and load balancing. allows Internet traffic bound for different gateways over a common intermediate hop, which is not the case for default route forwarding (see Figure 4). Redundant tunnels can be used as (a) (b) (c) Figure 4: Multiple gateway support: (a) A default route points to only one gateway at once. (b) With tunneling two nodes can share an intermediate hop while still maintaining tunnels to different gateways, (c) or one node can have tunnels to two gateways at once. as backup routes if the connectivity to one gateway is lost and to do a soft hand-over between gateways. 3.1.5 Efficient forwarding With tunnling, a source node needs to perform two look-ups in the routing table (first, to find the host route entry that maps the destination to the tunnel entry point and thereafter the tunnel entry point itself, i.e., the gateway route). On intermediate nodes, only one regular look-up is needed, which is a clear advantage over the default route approach that most likely needs three routing table accesses, both at the source node as well as at intermediate nodes. 3.1.6 Security Tunnel forwarding is not as security sensitive as default routes; only the source node adds redirection state on the source node itself. on the other hand can be redirected at any intermediate hop and all Internet traffic diverted by a malicious node. 4. EVALUATION In this section we present simulation results that compare the performance of default route and tunnel forwarding using constant bit rate (CBR) UDP traffic and (FTP) TCP traffic. We show that tunnel forwarding achieves constantly better performance than the default route route counterpart. However, default route forwarding can perform better if we apply modifications that are conflicting with the AODV specification. 4.1 Simulation Methodology We use ns-2 version 2.26 and the ns-2 AODV-UU implementation of AODV. We have implemented gateway forwarding for both default routes and tunneling. We have run simulations where we scale the number of nodes from 10 to 20 nodes, incrementing by two at a time. Two gateways are used in the simulations. We keep node density fixed at 2 10 5 nodes per m 2 Thus, the area size (with an x:y ratio of 1:2) grows with increased number of nodes. We found this density to be a good balance between network size and number of nodes so that routes on average are longer when the number of nodes and area size increase. Ad hoc nodes move according to the random waypoint model, 20 m/s max speed. 50 movement patterns were randomly generated for each of the 6 configurations (network size and nodes) and used for all comparisons. One movement run lasts for 200 seconds. Performance averages were taken over all 50 runs for each network size and forwarding strategy. 4.2 Performance with CBR Traffic In our first experiment we examine CBR performance, where there is no feedback loop, i.e., no acknowledgments like in TCP. Since there is no return traffic, it does not matter to which gateway the traffic is forwarded and dramatic differences between forwarding strategies are therefore not expected (see section 2.1). Two CBR sources, sending 512 byte packets at a rate of 10 per second, are randomly selected among the ad hoc nodes and communicate with a randomly selected fixed node. The CBR sources start 5 seconds into the simulation and continue until the end at 200 seconds. In Figure 5 we see a comparison of delivery ratios (data packets received / data packets transmitted) for half tunnel forwarding and default route forwarding. Because of the randomness in movement patterns, the variance is high, but the differences between curves are significant. As can be seen from the left figure the delivery ratio decreases with the number of nodes. This is expected since the number of hops to the gateway increase with the size. Hence will the probability of connectivity loss increase with hop length and consequently the control traffic increase to handle the losses. This overall pattern occur in all our measurements. We note that tunnel forwarding consistently achieves better delivery ratio than the default route approach. Our

Delivery ratio (%) 100 95 90 85 80 75 70 Control traffic / data (%) 12 10 8 6 4 2 0 Figure 5: CBR Delivery ratio and normalized control traffic using 2 CBR sources and increasing path lengths. hypotheses is that the default route solution occasionally suffers from incorrectly replicated or missing state on the nodes along a default route, as we suggested in section 2.1. If this is the case, the AODV protocol will send a route error message to upstream nodes to invalidate their default routes and force them to be rediscovered. This would explain the larger amount of control message overhead of default route forwarding compared to tunneling in Figure 5. The inconsistencies in default route paths are likely to occur more frequently when we have more nodes and a larger simulated area, thus on the average longer routes. The delivery ratio in Figure 5 supports this view, since the gap between tunnel forwarding and default route forwarding increases with more nodes. To verify our hypotheses we changed the AODV implementation so that a node does not send a route error in response to an incoming data packet that it can not route. Instead it forwards the packet on a default route (if it has one), ignoring any state that is missing or is conflicting. The modification also drops any route replies that wants to reconfigure an existing default route, otherwise it operates as normal. The simulations with this modification (called default route mod. in the figures) show that the modification brings the CBR delivery ratio much closer to tunnel forwarding as we expected. However, this modification violates the AODV specification and we see potential security issues with forwarding any incoming packet as long as there is a default route. 4.3 Performance with TCP Traffic When evaluating TCP performance in a multiple gateway environment it is important that the return traffic (i.e., the acknowledgments) from the fixed network is sent through the same gateway as the forward traffic. Otherwise TCP acknowledgments might get lost in case they are sent to the wrong gateway, because connectivity might not be possible between that gateway and the source node. To support this we modified ns-2 s Mobile IP to work with the AODV implementation. Mobile IP s agent discovery was removed and replaced with AODVs RREQ mechanism just to simplify the set-up. For other parts Mobile IP works as specified. Whenever a mobile ad hoc node discovers a new gateway it will register with the agent at that gateway. It is crucial that source nodes in the future keep track of the gateway they are currently registered with and to make sure that the return traffic is sent the same way. For this experiment the two gateways in our scenario are assigned as Home Agent (HA) and Foreign Agent (FA), respectively. The scenario configuration otherwise remains the same. In Figure 6, we see that the expected drop in performance with the number of nodes, caused by the increased probability of losing connections. When comparing strategies we see that tunnel forwarding constantly achieves a higher TCP throughput than default route forwarding. The inconsistencies of default route forwarding are likely to sometimes cause TCP timeouts which explain the decrease in throughput. The TCP goodput (ratio of TCP packets successfully delivered to the total number of TCP packets transmitted) in Figure 6 and the control traffic overhead in Figure 7 supports this view. Similar to the throughput, we see a drop in the goodput when paths get longer, caused by an increase in retransmissions. Surprisingly, the goodput of tunnel forwarding is slightly lower than default route. An explanation for this is that with less timeouts for tunneling it will send more packets than default route and thus also retransmit more packets. This would reduce the goodput of tunneling while it still has a higher throughput than default route. At the same time, default route forwarding is not retransmitting that much, indicating that the decreased throughput is caused by timeouts. Control traffic is also likely to increase, since with tunnel forwarding, AODV spends more time delivering packets than idle in timeouts. In combination this will give less goodput for tunneling. Interesting is that the modified default route forwarding does not show a similarly strong improvement in this experiment as in the CBR case. This is in line with our assumptions that default route forwarding does not work well with TCP in multiple

0.4 1 TCP Throughput (Mbps) 0.3 0.2 0.1 0 TCP Goodput 0.99 0.98 0.97 0.96 0.95 Figure 6: TCP throughput and goodput using 2 TCP sources and increasing path lengths. Control traffic / data (%) 10 9 8 7 6 5 4 3 2 1 Normalized control traffic in TCP sce- Figure 7: nario. 0 gateway scenarios. After closer examination we found the explanation to be inconsistent state in the default routes that stops the end node to register a change in gateway. Thus, the source node is not triggered to send a Mobile IP registration message to the new gateway, because it thinks its default route still points to the old gateway. In the full version of this paper [7], we explore this issue further. 5. RELATED WORK Belding-Royer et al. proposes Globalv4 [2] with Mobile IP extensions for AODV. Destinations are first searched in the ad hoc network. If none are found, host routes are setup to the gateway. This solution suffers from long route discovery delays and lack of route aggregation that tunneling provides. Similar solutions for integrating Mobile IP with ad hoc networks can be found in [10, 12]. Globalv6 by Wakikawa et al. [11], can also work with Mobile IP, but it is not mandatory. Nodes may acquire a prefix from a gateway and through IPv6 s auto-configuration it constructs a globally routable IP address. Globalv6 employs a similar technique as Globalv4 to determine the locality of destinations. Routing towards the gateway is done on a hopby-hop basis using a default route. Cascading effects are avoided by requiring intermediate nodes to configure host route entries for Internet destinations, with the downside of losing route aggregation. Jönsson et al. studies in [5] the integration of Mobile IP in mobile ad hoc networks. from ad hoc nodes to the foreign agent is proposed as a way to achieve default route like behavior. This is the tunneling approach. However, the paper does not explore the benefits of this approach but instead studies different approaches to disseminate Mobile IP information in the ad hoc network. The Dynamic Source Routing protocol (DSR) is proposed by Johnson et al. [4]. DSR is interesting because it uses source routing and therefore supports the type of indirection that tunneling provide to operate efficiently with (multiple) gateways. would work transparently with DSR, but it would be an unnecessary addition. We also point to the LUNAR protocol [1] which tunnels all network traffic directly over the wireless link layer. LUNAR uses a kind of ARP forwarding as route request messages to discover destination nodes; The ARP reply corresponds to a route reply message and is used to build a data delivery tunnel between the involved mobile nodes. 6. CONCLUSIONS We have presented alternative ways to attach ad hoc networks to Internet gateways. In regular LANs, a default route is used for forwarding packets to the Internet. Ad hoc networks have different characteristics which make the concept default route inefficient and in some situations it will work incorrectly. We have compared the architectural aspects of default routes with our proposal tunneling in different ad hoc environments and with different protocols. to gateways is implemented within the AODV

protocol framework and the same implementation has been evaluated in ns-2 simulations. Default route forwarding was also implemented in ns-2. From our comparison, implementation experience and evaluations we conclude that: tunneling is architecturally simple and an elegant solution compared to default routes, tunneling does not require any changes to the ad hoc routing protocols or to intermediate nodes. Only a simple indirection step in the source node s routing table is needed, the performance of tunnel forwarding is better or at least comparable to the improved default routes in our simulation scenarios, tunneling will work with multiple gateways and that default routes will operate incorrectly in this environment. Multiple gateways will provide fault tolerance and load balancing, tunneling has double destination IP addresses which may create a substantial overhead for very small packets which is not the case for default routes. [7] E. Nordström, P. Gunningberg, and C. Tschudin. Evaluating Gateway Forwarding in Ad hoc Networks. Submitted to Mobihoc 2004. [8] C. Perkins. IP mobility support for IPv4, January 2002. IETF Internet RFC 3220. [9] C. Perkins, E. Belding-Royer, and S. Das. Ad hoc on-demand distance vector (AODV) routing, July 2003. IETF Internet RFC 3561. [10] Y.-C. Tseng, C.-C. Shen, and W.-T. Chen. Integrating mobile IP with ad hoc networks. Computer, (5):48 55, 2003. [11] R. Wakikawa, J. Malinen, C. Perkins, A. Nilsson, and A. Tuominen. Global connectivity for IPv6 mobile ad hoc networks, (work in progress), October 2003. IETF Internet Draft, draft-wakikawa-manet-globalv6-03.txt. [12] C. Åhlund and A. Zaslavsky. Software solutions to Internet connectivity in mobile ad hoc networks. In 4th International Conference on Product Focused Software Process Improvement (PROFES), 2002. Items for future work include how to handle DNS look-up for source and gateway nodes. There are several ways and algorithms to solve the locality problem at the gateway and they should be evaluated. There exists techniques for compressing headers that may be attractive to explore whether they work for this ad hoc environment. The possibility to use multi-homing for soft hand-over would also be interesting to explore further. 7. REFERENCES [1] LUNAR - a Lightweight Underlay Network Ad-hoc Routing protocol and implementation, webpage:. http://www.docs.uu.se/selnet/lunar. [2] E. Belding-Royer, Y. Sun, and C. Perkins. Global connectivity for IPv4 mobile ad hoc networks, November 2001. IETF Internet Draft, draft-royer-manet-globalv4-00.txt, (work in progress). [3] J. Broch, D. A. Maltz, and D. B. Johnson. Supporting hierarchy and heterogeneous interfaces in multi-hop wireless ad hoc networks. In Proceedings of the Workshop on Mobile Computing. IEEE, 1999. [4] D. B. Johnson, D. A. Maltz, and Y. Hu. The dynamic source routing protocol for mobile ad hoc networks (DSR), April 2003. IETF Internet Draft, draft-ietf-manet-dsr-09.txt, (work in progress). [5] U. Jönsson, F. Alriksson, T. Larsson, P. Johansson, and G. Q. Maguire Jr. MIPMANET - Mobile IP for Mobile Ad hoc Networks. In 1st ACM international symposium on Mobile ad hoc networking and computing (Mobihoc 00), 2000. [6] A. Nilsson, C. E. Perkins, A. J. Touminen, R. Wakikawa, and J. T. Malinen. AODV and IPv6 Internet access for ad hoc networks. Mobile Computer and Communications Review, 6(3):102 103.