Enhancing AODV Routing Protocol 4sing Mobility Parameters in VANET =

Transcription

1 Enhancing AODV Routing rotocol 4sing Mobility arameters in VANET = Omid Abedi, Mahmood Fathy, Jamshid Taghiloo Department of Computer engineering, Iran University of Science and Technology, Tehran, Iran omid_abedi>comp.iust.ac.ir, mahfathy>iust.ac.ir, _taghiloo>comp.iust.ac.ir Abstract VANET is new generation of ad hoc networks that implement between vehicles on a road. Because of high mobility, routing in VANET has more problems than MANET. Thereby, in this paper we propose a modification on AODV as MANET routing protocol to make it adaptive for VANET. When a node is mobile, it has three mobility parameters: position, direction and speed. In our method, we have used direction as most important parameter to select next hop during a route discovery phase. With respect to mobility model, if nodes has same direction with source and/or destination nodes, our solution might selects them as a next hop. osition is another parameter that we used for next hop selection. 1. Introduction Vehicular Ad-Hoc Networks (VANETs) are special type of Mobile ad Hoc Networks (MANETs), where wireless-equipped vehicles form a network spontaneously while traveling along the road. Direct wireless transmission from vehicle to vehicle make it possible to communicate even where there is no telecommunication infrastructure such as the base stations of cellular phone systems or the access points of wireless dedicated access networks, needed in the previous Intelligent Transportation Systems (ITS) [1,2]. This new technology has been attracting lots of interest in the recent years for ITS and also causes oint efforts of governments, standardization bodies, car manufacturers, universities and research centers in several proects world-wide. The US FCC has allocated 75 MHz of spectrum in the 5.9 GHz band for Dedicated Short Range Communication (DSRC) to enhance the safety and productivity of the nation's transportation system [1,3]. The FCC's DSRC ruling has permitted both safety and non-safety (commercial) applications, provided that safety is assigned priority. The USDOT and IEEE have taken up the standardization of the associated radio technology Wireless Access for Vehicular Environments (WAVE), now described as IEEE p [4]. In addition some other proects outside the US like: ReVENT proect [5] in Europe, InternetITS [6] in Japan or Network on Wheels [7] in Germany are aimed to solve existing challenges. So in a near future, vehicles could benefit from spontaneous wireless communications. There are many challenges in VANETs, as described in a survey [8]. According to FCC frequency allocation we can categorize two main classes of applications for vehicular ad hoc networks. The first category which was mentioned above is to improve the safety level in roads, and the second predicted to grow very fast in the near future, is commercial services i.e., comfort applications. In both beforehand mentioned categories, related (i.e. safety or comfort) messages should be exchanged between vehicles. There are many routing protocols for ad hoc networks [9], [10], [11]. One of the most important of them is AODV [12], [13], [14]. AODV is an on demand routing protocol. This protocol finds routes for a node only when it has data packet for transmission. AODV routing consists of three phases: route discovery, data transmission and route maintenance. Route discovery phase starts when a node wants to transmit data and has no route to destination. Now, AODV call route discovery process. In this phase, source node broadcasts a Route Request acket (RREQ) to its neighbor. Nodes that receive RREQ packets divide into three categories: the receiver node is the destination of route, the node that has a route to destination or none of both. In the two first situations, receiver unicast a Route Reply (RRE) packet to the route that received Route Request (RREQ) packet from it. The route that RRE packet traverses, selected as one of the main routes for source that has been sent RREQ packet. In the last situation receiver generate another RREQ packet and broadcast it to its neighbors. * This work is supported by Iran Telecommunication Research center /08/$ IEEE 229

2 The Last situation repeats until one of the first two situations occurs. After routing phase, routing process calls data transmission phase, then it transmits data packets across selected route. In this phase, it is possible that a link is broken and results in route expiration. In this situation, the maintenance phase calls to repair broken route or find a new route to destination. Now, node that its link was broken, unicast a Route Error (RERR) packet to the source node. The Source node after receiving RERR packet, searches in its routing table if find another route to old destination select that route as new main route for data transmission, else rebroadcasts new RREQ packet and seeks new route to continue data transmission. If source node cannot find new route to destination, data transfer stops and failure happen. One advantage of AODV is that for any pair of source and destination finds more than one route. Although this appears as advantage but more often this advantage acts as disadvantage. Finding several route need to exchange more control packet. This leads to increase routing overhead. In addition, it increases bandwidth consumption. Obviously, some of scenarios use all of discovered route and others only use part of discovered route and rest of routes are wasted. All of unused routes are routing overhead and all control packets for this route are wasted. 1.1 Related work AODV is one of the most important routing protocols in MANET. There are many improvements on this protocol. [15], is a version of AODV (SAODV) that improves AODV security. The SAODV routing protocol proposed in [15] used to protect the routing messages of the original AODV. SAODV uses digital signatures to authenticate non-mutable fields and hash chains to authenticate the hop-count field in both RREQ and RRE messages. In addition, many protocols eliminate broadcasting in network such as Neighbor Elimination Scheme (NES) has been proposed in [16], [17] and geoflood [18]. Another method that can be used for stable routing is mobility prediction [19], [20]. In that, by using mobility parameters such as node s speed and node s position, node s movement can predict. Therefore, we can select routes that are more stable than other route. In addition, before a route is disconnected, new route selects. By prediction, we can eliminate transmissions of control packets. Thus, reduce routing overhead and will achieve stable routes. Simulation results for some existing ad hoc routing protocols (AODV, DSDV, DSR, TORA) found in numerous papers [19] [20] [21] have concluded that AODV is one of the best ad hoc routing protocols with overall better performance in terms of three metrics: packet delivery ratio, routing overhead and path optimality. In these papers shown that AODV is a good routing protocol for scenarios with high mobility. Therefore, seems it is ideal as VANET routing protocol. Rest of paper organized as below: In section 2, we describe mobility model that used for this method. The main idea of this paper is proposed in section 3, in section 4 we evaluate our method with simulation, and finally we have a conclusion in section Mobility model We want improve AODV as MANET routing protocol and make it useable for VANET with high performance. Therefore, we must use one of VANET s mobility models. Manhattan is one of the most important mobility models for VANET. In Manhattan mobility model, several horizontal and vertical streets co-exist in the simulation field and mobile nodes are moving on the lanes of the streets. For each street, it has several lanes in both directions. (Users can also define a street with only one direction). lease note each lane should be separated from other lanes by some distance. That is to say, while designing the map file, the lanes are not supposed to overlap. However, the vertical and horizontal streets may cross with each other at the intersection. The mobile nodes are suppose to move ahead, turn left or turn right with certain probability on the intersection. In figure 1, Manhattan has a vertical and a horizontal street with an intersection. Each street has two directions be made of a lane. If one direction of a lane has positive value 1, then the lane on the opposite direction must have the negative value -1. Streets of each map have the maximum and minimum allowed velocity (V min, V max ). In this mobility model, acceleration is an input parameter. To calculate node s next speed, model uses acceleration and current speed of node, relation 1 shows it: v( t) v( t 1) + β * a( t) = (1) Where: 1 β 1 If be less than zero, it means that the node is moving with de-acceleration (negative acceleration) Otherwise, it moves with positive acceleration. 230

3 Figure 1. Manhattan map If current speed of node is greater than the maximum allowed velocity for its lane, current speed decreases to V max. Otherwise, if current speed of node is less than minimum allowed velocity for its lane, current speed increases to V min. If v (t) > V max Then v (t) = V max If v (t) < V min Then v (t) = V min 3. Method description In our method, we use node s direction as most important parameter for next hop selection. Another parameter, which affects next hop selection, is node s position, but importance of this parameter is less than direction. These mobility parameters have been obtained from GS[22]. When a source node wants to send a packet to destination node, first, routing protocol gets direction of source node and destination node. Then, with respect to direction of source node and destination node, recognize intermediate node that can be participate in route between source and destination. Because of using Manhattan mobility model, nodes can move in three situations: 1- Source node and destination node move in same direction. 2- Source node and destination node move in opposite direction. 3- Source node and destination are orthogonal. The later situation demands more complicated arithmetical computation, therefore, for simplicity; in this paper we have only considered two first situations and have prolonged it to our future efforts. As mentioned before, nodes in VANET move with high speed. Thereby, links between nodes maybe broken as soon as possible. Here route stability is less than MANET with low mobility or nodes with low speed. In other hand, if two nodes that are moving in opposite direction communicate together, their link breaks sooner than state which these node move in same direction. Therefore, if source and destination are moving in same direction, the protocol must only selects intermediate nodes that move in same direction with source and destination. However, if source node and destination node are moving in opposite direction, there is no restriction on intermediate node s direction. In supposed method, we also try to select intermediate nodes that are moving in suitable position between source and destination. Furthermore, their direction is important. Corresponding pervious description, a node can be select as next hop in route between source and destination that has two conditions: A. It moves in same direction with source and/or destination. B. Intermediate node s location is between source and destination. Algorithm No.1 describes next hop selection according to the above criteria. Source node, destination node and candidate node for next hop, are input parameters for this algorithm. The algorithm determines whether a candidate node can be an intermediate in route between source and destination. Get_Direction function returns direction of input node and Get_osition function returns location of any input nodes in network. Routes established by this way are more stable and have less overhead than original routing method. There is one problem; maybe we cannot find any intermediate node as next hop for routes having these restrictions. For solving this problem, we change our strategy. Firstly, we put lower bound on number of discovered routes and then divide algorithm into two steps: in step one, protocol searches for nodes that have both condition of position and direction. After this, if the results satisfy lower bound of routes, algorithm without doing anything will finish and do other routing phases such as route maintenance and data transmission. Otherwise, algorithm goes to step two. In this step, algorithm removes position condition and all nodes that are in same direction with source and/or destination, can be selected as intermediate nodes for route. Now, we summarize these two steps in route discovery phase as below: 231

4 Bool NeXtQYop (node, source, destination) NN Step1O Ds = WetQDirection (source)s Dd = WetQDirection(destination)S Dn = WetQDirection (node)s If ((Dn RR Ds) (Dn RR Dd)) NNStep 2O s = WetQosition(source)S d = WetQosition(destination)S n = WetQosition(node)S If ((svrnvrd) TT (dvrnvrs)) Algorithm No.1 Next Hop Selection Step one: Nodes that move in same direction with source or/and destination and their position is between source node and destination node can be select as intermediate nodes in route. Step two: Return TRUES Else Return FALSES U Else Return FALSES } If lower bound of routes does not satisfied, the position condition remove from routing restriction and only limitation on routing is node s direction. Thereby, only if the node moves in same direction with source or/and destination, it can select as intermediate node. Algorithm No.2 summarizes our solution for problem mentioned above. Attempt=O NN Step 1O If (((DnRRDs) TT (DnRRDd))UU ((dvrnvrs) TT (svrnvrd)) U Else 4. Simulation SendQRRERQpacket(node)S NRTX AttemptXX If (attemptv2) U Vait (w_t)s NN Wo to Step 1S Else if (NRVMin_route_thershold) NN Step 2O If ((DnRRDs) TT (Dn RR Dd)) SendQRRERQpacket (node)s NRXX } Algorithm No.2 DAODV 4.1 Simulation environment Glomosim is one of the well known simulators which is used to evaluate various routing protocols in wireless networks. It implements within the GloMoSim library [23]. The GloMoSim library is a scalable simulation environment for wireless network systems using the parallel discrete-event simulation language ARSEC [23]. Our simulation has been done in a 1000m 1000m area and for 700 second. Simulation area contains 3 horizontal streets, 3 vertical streets and 9 intersections. Each street has 2 lane and we do not supposed any traffic lights at intersections. The initial placement of 232

5 nodes is base on input trace file and mobility model is Manhattan mobility model. Maximum speed when the number of nodes changes is 20m/s and acceleration is 5m/s 2. When acceleration of nodes or speed of them change, number of nodes is 40 nodes. Transmission range of each node is 250meters and used as MAC layer protocol. Results are average of ten simulations run. We select CBR as application to send and its packet s size is 5000 bytes. Number of packets that we want transmit is packets. The parameter Min_route_thershold mentioned in algorithm No.2 has been set to 5 routes. 4.2 Simulation parameter A. Link Expiration Time (LET) We suppose that transmission range of every node is R. We also need node s speed to calculate LET. Suppose we want calculate LET between node i and. distance between them is d i, and velocity of each node is v i and v. Base of these parameters and if nodes move in same direction, we have: R + α* d i, LET = v v i (2) If two nodes are moving toward each other, this means those nodes firstly must reach another, then, they can far away. For this situation " is +1. However, if two nodes are move far away firstly, this means, these nodes never reach another. Therefore, " is -1 to satisfy this situation in equation 2. If two nodes are moving in opposite direction, equation 2 will be change to equation 3: R + α* di, LET = v + v i (3) In addition, here nodes can go far from each other or can move toward another. If nodes move toward each other, " is +1 and otherwise " is -1. After calculation of LET, we can obtain RET (Route Expiration Time). RET for a route is minimum LETs that make that route: RET = min{let 1, LET 2,, LET n } B. Broken links Route is more stable if it has less broken link in any connection. We use this parameter to show route stability. If the number of broken link per route is low, route is more stable. Otherwise, high broken link leads to exchange more control packet and have more packet loss than route with lower broken link. C. Route length Route length determines by number of hops that every source nodes needs to traverse until reach a destination. If the number of hops is low, the route is short and if the number of hops is high, the route is long. In wireless network if a route uses a few numbers of nodes, this means, more nodes can participate in other routes. Therefore, network resource can be saved and be optimized. 4.3 Simulation results A. Link stability A route is stable when its nodes have connection for expected time that needed for data transmission. Routes life time strongly depends on their links connectivity. A link is stable if its nodes satisfying these three conditions; their moving directions are same, their positions are in acceptable states, and finally difference of their velocity is endurable with regard to their positions and direction Broken links &% &$ &# &" & % $ # " & " ' # number of nodes "#$ "#$ Figure 2. Broken link vs density In this paper, we have concentrated in two first parameters, position and direction. Figure 2, depicts number of broken links between source and destination with respect to change network density in our simulation environment. In this experiment, we have compared our method, DAODV, with original AODV. This figure shows that our method improves the original one about fifteen percent in average, it means that links which selected with respect to nodes directions and positions are more reliable. As depicted in figure 3, DAODV behaves uniformly on changing of node s speed. In AODV, number of 233

6 broken links increase faster than DAODV in high velocities. Broken links &$ &# &" & % $ # " & " ' # ( speed(s) "#$ "#$ Figure 3. Broken link vs speed B. Route length Number of hops between source and destination determines route length. Our solution tries to reduce number of hops to shorten route lengths between source and destination. Number of hops ) $ ( # ' " & Route length vs speed & " ' # ( speed(m/s) "#$ "#$ Figure 5. Route length vs speed As depicted in figure 5, our method is also sensitive to nodes velocities. It works well for the applications, which demands high-speed nodes, and selects shorter routes than AODV. C. Route Expiration Time (RET) Figure 6 shows Route Expiration Time (RET) versus network density. As depicted in this figure, increasing the number of nodes leads to decrease RET. Number of hops $ ( # ' " & "#$ "#$ & " ' # Number of nodes RET(s) &# &" & % $ # " & " ' # number of nodes "#$ "#$ Figure 4. Route length vs density Figure 4 and 5 respectively illustrates our gained results with regard to increments in number of nodes and their velocities. In these figures, the number of hops is counted for all routes. Figure 4 demonstrates that in our method, route lengths are shorter than AODV routes. The improvement is about 30% in average. Changes in amount of nodes in network do not intensively influence the efficiency of our solution. Figure 6. RET vs density This is normal because by increasing the number of nodes as depicted in figure 4, number of hops may be increase. Therefore, the links, which constructs route, will increase. Figure 6 show that DAODV has more stability than original one in route selection. Figure 7 shows that DAODV by changing speed of nodes is also more stable than AODV. 234

7 5. Conclusion Figure 7. RET vs speed In this paper, we have proposed DAODV protocol that improves the performance issues on common AODV protocol. The main goal of DAODV is to establish more stable routes especially in the applications that demand high mobility of nodes such as VANETs. The proposed method uses two important parameters of movement (direction and position) to select next hops during route discovery phase. Finally, our simulation experiments showed that DAODV protocol works well in various traffic situations. On the other hand, our solution behaves better in situations those nodes having higher speeds. The overhead of each route in AODV is less than DAODV, but the overall overhead of DAODV is less, because its discovered routes have been reduced. Now, we are trying to implement other movement parameters in selecting next hops during route discovery phase. 6. References [1] Dedicated Short Range Communications roect, DSRCHomeset.htm. [2] Internet ITS Consortium, [3] ASTM E , Standard Specification for Telecommunications and Information Exchange Between Roadside and Vehicle Systems 5 GHz Band Dedicated Short Range Communications (DSRC) Medium AccessControl (MAC) and hysical Layer (HY) Specifications, ASTM Int l., July [4] X. Yang, J. Liu and F. Zhao, "A Vehicle-to-Vehicle communication rotocol for Cooperative Collision Warning", MobiQuitous'04, [5] The ReVENT roect, [6] Internet ITS Consortium, [7] The NOW: Network on Wheels roect, [8] J. Lou, J-. Hubaux, " A Survey of Inter-Vehicle Communication", Technical Report, School of Computer and Communication Science, EFL, Switzerland, [9] D. Johnson et al., Dynamic Source Routing for Mobile Ad Hoc Networks, IEFT MANET Draft, April [10] S. Jaap, M. Bechler, L. Wolf. Evaluation of routing protocols for vehicular ad hoc networks in city traffic scenarios, 11th EUNICE Open European Summer School on Networked Applications, Spain, July 05. [11] Royer et al., A review of current routing protocols for ad hoc mobile wireless networks, IEEE ersonal Communications, Apr 99. [12] C. E. erkins and E. M. Royer, Ad-hoc On-Demand Distance Vector Routing, In roceedings of the 2nd IEEE Workshop on Mobile Computing Systems and Applications, pages , New Orleans, LA, 1999 [13] C. erkins, Ad Hoc On Demand Distance Vector (AODV) routing, Internet-Draft, draft ietf - MANET - aodv-00. txt, [14] C. erkins, E. Royer, and S. Das., Ad hoc on-demand distance vector (AODV) routing. Internet Draft, Internet Engineering Task Force, Mar [15] M. Zapata and N. Asokan, Securing Ad-hoc Routing rotocols, in roc. of ACM Workshop on Wireless Security (WiSe), Atlanta, GA, Sept [16] W. eng and X.C. Lu, On the reduction of broadcast redundancy in mobile ad hoc networks, In ACM MobiHoc 2000, pages , Boston, Massachusetts, USA, August [17] I. Stomenovic and M. Seddigh, Broadcasting algorithms in wireless networks, In roceedings of the International Conference on Advances in Infrastructure for Electronic Business, Science, and Education on the Internet SSGRR, L Aquila, Italy, [18] J. Arango et al., An Efficient Flooding Algorithm for Mobile Ad hoc Networks, University of Arizona, [19] S. Lee, W. Su, and M. Gerla, Ad hoc Wireless Multicast with Mobility rediction, roceeding of IEEE ICCCN'99, Boston, MA, 1999, pp [20] W. Su, Motion rediction in Mobile/Wireless Networks, hd Dissertation, UCLA Computer Science Department, Los Angeles, CA, [21] M. Takai et al., GloMoSim:A Scalable Network Simulation Environment, Technical report ,UCLA, Computer Science Department, [22] Global ositioning System Joint rogram Of_ce, URL:http//gps.losangeles.af.mil. [23] R. Bagrodia et al., ARSEC: A arallel Simulation Environment for Complex Systems, IEEE Computer,