Performance evaluation of associative based routing in Adhoc networks

Performance evaluation of associative based routing in Adhoc networks Ashwani Kush Department of Computer Science University College Kurukshetra University Kurukshetra 132 119 India Phalguni Gupta Department of Computer Science & Engineering Indian Institute of Technology, Kanpur Kanpur 208 016 India Ram Kumar Department of Computer Science & Applications Kurukshetra University Kurukshetra 132 119 India Abstract A Mobile Ad hoc Network is an autonomous system of mobile hosts which are free to move around randomly and organize themselves arbitrarily. In order to facilitate communication within the network, a routing protocol is used to discover routes between nodes. The primary goal of such an Ad hoc network routing protocol is correct and efficient route establishment between a pair of nodes so that messages may be delivered in a timely manner. In this paper, in addition to defining major areas of research in the area of network support for mobile computing, an attempt has been made to give emphasis on Associative E-mail: akush20@hotmail.com E-mail: pg@cse.iitk.ac.in E-mail: rkckuk@rediffmail.com Journal of Interdisciplinary Mathematics Vol. 9 (2006), No. 2, pp. 347 361 c Taru Publications

348 A. KUSH, P. GUPTA AND R. KUMAR based routing as new paradigm in place of existing shortest path routing algorithms. The shortest Ad hoc route derived at time T may no longer be valid at time T + 1, since nodes are moving. ABR (Associative Based Routing) and OABTR (Optimized Associativity-Based Threshold Routing) protocols are compared to give a view of Associative routing with a set of parameters. Mobile Ad hoc Networks community is trying to standardize the protocols and this study will provide a great help to them. Keywords : Ad hoc networks, routing, protocols, flooding, wireless, link state, MAC. 1. Introduction Since their emergence in 1970 Wireless Networks have become increasingly popular in the computing industry. This is particularly true in last decade which has seen wireless networks being adapted to enable mobility. Mobility in Ad hoc network doesn t pose any problem, unless two nodes that need to communicate have moved out of range from each other [2], [10]. In order to maintain communications, intermediate mobile nodes should cooperate to find a new multi-hop path to forward data from the source mobile to the destination mobile [11]. Mobility implies that networks need to cope with moving users. During a single session, users may connect from different network attachment points or use different networks or even more than one network simultaneously [8]. Hence, tomorrow s network will have to support movement of users, servers, services and even infrastructures to support mobile computing. Rest of the paper has been divided into 5 parts. Section 1 gives introduction to Ad hoc networks and issues related to research areas of mobile computing. Section 2 describes types of Ad hoc networks protocols. Section 3 introduces ABR (Associative Based Routing) Section 4 discusses OABTR (Optimized Associativity-Based Threshold Routing), Section 5 represents comparison of the two protocols, and Section 6 gives conclusions. Figure 1 gives a description of Ad hoc network, Node C moves out and a new route is established via node B. Some of the important areas of research in mobile computing are: Mobility: Protocols to support routing to and from mobile hosts. Addressing the issue of unicast routing to mobile hosts.

PERFORMANCE EVALUATION OF ASSOCIATIVE BASED ROUTING 349 Figure 1 Ad hoc network Dealing with large scale mobility, location management, scalable mobile multicast routing. Development of scalable and reliable multicast for mobile networks is another research problem. Design of authentication and access control schemes for mobile users. Location or position information as an additional parameter in protocol design. Providing routing protocols based on position or location [1, 3]. Providing support for querying services available within a given proximity Providing services with a specific geographic scope. Use of location in determining router functionalities, designation of responsibilities and services to entities within a geographic region. End-to-end QOS to moving users. Current end-to-end QOS schemes are designed to guarantee flow behavior with the assumption of relatively static route path between the sender and the receivers. With mobility, the flow path may change with each move; hence end-to-end delay will be impacted by mobility. Any contract that the network is willing to offer to a user must be valid in spite of mobility. Hence, services classes or contracts have to account for mobility [4]. Reservation schemes that support service classes in mobile networks need to consider spatial resource demands in all possible locations that a user may move during a given session[5] Solving these research problems will lead to interesting new services such as Internet Cellular Phone [6]. Adaptive protocols. Protocols have to adapt to a different set of parameters in mobile wireless networks. Protocols need to be

350 A. KUSH, P. GUPTA AND R. KUMAR designed for adaptation to parameters such as latency, burst error, disconnection during handoff, and asymmetry of the link, location, and cost. Mechanisms for introducing general purpose adaptation in a secure manner are needed. Work done in the area of extensible kernels is applicable here. Support for data and functionality migration. Network abstractions for data and functionality migration to and from the mobile host are needed. Data should be able to follow the mobile host. This should be balanced against the cost of migration [7]. 2. Routing protocols The significance of a routing algorithm is to compute a path from one node to another within a finite time. If more than one path exists then some selection criterion is needed. Most commonly used criterion is Shortest Path. Many protocols have been designed on this view and they are categorically divided in two ways as Table Driven Routing Protocol or Proactive On Demand Routing or Reactive Figure 2 Types of protocols In Table Driven Routing Protocols each node maintains one or more tables containing routing information to every other node in the network. All nodes keep on updating these tables to maintain latest view of the network. Some of the famous Table Driven or Proactive Protocols are: DBF

PERFORMANCE EVALUATION OF ASSOCIATIVE BASED ROUTING 351 (Bellman Ford Algorithm), GSR (Global State Routing), WRP (Wireless Routing Protocol), ZRP (Zone Routing Protocol), STAR (Source Tree Adaptive Routing). In On Demand routing or reactive protocols, routes are created as and when required. When a transmission occurs from source to destination, it invokes the route discovery procedure. The route remains valid till destination is achieved or until the route is no longer needed. Some famous on demand routing protocols are: DSR (Dynamic Source Routing), DDR (Distributed Dynamic Routing), TORA (Temporally Ordered Routing Algorithm), RDMAR (Relative Distance Micro Discovery Ad hoc Routing Protocol), AODV (Ad hoc On Demand Distance Vector). Brief analysis of protocols has been represented in Table 1 below. Table 1 Summary of routing protocols Protocol Year Loop Multiple Distri- Cate- Security Multi- Requires free Routes buted gory cast sequence data WRP 1996 Yes No Yes proactive No No Yes GSR 1998 Yes No Yes proactive No No Yes DBF 1992 Yes No Yes proactive No No Yes ZRP 1999 Yes No Yes proactive No No No DSR 1996 Yes Yes Yes Reactive No No No DDR 1999 Yes Yes Yes Reactive No No No TORA 1997 No Yes Yes Partially No No Yes RDMAR 1997 Yes Yes Yes Reactive No No No STAR 1999 Yes Yes Yes Proactive No No No 3. ABR (Associative Based Routing) All the above protocols are designed by keeping in view of the concept of shortest path. However shortest path routing cannot be appropriately applied to wireless Ad hoc networks. Since nodes are moving in an Ad hoc network, a shortest route derived at time T may no longer be valid at time T + 1. A new routing paradigm is needed. This new paradigm has been Associative Based Routing suggested by C.K. Toh [15]. Associativity is related to connection stability, spatial, temporal and signal stability of a mobile host. Associativity is calculated by one node s

352 A. KUSH, P. GUPTA AND R. KUMAR connection relationship with its neighboring nodes. A node s association with its neighbor s changes as it is moving and its transition period can be identified by associativity ticks or counts. The Associativity Based Routing considers a link between two mobile nodes to be stable if it exists for a long time, based on which the associativity ticks, are calculated and the route with the maximum value, is chosen. ABR protocol is free from loops, deadlocks and packet duplication. The rule of Associativity [9] states that a MH s (Mobile Host) association with its neighbor changes as it is migrating and its transiting period can be identified by the associativity ticks. Figure 3 Life of a mobile route As has been shown in Figure 3, in an Ad hoc mobile network initially there will be a period of instability. After that the route will be stable for some time during which the route can be used and then it will again become unstable (dying state), and the route cannot be used. Associativity ticks are updated by the MH s data-link layer protocol. A MH periodically relays beacons identifying it and constantly updates its associativity ticks in accordance with other MHs sighted in its neighborhood. 3.1 Working of ABR Every node in an Ad hoc network moves for some time, which is a transition period and then remains stable in one cell before it again starts moving. In ABR, a node detects the changes. So the associativity of a node with its neighbors changes frequently if the node is in transition. The nodes keep track of Associativity information using ticks for itself as well as for its neighboring nodes. The node periodically sends the beacon messages identifying it self and constantly updates the associativity ticks for its neighbors. Associativity stability of a node with its neighbors is found if the Associativity ticks exceed a threshold value. Low Associativity ticks for a mobile node with its neighbors indicate the high mobility of the node. High Associativity ticks indicate the dormant state of the mobile node, which is ideal for Ad hoc routing.

PERFORMANCE EVALUATION OF ASSOCIATIVE BASED ROUTING 353 When a node or its neighbor moves to a new location, the node resets the Associativity ticks. The property of associativity can also be used when Base Stations are available. Formula used for calculating Associative- Threshold [9] is A threshold = 2r pv 20 where r = Transmission range, v = Migrating speed, p = Beaconing interval. Association stability results when number of beacons recorded > A threshold. In a wireless network, where each cell is 10m in diameter and each MH is beaconing once every second, a mobile node migrating at pedestrian speed of 2m/s across a wireless cell with one or more neighboring MHs will record an associativity tick of no more than 5. So any associativity tick greater than this imply periods of stability. 3.2 Protocol overview Associative Based Routing (ABR) is a bandwidth efficient distributed routing protocol used in Ad hoc networks. ABR is a source initiated On- Demand Routing Protocol, which uses both point-to-point and broadcast routing. In ABR the destination node takes the decision of choosing a route based on the property of Associativity and the selected route is used and all other routes are discarded. This results in long-lived routes. The protocol consists of three phases: (a) Route Discovery phase. (b) Route Re-Construction (RRC) phase. (c) Route Deletion Phase. 3.2.1 Route Discovery The Route Discovery is initiated when a node wants to communicate with a node for which it has no valid route. This phase consists of sending a broadcast query (BQ) and Await-reply (REPLY) from the destination node or any node, which has route to the destination. The Query packet contains the Source ID, Destination ID, Intermediate IDs, Sequence Number, CRC, LIVE field and a TYPE field that identifies the type of the message. A node upon receiving a request message checks to

354 A. KUSH, P. GUPTA AND R. KUMAR see if it has already processed the request based on Sequence Number and discards the packet if it has already processed. Otherwise the node checks to see the Destination the packet is intended for. The node appends its ID to the list of Intermediate-IDs and rebroadcasts the packet. The Destination upon receiving the route query packet can find the best route to the source and then send a Reply packet. This makes all the intermediate nodes know that this is the best route and makes all the other routes to the destination as invalid. 3.2.2 Route Re-Construction (RRC) The Route Re-Construction or the route maintenance phase performs the following operations: (a) Partial route discovery. (b) Invalid route erasure. (c) valid route update. (d) new Route Discovery (worst case). These operations are invoked depending on the movement of the source, the destination or the intermediate nodes. When the source (SRC) moves the RRC process is initiated i.e., it results in BQ REPLY process as the protocol is source initiated. When the Destination (DEST) moves the immediate upstream router (pivoting node) erases its path to the destination and sends a LQ (Localized Query) as LQ{H} where H is the hop count from the upstream node to the DEST. If the DEST receives the LQ packet then the DEST selects the best partial path and sends a REPLY to the pivoting node. If the pivoting node doesn t get a reply for a LQ TlMEOUT period then the pivoting node backtracks to the next upstream node. This process is continued until a route is found to the DEST or until the pivoting node is more than half the HopCount from the SRC to DEST. 3.2.3 Route Deletion phase If the SRC no more needs the route to DEST then it sends a RD (Route Delete) message and all the intermediate routers on the way to destination delete the route from their routing tables. Performance 1. ABR is a source initiated Ad hoc On-Demand routing protocol.

PERFORMANCE EVALUATION OF ASSOCIATIVE BASED ROUTING 355 2. The protocol is best suitable for Small sized Ad hoc networks. 3. ABR provides fast route discovery and provides a shortest path basing on the property of Associativity. 4. The protocol exploits spatial, temporal, connection and power characteristics to construct a route that is long-lived. 5. There is no need for source routing. 6. ABR does not employ route caches thus reducing control overhead. 7. The protocol does not interrupt the source node each time a route is truncated, this is a problem commonly found in other protocols. 4. OABTR [13] This protocol proposes an Optimized Associativity-Based Threshold Routing (OABTR) protocol for Ad hoc mobile networks with excessive traffic. The routes are selected based on their lifetime and hence there is no need to restart frequently, resulting in higher attainable throughput. This protocol is based on source-initiated on-demand routing. Here the multipoint relays are used to calculate the route towards any destination in the network. The protocol is particularly suitable to the large dense networks with high nodal mobility and topological changes. OABTR is designed for Ad hoc mobile networks where mobile computers act as routers and packet forwarders in a wireless environment with no base stations. This protocol allows networks to be formed and deformed quickly, without users intervention and without the need for fixed network infrastructures. Unlike some existing link-state and distance vector-based approaches, OABTR does not attempt to consistently maintain routing information in every node. A route whose associativity is closest to the threshold value is chosen. OABTR is a compromise between broadcast and point-to-point routing and it uses the connection-oriented packet forwarding approach. The protocol exchanges topology information with other nodes of the network at regular intervals. To do efficient flooding of its control messages in the network, this protocol selects multi-point relays among the one hop neighbors with symmetric i.e. bi-directional link. Therefore, this automatically avoids the problems associated with data packet transfer on unidirectional links; as the problem of getting the ACKnowledgement for the data packets at each hop, which cannot be received if there is a unidirectional link in the selected route.

356 A. KUSH, P. GUPTA AND R. KUMAR OABTR takes care of following problems: 1. Routing information need not be stored on each and every node. 2. Routing needs to be done only on the request of the source or the sender who wants to transmit data. 3. Routing information need not be constantly exchange. It is done only on a demand [14] basis thereby reducing network traffic which is an important parameter in mobile networks. 4.1 Link/Connection Status Sensing Packet Exchange Mechanism OABTR uses a combination of BEACON and ECHO packets to ascertain connection (and indirectly link) status. On initialization, MHs would send a BEACON packet (once every BEACON PERIOD time units) to all its adjacencies to alert any adjacencies of the existence and identity of the broadcasting interface, which is its IP address. Reception of a BEACON at an interface implies either reconfirmation or creation of IN status of a connection at that interface, depending on whether or not the connection already exists, respectively. Once present, the status remains for MAX BEACON TIME (MBT) time units, at which time it times out if no subsequent BEACONs have been received; i.e. the link is declared DOWN and is removed from the data structures. ECHO packets are used to respond to BEACONs, to let a BEACONing router know that someone hears its BEACON. An ECHO packet contains the identity of the interface broadcasting the ECHO and the identity of the BEACONing interface to which it is responding. Reception of an ECHO at an interface implies either reconfirmation or creation of Bidirectional status of a connection at that interface, depending on whether or not the connection already exists, respectively. OABTR supports Multipoint Relaying (MR); a mechanism designed to minimize the overhead of packet flooding throughout a MANET by optimizing/reducing the number of duplicate retransmissions. In spite of the control overhead expenditures, MR has proven to be very efficient considering the high traffic found in the mobile ad-hoc network. A Multipoint Relay (MPR) is a router, which is selected by a one- hop neighbor to forward or retransmit that neighbor s packets. Every router has a set of nodes one hop away N1 (its one-hop neighbor set) and a set of nodes two hops away N2 (it s two-hop neighbor set). The objective of a router participating in MR is to select a minimal subset M from N1 so that their retransmissions cover N2.

PERFORMANCE EVALUATION OF ASSOCIATIVE BASED ROUTING 357 5. Comparison OABTR supports multipoint relaying. ABR uses Associativity ticks only. It was observed Multipoint relaying reduces overhead of packet flooding by reducing number of duplicate transmissions. In OABTR two threshold values were used, one for minimal threshold to select a link and other optimal threshold to select a particular route at the destination. ABR performs better in this context. OABTR uses reliability and packet transmission rate as metrics. ABR covers much more aspects to prove its point like Packet Delivery, Route discovery time, end to end delay, packet loss, communication throughput, route reconstruction time etc. OABTR shows graph of reliability with threshold values at 600 sec, 300 sec and 900 sec and is best performance at 900 sec. This is very close to ABR performance using Associativity Ticks. So no major advantage is gained in OABTR. OABTR gives comparison of packet transmission time with DSR, whereas no comparison has been made with ABR. ABR definitely performs better than DSR. Only merit of OABTR is Bi-directionallinks and multipoint relays but this also leads to increased packet size. Drawback associated with ABR is: Periodic beaconing at each Mobile Host in ABR may lead to power consumption problems. The signaling traffic grows with increasing mobility of active routes in ABR. The Graph 1 is study of OABTR [13], which compares it at different thresholds. It has been found that with more threshold value it is very close to actual ABR ticks. So it is not better than ABR. In order to further prove the point more comparative study has been conducted comparing the two protocols with DSR [16] and AODV [14] as on-demand protocols. Metrics used are control overhead, data throughput, and end-toend delay. For energyefficient communications, low overhead and high throughput is desired. The control routing overhead is total number of packets transmitted during simulation. It gives scalability, the degree to which it will function in congested or low bandwidth environments. Throughput is ratio of number of packets originated by number of packets received. This gives a picture of completeness and correctness of routing protocol. End to end delay is packet time from Source to Destination.

358 A. KUSH, P. GUPTA AND R. KUMAR Graph 1 Y axis: hit ratio i.e. the ratio of hits when the optimal route was chosen X axis: time in minutes for which the behavior of the network was observed Simulation results show ABR, DSR and AODV outperform OABTR in all three areas, DSR, ADOV performs better than OABTR, yet ABR performs slightly better than others in each case. Unlike DSR s path metric which could potentially lead to early node failure, ABR s Quality of Service CQOS) based metric uses route relaying load to avoid congested network areas. This increases the overall network life by distributing the energy consumed in route processing more evenly among the network nodes. OABTR worked close to ABR in this case. ABR s QOS routing metric is better suited to adopting power-conserving scheme which seeks to balance routing performance and energyefficiency. DSR, AODV do not have this facility. OABTR could not be evaluated in this scheme. Table 2 gives a description of studies carried out in comparing DSR, OABTR and ABR. Overall it is clear that ABR is better of the other two and in most of the cases it behaves better than the other two counterparts. 6. Conclusion The paper is an attempt to compare the two existing protocols based on the concept of associative based routing in an Ad hoc network. The earlier claims of OABTR better performing than ABR have been proved to be on the wrong side and it has been observed that in most of the cases the ABR is performing much better than its counterpart. Moreover concepts like power factor have not been taken into account by OABTR, whereas ABR discusses them and also uses them as one of the metrics. ABR uses more metrics to prove its points than OABTR. More comparisons have been made using DSR and AODV and it has been found that in almost all cases ABR performs much more satisfactorily. For these reasons, ABR is recommended for implementation in any Ad hoc network concerned primarily with maximizing energy efficiency.

PERFORMANCE EVALUATION OF ASSOCIATIVE BASED ROUTING 359 Table 2 Parameters AODV DSR OABTR ABR 1. Scalable No No No No 2. Routing Type Source Routing Source Routing Broadcast & Point to Point Point to Point 3. Routing Metric Shortest Path Shortest Path Associativity Associativity 4. Could suffer from Yes Yes No No shortest Path Problem 5. Beaconing Requirement Yes No Yes Yes 6. Supports additional Yes Yes No No Routes to Destination 7. Adaptation to new No Difficult Not Clear Easy Energy Schemes 8. Control Overhead Less than ABR Comparable to ABR Less than ABR Better of Three 9. End to End Delay Better than DSR Comparable to ABR Less than ABR Better of Three 10. Data Throughput Comparable to ABR Comparable to ABR Less than ABR Better of Three

360 A. KUSH, P. GUPTA AND R. KUMAR References [1] T. Imielinski, Mobile computing: dataman project perspective, in Mobile Networks and applications, January 1996, pp. 359 369 [2] MANET Working Group, URL http://www.ietf.org/html.charters/manet-charter.html [3] J. Navas and T. Imielinski, Geographic addressing and routing, in Proceedings of Mobicom 97, September 1997. [4] A. Talukdar, B. R. Badrinath and A. Acharya, On accommodating mobile hosts in an integrated services packet network, in Proceedings of IEEE Infocom, April 1997, pp. 1048 1055. [5] A. Talukdar, B. R. Badrinath and A. Acharya, MRSVP: a reservation protocol for an integrated services packet network with mobile hosts, Rutgers University Tech. Report, July 1997. [6] A. Talukdar and B. R. Badrinath, IPv6 + Mobile IP + MRSVP = Internet Cellular Phone, in Proceedings of IWQOS, May 1997. [7] B. R. Badrinath, A. Acharya and T. Imielinski, On structuring distributed algorithms for mobile networks, Computer Communications, April 1996, pp. 309 320. [8] M. Stemm and R. Katz, Vertical handoffs in wireless overlay networks, ACM/Baltzer Journal on Mobile Networks, May 1997. [9] C.-K. Toh, Long-lived Ad-hoc routing based on the concept of associativity, March 1999, IETF Draft, 8 pages, URL http://www.ietf.org/internet-drafts/draft-ietf-manetlonglived-adhoc-routing-00.txt [10] Z. J. Haas et al., Guest editorial, IEEE JSAC, Special Issue on Wireless Networks, Vol. 17 (8) (August 1999), pp. 1329 32. [11] S. Corson and J. Macker, Mobile Ad hoc networking (MANET): routing protocol performance issues and evaluation considerations, RFC 2501, January 1999. [12] C. K. Toh, Ad hoc Mobile Wireless Networks, Prentice Hall, New Jersey, 2002. [13] S. Manikandan, R. Naveenan and R. K. Padamanaban, Optimized Associativity-Based Threshold Routing for Mobile Ad hoc Networks, Guindy, Anna University. [14] C. E. Perkins, Ad-hoc on-demand distance vector routing, MILCOM 97 on Ad-hoc Networks, Monterey, CA, November 3, 1997. [15] C. K. Toh, Associativity based routing for Ad hoc networks, Journal on Wireless Personal Communications, Vol. 4 (First Quarter), 1997.

PERFORMANCE EVALUATION OF ASSOCIATIVE BASED ROUTING 361 [16] J. Broch, D. Johnson and D. Maltz, The dynamic source routing protocol for mobile Ad hoc networks, IETF Internet Draft, December 1998, URL http://www.ietf.org/internetdrafts/draft-ietf-manetdsr-01.txt