Setup Ad Hoc Routing Protocol over On-Demand Backup Node

Transcription

1 etup d oc Routing Protocol over On-emand ackup ode Ying-ong Wang, hi ing hung, hih-hieh huang epartment of omputer cience and nformation ngineering, Tamang University 151 Ying-chuan Road, Tamsui 25137, Taipei sien, Taiwan Tel: , inhon@mail.tku.edu.tw, cmchung@cs.tku.edu.tw, g @tkgis.tku.edu.tw TRT n ad hoc wireless network provides infrastructueless data networking; users have network services while they are continually moving. nstructureless, and mobility are properties of the ad hoc wireless networks. urthermore, each move of the mobile host affects the change of network topology, and affects the change of the route of transmission. There are numerous routing protocols have been developed for ad hoc wireless networks, and they may be generally categorized by table-driven and source-initiated on-demand. owever, the properties of the ad hoc wireless networks still strike those routing algorithms. The d oc ackup ode etup Routing Protocol (RP) is proposed to lead us pay more attention on the intrinsic properties of the ad hoc wireless networks. t provides more complete consideration will take the quality of routing much well. ccording to the proposed RP, the destination can receive some more routes in a period of time. Those routes, which are more than one routes, from the source node to the destination node may give us ways to analyze and find some good backup routes to get more help for reconnection when link failure happened. The backup route information can be saved in a specific on-the-route node. s soon as the backup routes can be rapidly found out and trace back these nodes when there would be situations such as disconnection or loss of connection. oreover, RP provides backup node mechanism to reconnect maintenance quickly for fitting the property of the ad hoc wireless networks.. TROUTO Wireless networks have been developed for a long time and become more and more popular in many areas, such as military, academic research, business, and so on. The ad hoc networks, is one kind of mobile wireless networks, which provides infrastructueless data networking. n ad hoc wireless networks, users have network services while they are continually moving. ach move of the mobile host affects the change of network topology, and affects the change of the route of transmission; therefore, those affections lead to a link failure of the route. esides, there is a radio wave communication for mobile hosts to communicate with each other. While the radio wave exposes itself in the air, the coverage of communication must be limited. ccordingly, when the environment is bad or the far distance is from a host to the other hosts, it is possible to make low quality of transmission, even of disconnection. Power limit will be also a restriction of the ad hoc networks. The mobile node in the ad hoc wireless network may be a notebook or P. t is unlike common static equipment, such as P, that the equipment usually has to depend on the finite energy source primarily by batteries. n order to save the consumption of electricity, power off or disconnection is the measure we often take. s above, when a mobile host takes a measure of power off or disconnection, the mobile host will disappear in the network and also the network topology by change. ometimes link failure is inevitable; thereupon, when network topology s speed of reconnection changes, it affects not only the communication among the nodes in networks but also the quality of packet sending. Once link failure happens, the packet would again keep delivery swiftly if there would be backup routes replaced. Otherwise, it is necessary that it will cost much time to find out a new route to the destination node. There are numerous routing protocols have been developed for ad hoc wireless networks, and they may be generally categorized by table-driven and sourceinitiated on-demand [1]. n the ad hoc networks, tabledriven routing protocols attempt to maintain consistent and up-to-date routing information which is from nodes to the other ones, such as estination-d istance-vector Routing (V) [2], lusterhead ateway witch Routing (R) [3], and the Wireless Routing Protocol (WRP) [4]. Those kinds of routing protocols require each node to maintain one or more routing tables which record routing information, and propagate update packets through the network to maintain a consistent network information when network topology changing or is in a period of duration. Table-driven routing protocols may cause heavy overhead in high mobility network environment, because network topology changes more frequently to update routing table fresh. Unlike the table-driven routing protocols, sourceinitiated on-demand routing protocol creates route when the source node only requires a route to the destination node. d oc On-emand istance Vector Routing

2 (OV) [5], ynamic ource Routing (R) [6], Temporally Ordering Routing lgorithm (TOR) [7], ssociativity-ased Routing (R) [8] and ignal tability Routing (R) [9] are all on-demand routing protocols for ad hoc wireless networks. ource-initiated on-demand routing protocols often process include route discovery to establish a route first when the source node requires, and then maintain the route when network topology changing happened. The issue how to find a route fast and stably is always being discussed in this area. owever, compare these two categories of routing protocols, source-initiated on-demand routing protocols often have performance well [1]. These two kinds of routing protocols do not think about backup route; therefore, while link failure is happening, it approximately has to find a new route. There are many limitations in the ad hoc wireless networks that are including high power consumption, low bandwidth, and high error rates. Whether the table-driven or on-demand is used, dynamic mobility characteristic of the ad hoc networks is the most important factor which effects the route s life cycle. n this article, a new ad hoc on-demand routing protocol was proposed. The Remainder of the paper is organized as fellow: n ection 2, we bring up the motivation of our research first. We illustrate the new routing protocol we proposed in ection 3. Then, in ection 4, we will give an example about our routing protocol. The paper then concludes with challenges and future works for our keep on research.. OTVTO or ad hoc wireless networks routing protocol, there are many methods had been proposed to establish stably, shortest, or fast routes. owever, instructureless and mobility are properties of the ad hoc wireless networks, they are still charge those routing algorithms. esides, the performance has not been improved much. Promoting the quality of ad hoc routing, we must manage to avoid those situations that are mentioned above. The directions are: 1. voiding excessive long route, 2. hoosing stronger route, 3. hoosing low mobility nodes, 4. peeding up maintenance or re-establishing routes, and 5. mplement the design of backup route. ccording to the five directions above, the major idea is to adapt the mechanism of backup routes. oreover, a new routing protocol generates by the idea. The method that we proposed leads us pay more attention on the intrinsic properties of the ad hoc wireless networks which the more complete consideration will take the quality of routing much well. f the source node needs route to reach the destination node, it would influence not much on Qo while spending some time searching the route within the tolerance time. That is, when the source node broadcasts RQUT packet to find out the route to the destination node, the route RQUT arrives first the destination might not be the shortest path or most stably route. ince, the throughput between a node and the other node affects RQUT s order to reach the destination, and so does the stability of the connection. f we wait a little time for the destination to receive some more routes, those routes, which are more than one routes for source node to destination node, may give us ways to analyze and find some good backup routes to get more help for reconnection when link failure happened. nd, we save backup route information in certain on-the-route node. s soon as the backup routes can be rapidly found out and trace back these nodes when there would be situations such as disconnection or loss of connection. The destination node replies the first arrival route as primary; therefore, we can design a period of time, Tc, to gather more routes as backup route analysis, and we can also analyze the backup routes saving methods of mechanism.. O UP O TUP ROUT PROTOO ome concepts of the d oc ackup ode etup Routing Protocol (RP) are basic on the ynamic ource Routing (R) [6]. The RP is an on-demand routing protocol that doesn t requires any table and replies the complete route from the destination to the source. The difference from the R is that RP allows intermediate nodes to receive and transmit the same request packet from the source reduplicate that is to gather more information to establish backup nodes. n RP, there are three phases, route discovery, backup node setup, and route maintenance, and there are three kinds of cache required (ee ig 1). urthermore, there are some basic protocol packets that need to be defined in the RP (ee ig 2). nd, there are statements of the three phases as fellow.. Route iscovery Phase When source node needs the route to destination, will enter the route discovery phase and check if there are or not adequate fresh routes to in resh_routes_ first. (ee ig 3) f some fresh routes to in resh_routes_ are found, will run Route_onfirm_Process instead. Otherwise, runs ew_route_iscovery_process to find a new route to the destination node. * ew_route_iscovery_process ource node broadcasts Route iscovery (R-request) to nearby nodes. n the R-request, there is a sequence number field to distinguish from the other route discovery process, and there is a route content field to record the address along the path from to. fter the on-the-way node receives R-request, the node inserts its address into the route content field of the R-request, and then broadcasts to the other neighbors. urthermore, the node received the Rrequest will record the information in the R-

3 request_ and sends R-request and sequence numbers to the other nodes. f there is a node that receives the same sequence number of R-request from its neighbor X, the node will check whether the route content of R-request has recorded its address; the node will discard this Rrequest. Otherwise, the node will insert its address into the R-request, and then check if X is or not in the Rrequest_. f it is, the node will clear X from the R-request_, and then forward the R-request again to the neighbor nodes recorded in the Rrequest_. f X is not in the R-request_, the node forwards the R-request to the neighbor nodes according to the R-request_. f there is no record in R-request_, the R-request will be discard. Unlike the R, RP only discards the duplicate R- when the node appeared in the route content field of the R-request or when there is no more downstream node recorded in the Rrequest_. ence, RP can prevent loop infinite. fter the destination node receives a R-request, will send the route content of the first arrival Rrequest back to the source node, and wait more Rrequests in a short span of time, and then enter the step of ackup ode etup Phase. Route iscovery Phase ew_route_iscovery_process igure 1. The main architecture of the RP Packet ame unction ain ields Route iscovery (R-request) Route onfirm (R- request) Route iscovery Reply (R-reply) ackup etup Packet (-packet) Route_onfirm_Process ackup ode etup Phase Route aintenance Phase Route rasure (R- request) Routing Phase ccess To find and record the route content from the source node to the destination node. To re-establish the path from the source to the destination node according to the fresh route in the resh Route. The destination node replies the route content back to the source. The destination node transmits the backup information to the backup nodes to setup backup routes. The source node announces other nodes to erasure routing information if no need. resh Routes R-request ackup Routes isk ackup node, ackup Route Route rasure igure 2. The main protocol packets of the RP resh Routes Y Read the route of the cache Route onfirm end Route onfirm (R-request) Y Return the route of the R-reply ode needs routes to node egin Route iscovery Phase igure 3. Route discover phase of RP * Route_onfirm_Process f there were a fresh route to the destination in the resh_routes_, the source node would add the fresh route from to into the Route onfirm (R-request), and then transmit R-request through this route. On the way to transmit R-request, intermediate node checks its resh_route_ whether there is any other fresh route to when it receives the R-request. f there is a fresh route, the node will copy R-request and put this route in the route content field of the R-requests, and then transmit the R-request by this fresh route. Of course, if there is not, R-request would be transmitted downstream according to its original route content field. ventually, after is receiving the R-request, Route iscovery Reply (R-reply) will send back to, and then will send packets by this route.. ackup ode etup Phase s there any fresh route in the cache Receive R-reply nding Route iscovery Phase roadcast Route iscovery (R-request) ew Route iscovery fter route discovery phase, the destination gathers several routes within a period of time, Tc. The destination will get backup nodes and backup routes. aving gathered some routes, we compare with nodes two by two (from beginning to the end) to find whether there is any the same section of any two paths. The last node of this section is ackup ode (destination is exception), and a backup node subset can be gathered from any two paths. fter comparison, we make union of all those backup node subsets. urthermore, the ackup etup Packet (-packet) with each backup node and the partial path from the backup node to the destination node are formed. nd then the destination Return O such route from to

4 node will make use of -packet to separately setup the ackup_route_ of those backup nodes.. Route aintenance Phase s a link failure happens, a node is not able to keep on transmitting. Therefore, this node will not pass a link failure message to upstream node until the message is gave back to backup node. fter the backup node receives the message of a link failure, it will fetch the backup route of ackup_route_ to replace the route after the backup node, and inform the source node to change the route. nd then, node sends packets to use the new route. esides, backup route that has been fetched by the ackup_route_ would be labeled as non-backup route. f there has been no more backup route in ackup_route_, the node will have lost the identification of backup node. Under the circumstances, there is no backup node exists. The source node will receive the link failure message and reenter the Route iscovery Phase to establish new route toward the destination. When the route from to is invalid, will store the usable routes into the resh_routes_ and broadcast Route rasure (R-). When backup node receives R-, it will save backup routes of ackup_route_ into the resh_routes_ as well. V. XP The process and the principle of RP are illustrated above. n this section, an example shows how the RP runs in practical case (see fig 4). When source node needs a route to pass packets to destination node, processes route discovery phase first. n the route discovery phase, broadcasts R- with a sequence number to its neighbor nodes. When the intermediate node receive the R-Packet (we take node for example) node receives R- Packet from, will check whether its address has recorded in the route content field of the R- or not. f it won t, will insert its address into the route content field and send the R-Packet to its neighbor nodes,,, and after recording <sequence number, downstream node> pair into the R-_ (n this case which is <#, >, <#, >, <#, >, and <#, >). f receives the same sequence number R- without its address in the route content field from one node, will also insert its address into the route content field and send this R-Packet to those nodes (,, and ) of the R-_, except for. Otherwise, loop infinite will be occurred if sends R- to again. urthermore, will discard the R- in those situations that R- has already contained or contained no pair in the R-_. ource ode estination ode igure 4.. n example of RP. ntermediate ode onnection When the destination node receives the first arrival R-, node will reply the R-Reply packet that is used by the route of the first arrival R-. or example, --> --> --> --> -->. Then the timer, Tc, will be started. Within a period of time, Tc, may gather some more routes from to, such as followings: --> --> --> --> --> --> --> --> --> --> --> --> --> --> --> --> --> --> --> --> --> --> --> --> --> --> --> nalyzing and comparing those routes with nodes two by two from beginning to the end, we can find out the route tree and backup node set {,,,, }(ee ig 5). ote that, the source node is also a backup node, because there are two ways for to send to. nd then, node sends -Packet with backup routes to each backup node. or this case, will set to be backup node and store { --> --> -->, --> --> -->, --> --> --> } in its ackup_route_. ackup ode ource ode igure 5. n example of route tree and backup node subset in RP ntermediate ode estination ode Path of Routes

5 fter the source node receives R-Reply from the destination node, then begins to send packets to. When link failure happens to the duration sending packets, there are some nodes could not send packet to the downstream node by the current path. This node will not pass over a link failure message to the upstream node until the message is gave back to backup node (ee ig 6). or example, node moves out and a link failure from to happened, node will send alert message back to the node. fterwards, backup node will check its ackup_route_, and pick one path ( --> --> --> ) to replace the current route, and inform to change the route toward (the new route is --> --> --> --> --> ). ater on, there might be no backup route in s ackup_route_; thereupon, will lose the identification of backup node. ackup ode ource ode ig 6. n example of link failure. V. OUO ntermediate ode n this article, a new ad hoc on-demand routing protocol, RP, is proposed. t seems that R and RP provide the shortest path. Thus, the destination node sends back the R-Reply with the route of the first arrival R-. oreover, RP provides backup node mechanism to reconnect maintenance quickly for fitting the property of the ad hoc wireless networks. ote that, Tc is a critical parameter that affects the performance of the RP. Tc begins with the s receiving first R-request or R-request from, and then analyzes some routes after Tc s finishing gathering. f the Tc is too large, there will be some redundant backup routes processed, and there will be some routes maintenance phase procrastinated. On the other hand, small Tc will reduce the gathered routes for analysis, even helpless for the route maintenance phase. To prevent the intermediate nodes from receiving and sending redundant R-s in the route discovery phase, they must stop receive any R-s after estination ode Path of Routes s receiving first R-request to the end of Tc. There are three measures as below: 1. estination node broadcasts message to each node. 2. ll nodes are set for synchronous clock. 3. odes receive R-request from and then start Tc by itself. owever, the first way will make the network overhead increase, and the second way is extremely difficult to set synchronous clock. The last way is more convenient and easy to implement, and we just need to choose an adequate Tc for the RP. n the future for RP, there are also many ways to be considered that can improve the stability of the routes. Taking ssociativity-ased Routing (R)[8] and ignal tability Routing (R)[9] for example, the intermediate nodes can choose more stable path first. nd, we will keep about some hot issues of the ad hoc networks, such as the Qo and mulicast to enhance the capability of the RP. RR [1] Royer,..; hai-eong Toh, review of current routing protocols for ad hoc mobile wireless networks, Personal ommun., Vol. 6 ss. 2, pr. 1999, pp [2].. Perkins and P. hagwat, ighly ynamic estination-d istance-vector Routing (V) for obile omputers, omp. ommun. Rev., Oct. 1994, pp [3].-. hiang, Routing in lustered ultihop, obile Wireless etworks with ading hannel, Proc. O 97, pr. 1997, pp [4]. urthy and.. arcia-una-ceves, n fficient Routing Protocol for Wireless etworks, obile etworks and pp.., pecial ssue on Routing in obile ommunication etworks, Oct. 1996, pp [5].. Perkins and.. Royer, d-hoc On- emand estance Vector Routing, Proc. 2nd Wksp. obile omp. ys. nd pps., eb. 1999, pp [6].. ohnson and.. altz, ynamic ource Routing in d-oc Wireless etworks, obile omputing, T. mielinski and. orth, ds., luwer, 1996, pp [7] V.. Park and.. orson, highly daptive istributed Routing lgorithm for obile Wireless etworks, Proc. OO 97, pr [8] -. Toh, ovel istributed Routing Protocol To upport d-oc obile omputing, Proc th nnual nt l. Phoenix onf. amp. and ommun., ar. 1996, pp [9] R. ube et al., ignal tability based daptive Routing () for d-oc obile etworks, Pers. ommun., eb. 1997, pp