You are on page 1of 6

International Conference on Communication Theory, Reliability, and Quality of Service

Cooperation Enforcement in Vehicular Networks
Yao H. Ho1, Ai H. Ho1, Georgiana L. Hamza-Lup2, and Kien A. Hua1
School of Electrical Engineering and Computer Science, University of Central Florida, Orlando, FL
32816, USA
{yho, aho, and kienhua}
Department of Computer Science and Engineering, Florida Atlantic University, Port St. Lucie, FL
34986, USA

ABSTRACT The goal of this research is to address the security and
Operations of vehicular ad hoc networks rely on the collaboration cooperation issues for Connectionless Approach for Street
of participating nodes to route data for each other. This standard (CLA-S) in vehicular networks. There can be both selfish and
approach using a fixed set of nodes for each communication link malicious nodes in a vehicular ad hoc network. The selfish
cannot cope with high mobility due to a high frequency of link nodes are most concerned about their energy consumption and
breaks. A recent approach based on virtual routers has been intentionally drop packets to save power. The purpose of
proposed to address this problem. In this new environment, malicious node is to attack network using various intrusive
virtual routers are used for forwarding data. The functionality of
techniques. In general, nodes in an ad hoc network can exhibit
each virtual router is provided by the mobile devices currently
within its spatial proximity. Since these routers do not move, the Byzantine behaviors. That is, they can drop, modify, or
communication links are much more robust compared to those of misroute data packets. As a result, the availability and
the conventional techniques. In this paper, we investigate robustness of the network are severely compromised. Many
techniques to enforce collaboration among mobile devices in works ([1], [2], [7], [8], [9], and [10]) have been published to
supporting the virtual router functionality. Simulation results combat such problem - misbehaving nodes are detected and a
based on various system configurations are given. They indicate routing algorithm is employed to avoid and penalize
that the proposed technique is effective. misbehaving nodes. These techniques, however, cannot be
applied to CLA-S since any node in the general direction
1. Introduction towards the destination node can potentially help forward the
Vehicular Network (VNET) has attracted great research data packets.
interest in recent years. Similar to Mobile Ad hoc NETworks
(MANETs), a vehicular network is a self-organizing multi-hop The primary contributions of this paper are as follows: 1) We
wireless network where all vehicles (often called nodes) introduce a cooperation enforcement technique, called 3CE (3-
participate in the routing and data forwarding process. The Counter Enforcement), for the Connectionless Approach for
deployment of ad hoc vehicular networks does not rely on Street (CLA-S); 2) We apply the 3CE method to CLA-S; and 3)
fixed infrastructures such as router and base station, thereby We present simulation results to show that with the 3CE
posing a critical requirement on the nodes to cooperate with features, CLA-S can prevent malicious nodes and enforce the
each other for successful data transmission. Many works (e.g., cooperation among nodes to maintain the good performance of
[1], [2], and [7]) have pointed out that the impact of malicious the network. The remainder of this paper is organized as
and selfish users must be carefully investigated. Existing follows. We review the Connectionless Approach for Street
cooperation enforcement techniques ([1], [2], [7], [8], [9], and (CLA-S) in Section 2. In Section 3, we present our
[10]) can not be adapted for recent advance in routing cooperation enforcement technique for CLA-S. We give
protocols – connectionless oriented approach ([3] and [6]). In simulation results in Section 4 to demonstrate the benefits of
particular, we are interested in the Connectionless Approach the proposed techniques. Finally, we draw conclusion on this
for Street (CLA-S) [5], in this paper. This technique does not work in Section 5.
maintain a hop-by-hop route for a communication session to
minimize the occurrence of broken link. In CLA-S, the streets 2. Connectionless Approach for Street
are divided into non-overlapping grid cells, each serving as a To make the paper self contained, we first describe previous
virtual router. Any physical router (i.e., mobile host), work, Connectionless Approach for Street (CLA-S), in more
currently inside a virtual router, can help forward the data detail in this section. In CLA-S, the streets are divided into
packet to the next virtual router along the virtual link. This small “virtual cells.” These cells are divided according to
process is repeated until the packet reaches its final destination. intersections and blocks (see Figure 1). Instead of maintain a
Since a virtual link is based on virtual routers which do not hop-by-hop route between the source and destination node, the
move, it is much more robust than physical link. source only needs to maintain the location of the destination.
Using this location information, the source dynamically

978-0-7695-3190-8/08 $25.00 © 2008 IEEE 7
DOI 10.1109/CTRQ.2008.25

Authorized licensed use limited to: ACADEMIA SINICA COMPUTING CENTRE. Downloaded on April 21,2010 at 02:02:14 UTC from IEEE Xplore. Restrictions apply.

and Distn the Figure 1. it will compute the location of destination is discovered by the CLA’s location Forwarding Zones and save the result as a list of streets and discovery procedure where a simple broadcasting technique [4] the ranges of the streets that are encompassed by the is employed. The procedure is as follow. n will cancel the forwarding if n block. When a node (not the source node) receives a LR packet.. packet from both adjacent intersections. centerline of either a vertical street or a horizontal street (see Figure 2). The source will Forwarding Zones. The LD packet will propagate unit it reaches the destination node. D_Distn the distance between node n and the center of the cell denoted by the Destination Cell ID in the packet header. reason for this is that. compared to the single street coverage of another node on a 5) During this delay period. n might need to forward the effective radio range can cover the 2 intersecting streets data). n forwards the data. Reference Line. when at an intersection. we will use reference points to determine each Forwarding Zone. the center of the destination cell).. it will determine if it is on the grid path using a Reference Line (see Figure 2). we need to packet header (See Figure 3). we will set a 2) If n is not in the forwarding zones. destination cell ID). and Forwarding Zones Source and Destination (i.e. 2. The reference points (RP’s) on is to select a node farther away from m and closer to the a reference line are the interceptions of the reference line and destination node to forward the data packet. destination node ID. Source Cell ID and Destination Cell ID).e. Restrictions apply. Path Update. and Empty the same cell or if n is in a block cell and n hears the same Cell/Obstacle Recovery can be found in [5] and [6]. The significance of this equation determine the reference points. a node’s 4) If Steps 1. the source node can start communication sessions to the destination node by simply broadcast data packets which contain location information of Figure 2. the center of the source cell) and the destination (i.e. the delay for an 3) If n or any other node in the cell containing n has intersection node is set to one third of the normal DELAY. . shorter delay period. If the node n is at an intersection of two streets. source cell ID) and the location of the destination node (i. it is vertical). n does not forward the data. the delay of a node n is computed as cell ID of the node that is about to forward the data packet) in follows: the packet header. When the source node receives the LR packet. cell ID) of packets with the same Forwarding Area. Reference Line information. The LR packet includes the location of the source node (i.e..e. n delays the forwarding. When the destination received the LD packet. n does not forward. Delay for node n. When a node n receives a data packet from m. Once all reference points of a reference line have been determined.e.e. The detail information for CLA and CLA-S such as either hears the same packet from a neighboring node on Path Computation. or an intersection (a block is considered as horizontal if the street it is on has a horizontal orientation. and a unique request ID to determine the location of destination. Reference Line is the straight line that connects the source (i. if the forwarding decision has not been cancelled. A Forwarding Zone is an area that is determined by a reference point or the center of a source cell. otherwise.e.. α α DELAYn = − (1) 2 • D _ Distn 2 • Distn where α is a maximum delay constant in µsec. 6) At the end of the delay period. computes and selects a list of grid cells that form a When a node receives a packet with a new Forwarding Area “connecting” path between the source and destination.. Data Forwarding. distance between node n and the center of the cell denoted by the Current Cell ID (cell of previous relaying node m) in the Once the Reference Line has been established. In the above procedure. In the simulation. a vertical block.2010 at 02:02:14 UTC from IEEE Xplore.. This allows the node a quick and simple broadcast a Location Discovery (LD) packet that contain way to determine if it is in a Forwarding Zone for subsequent source node ID. and current cell ID (i. the destination node will reply with Location Reply (LR) packet. n does not forward. Reference Points. and 3 fail (i. Grid path.. the source node. 1) If n is the destination. The forwarded. 8 Authorized licensed use limited to: ACADEMIA SINICA COMPUTING CENTRE. Downloaded on April 21. location (i. A reference point can be on a horizontal block. the data forwarding procedure is as follows: Figure 3. The (because of a new reference line).

In order for malicious nodes to rejoin the network. By forwarding discard the packet.e. If a node n suspects the confirmed to be misbehaving.3 and 4. go to Step 2. and all packets exchanged Layer for broadcast to other nodes.. n’s MAC Layer passes the packet to its Counter) are updated according the counter update procedure. continue to forward (i. the 3C Module of node n source node is misbehaved. n invokes its Misbehavior discards the packet.e. its 3C Module resets the time to live (TTL) timer of misbehaving. Thus.. Unlike many techniques that and increases the Forward Request Counter. otherwise. between nodes are examined by the nodes’ 3C Module.. If the arriving Location nodes to determine the network condition in the local area (i. We will Detection Procedure to check the 3C header contained in the discuss when to initiate the misbehavior detection and the Location Discovery packet. a Location Discovery packet is broadcasted to find the three counters (i. If so. and 4) Time to live (TTL) timer. 1. The 3C Module determines if this Location Discovery An entry’s TTL timer gets to reset when the entry’s packet that packet belongs to one of the initiator’s (i.e. the source is not from a malicious node enter the 3C Module. and Location Discover the 3C Module in n uses the information in the packet’s 3C Counter) of the source node. a 2. it processes the packet Entry’s TTL (time to live) timer expired. Forwarding Counter. and Location Discover Counter) to this Location nodes can forward data packet according to the general Discovery packet before passing the packet to the MAC direction towards the destination. its neighboring nodes will penalize this node by the corresponding entry. Initiate Location Discover: When a Location Discover An entry is deleted from the pending queue when one of the procedure in the routing layer passes an initiated Location following is true: 1) A communication session ended. In Figure 4. If the source node is identified as node n. packet’s source node’s 3C Module adds the 3C’s header to the Location source ID = packet’s forward node ID). the once malicious node will again be Location Discovery packet that received by a node: 1) allowed to join the network. If m is the source node of this Location Discovery packet. Discovery packet matches an entry in the pending queue of n’s neighboring nodes). i. a malicious node can increase its down to lower layers). Based on this header. before a node can start a data transmission session to another 3. go to Step 3...e. nodes 5.e. then n exchanges 3C information 3. Once its location is determined. If the Location Discovery packet is not found in the Modules of non-malicious nodes still allow malicious nodes to pending queue. 3C otherwise. 3. the location of the destination node is needed session list. our Module then passes the Location Discovery packet to the approach allows malicious nodes to rejoin the network by routing layer for further processing (Step 5). Forward Request Discovery packet broadcast from a node m to any of its one- Counter.e. If the Location Discovery packet does not belong to the node can reply the Location Discovery packet.. Downloaded on April 21. When Routing Layer 9 Authorized licensed use limited to: ACADEMIA SINICA COMPUTING CENTRE. the 3C Module increments the Location Location Discovery packet. ratio of the Location Discover Counter to Forward Counter rises above a threshold β. Forward or Reply Location Discovery Packet: Depending on the different routing protocol used in the routing layer. Depending on different routing protocols. Forward the destination. then passed on to the routing layer (Step 5). Counter to Forward Counter fells below a threshold α and the respectively. or discard the Location Discovery packet according to its routing protocol. and 3) Entry belongs and updates the Location Discover Counter as follows: to an identified malicious node. go to Step 3. 3C Module for processing the Location Discovery packet and These counters are maintained by the node’s own 3C Module updating the Forward Request Counter as follow: so that malicious users cannot contaminate it. 2) Discovery packet to the 3C Module.. node ID. If the ratio of Request Forward procedure for misbehavior detection in Section 4. When its ratio of Request Forward reach the destination). 3C 4. This way. The Location Discovery packet is not forwarding Location Discovery packets it generates.e. 1. pass back data packets for other nodes. Forward Counter. Node n keeps records of Location Discovery packets it (i. node n can are given more incentive to act collaboratively. Forward 2.e. 2)Source node ID. Forward Counter. Forward Request Counter. Restrictions apply. If it does. The 3C Module adds a 3C header containing the values of node. The 3C Module determines if m is the source node that When a source node initiates a Location Discovery packet. Receive Location Discovery Packet: When a Location In a 3C Module. ratio of Location Discover Counter to Forward Counter fells Pending Queue contains the following information for each below threshold β. 3) Destination help forward its Location Discovery packets. Discovery packet as in various layers of the OSI model. header to determine if there is a need to start the detection neighboring nodes of the source node can decide to forward or procedure to examine m’s behavior. the initiated this Location Discovery packet (i. The 3C avoid the malicious nodes during the routing procedure. contributing its share of network workload. and Location Discover hop neighbor node n. header contains the value of three counters (i. node’s) ongoing communication session.4.2010 at 02:02:14 UTC from IEEE Xplore. otherwise.. the value of the three counters) with its neighboring received in a pending queue. or initiate a reply procedure (i. . intermediate Counter.e. If node m is discard the Location Discovery packet. three counters (i. go to Step 2. its neighboring nodes again Communication session ID.e. the 3C Module updates the pending queue participate in forwarding data. go to Step 3. Request Counter. forward the ongoing session.. we show the routing Counter to Forward Counter rises above threshold α and its protocol for CLA in the Routing Layer.. 3C Module Discover Counter by one and adds it to the ongoing In CLA-S.

Else. We formulate this using a simple model Receive Data Packet: When a node n receives a data packet. it then returns a data packet to ratiom ≥ relative forward rate. returns a Location Discovery packet to 3C Module. it increases the Forward relative forward ratio. 3C 2. If the Location Discovery packet does not match any entry Module need to exam the 3C header in the Location Discovery in the pending queue. Mode. If the data packet is not in the pending queue. the m might be misbehaving by not Counter by one and passes the data packet to the MAC forwarding packets. the changed where some area the number of nodes might be dense data packet is either discarded or forward (see Routing Layer or sparse. Else. it increases the Forward = 1. Downloaded on April 21. We discard any packets processes the Location Discovery packet and updates the that are not in the pending queue. information included in the packet. Ratio of Request Forward Counter to Forward Counter A Location Reply packet is generated by a node’s Routing < threshold α. n’s And 3C Module determines if the corresponding Location 3C Module does not invoke the detection procedure. If the routing layer local network condition around n. ∑ (absolute forward ratio ) + absolute forward ratio i i n 3. In Vehicular Networks. 1. it is the link to the next node. Else. n activates its Detection layer. s’s 3C neighboring nodes’ replies. 3C Module resets the time to live (TTL) timer Reqest Forward counterm belongs to this data packet in the pending queue and passes the data packet to the routing layer. the packet is discarded. matches any entry in the pending queue. In CLA-S.2 * n’s ratio of Location Discover Counter to Forward Counter by one. The 3C Module processes the data packet and at area of m might be congested which cause m not forward updates the Forward Counter as follows: packets. n’s 3C Module broadcasts source node can start a communication session. When a node n receives a Location Discovery packet destination. we do not need to invoke the Misbehavior 1. 3C Module determines if the data packet is in the pending Forward counterm absolute forward ratiom = queue. Ratio of Location Discover Counter to Forward Counter destination. the data packet is discarded. If the Location Discovery reaches its from a node m. go to Step 2. go to Step 3. . In CLA-S. if absolute forward ratiom < in the pending queue. Else. Depend on different routing protocols. packet matches any entry in the pending queue. a 3C packet (include n’s 3C information) to its one-hop nodes simply forward data packets without first establishing neighbor nodes. If not. 3C Module determines if the Location Discovery packet Detection procedure. Initiate Misbehavior Detection: In order to determine if there Else. Location Discovery packet passes to the Counter Update If so. If m is initiator of this Location Discovery packet. the relative forward ratio is merely the participate in the data forward procedure. 3C Module 2. And if absolute forward decides to forward data packet. where there are i number of neighboring nodes for n its MAC Layer passes the data packet to its 3C Module. Thus. neighboring nodes. this packet might reach its packet.2010 at 02:02:14 UTC from IEEE Xplore. 3C Module updates the Forward Request Counter as follows: The absolute forwarding ratio for m is: 1. the data packet i+n either discarded or forwarded. included in this Location Discovery packet for the following the Location Discovery packet or the Location Reply conditions: packet is passed to MAC Layer. the If one of the above condition is true. Counter. Then. 3C Module neighbor i’s (without m) and n’s absolute forward ratio as updates pending queue and increases the Forward follow: Request Counter before passing the packet to the routing relative forward ratio = layer for further process. 10 Authorized licensed use limited to: ACADEMIA SINICA COMPUTING CENTRE. Then.e. A malicious node can Forward Counter as follows: generate dummy packets to avoid evoking the Misbehavior 1. the 3C Threshold α = n’s absolute forward ration and threshold β Module discards the packet. every node has equal probability of drop packets. and others the channel might be congested cause to in Figure 4). When a source node s starts to send reply with its own 3C information. In this case. go to Step 2. 3C Module determines if the data packet matches any entry Detection procedure. (excluding m). procedure for further process. n’s 3C Module checks the 3C header 3. If so. Then. this Discovery packet matches any entry in the pending queue. n’s 3C Module checks if m is initiator (i. 3C Module checks if this Location Discovery > threshold β. Then relative forward ratio is the average sum of each 2. n calculates the absolute forward Module simply passes the data packet to MAC layer without ratio of m and the relative forward ratio of n and n’s update any counter. the routing layer generates a Location Reply source node) of this Location Discovery packet using the packet that corresponds to the Location Discovery packet. Layer when a Location Discovery packet arrived at a 2. it means that network condition 3C Module. destination. Else. When n receives its the data packet from routing layer to 3C Module. Once the location of the destination node is determined. If m is not the initiator. If so. Notice that all the nodes that are neighboring nodes of m and n can activate its Detection Mode due to they all have the similar relative forward rate.. If so. go to Step 3. 3C Module increases the Forward Counter by one. Thus. is a need to invoke the Misbehavior Detection procedure. Restrictions apply. When a node receives n’s 3C packet. network conditions are dynamically Forward Data Packet: Depend on the routing protocol.

In the experiments. In the reference scheme. During this delay period of time.. We references CLA (CLA-S-Reference).. the Misbehavior Detection total simulation duration for each run was 20 minutes (1200 procedure captures malicious users who deliberately discard seconds). in the proposed 3CE scheme. And the Detection packet’s source location is behind of n. Multiple simulation runs (100 runs per setup on First.e. turn left. In the Listening-State. a node in the Detection Mode is set destined at them. for defenseless scheme. The following mobility (i.e.1 Experimental Results exits the Detection Mode. a node probabilistically changes its degradation of data transmission. In other wards. by the CBR source. discussion assumes certain familiarity of the CLA-S. random period of time. 10%.2010 at 02:02:14 UTC from IEEE Xplore.. similar to regular data effective data transmission is performed. procedure to determine if m is a malicious node. the Misbehavior Detection procedure is invoked if there average) with different seed numbers were conducted for each is a suspicion of dropping packets was detected during the scenario and collected data were averaged over those runs. This measures the rate at which 2. Misbehaving node detection ratio (D): The ratio of the number of misbehaving nodes that were correctly identified 3. If n hears a Detection packet been forwarded by m. n invokes the Misbehavior Detection by 100 meters. to the total number of misbehaving node that have actually 4. 2. We observe in the We conducted various experiments to verify the effectiveness case of fewer malicious nodes (less then 10%). these nodes discard any packets not Detecting-State. Each node has a radio range of about 375 meters. the suspect node). we evaluated the proposed scheme based When a node n invokes its Misbehavior Detection procedure on the following metrics: to detect a suspect node m. Simulation Study node is 10% and 20% of the total nodes. In the defenseless scheme.. Next.. all the nodes act collaboratively and relay data for each other. A has been dropped four consecutively times. Second. n calculates a virtual link using the location information delivered to the destinations and the data packets generated (i.g. in forwarding data to rejoin the network. False accusation ratio (F): The ratio of the number of 3C 5. We focus on network layer directions of movement (e. significantly more data Discovery packet because m has passed n’s Misbehavior can be successfully delivered to the destinations since nodes Detection procedure. n waits for a t period of time (t = maximum delay time in acted uncooperatively during the simulation. No detection or prevention mechanism is to Listening-State. the routing layer). 512 bytes. n generates a Detection packet (i. identified. the improves the deliver ratio by more than 25% compare to the defenseless scheme and the proposed 3CE scheme. 3. It is also a good packet) based on the virtual link where its destination indicator of the degree of collaboration among the nodes. performance evaluation. location (i. and 30% of total number of nodes) and node purposes or to mount denial of service attacks. Each data packet is contained by the Misbehavior Detection Module are as follows. or continue attacks and do not address lower level threats such as physical in the same direction). they participate procedure to test a suspicious node. if n does not receive the Detection Modules that incorrectly accused benign hosts to the packet forwarded by m. node n enters the Detecting-State. n Finally.. All nodes employ 802. n exits All the experiments were conducted using GlomoSim [11]. At the end of the delay. 2. The location discovery phase. the Detection Mode. with a street width of 10 meters and building block size of 100 In the Detecting-State. If n hears a Detection packet from another node to test the is been punished when Location Discovery packets of the node node m (i. Once malicious Detection packet is generated by Misbehavior Detection nodes recognized themselves been punished. The proposed technique implemented three schemes.11 at the MAC layer. cell ID) contained in m’s Location Discovery packet. turn right. Restrictions apply. n resets the delay time. 10 m/s to 25 m/s or 22 mile/hr to 56 mile/hr). n broadcasts this Detection packet. The nodes The purpose of the Misbehavior Detection procedure is to move in the directions permitted in the streets. Downloaded on April 21. namely the reference scheme. Upon arriving detect uncooperative behaviors that result in disruption or at an intersection. relative to n. a certain fraction of nodes are misbehaving as they failed to participate in forwarding 11 Authorized licensed use limited to: ACADEMIA SINICA COMPUTING CENTRE. Initially.e. Experiments were based on a mobile ad hoc network with 200 nodes. are now required to participating in data forwarding. the procedure is as follows: 1. the CLA-S of the proposed 3CE (3-Counter Enforcement) scheme in with 3CE (CLA-S-3C) has very close throughput to the enhancing performance of vehicular network. packets that they are obligated to forward either for selfish 5%. n repeats the process again for two overall number of misbehaving nodes that 3C Module times (total of 3 times). The attacks rate sessions involving 1/10 of all nodes. And n forwards m’s Location By employing the proposed scheme. The field configuration is a 2000 by 2000 meters field At the end of delay period. a node n waits for a implemented so that the network is totally “defenseless”. Traffic applications are constant-bit- layer jamming and MAC layer disruptions.e. We varied the number of misbehaving nodes (i.e. misbehaving nodes are does the following: detected and punished. A malicious node can recognize itself 1. Packet delivered ratio (P): The ratio of the data packets 1. grid cell ID) is behind of m. n 4. . Figure 4 depicts the practical scenarios where the number of malicious 4. If n receives the detection packet which is forwarded by m. The Detection Mode has two states: Listening-State and procedure. 20%.e..

Gerla “GloMoSim: a library for parallel with very low false accusation ratio.1 improvement is accomplished with almost no false accusation 0 of cooperative nodes. Buchegger and J. They indicate that the CONFIDANT Protocol: Cooperation of Nodes – Fairness in Dynamic Ad Hoc Networks. technique can adapt by the CLA-S..A. Speed (m/s) 10 15 20 25 10 15 20 25 Vancouver. [4] A. Karp and H.” Proceedings of IEEE 10% misbehaving nodes 93% 91% 86% 88% 1% 2% 2% 3% Intelligent Vehicles Symposium (IV 2005). on average less four nodes were incorrectly 6th IFIP Conference on Security Communications and Multimedia (CMS accused. Hubaux. vehicle-to-vehicle networks).2 improved compare to a defenseless network. and O. Sheu. In Proceedings 11th International Conference on Parallel and low. T. [10] P. Both results indicate that the proposed detection 2002). Restrictions apply. mechanism is able to detect most of the in-cooperative nodes [11] X. A. Downloaded on April 21. and K. Kung.5 results indicated that the proposed technique is very effective 0. Michiardi and R. The simulation 0. U. Ho. Table 1. the proposed scheme 10 15 20 25 incurs minimum overhead. June 2005. page 243-254. K. A Connectionless 30% misbehaving nodes 91% 87% 84% 85% 2% 2% 4% 5% Approach to Mobile Ad hoc Networks. GPSR: Greedy Perimeter Stateless Routing for Wireless Networks. Hua. proposed misbehaving node detection mechanism is very CH. As of efficiency. A. the 3CE’s detection ratio is about [2] L. and Kien A. pp. and H.H. Proc.L. Aved. S. M.3 significantly strengthened as the network throughput is greatly 0. Since the CLA-S addresses highly dynamic networks (i. REFERENCES We list the results of misbehaving node detection ratio for [1] S. Ozyer. pp.e. We observe that this ratio is higher when the speed of Distributed System. Detection ratio and False Accusation ration of CLA-S with 3CE. and were thus incorrectly classified by 3CE’s Misbehaving [9] S. Canada. Yao H. Hua. Alexandria. 5% misbehaving nodes 89% 88% 83% 81% 0% 2% 3% 2% [5] Ai H. the existing misbehaving-node detection techniques are not suitable.2010 at 02:02:14 UTC from IEEE Xplore. Egypt. August 2000. 2004.” Proceedings of International Wireless Detection Ration (D) False Accusation Ratio (F) Communications and Mobile Computing Conference (IWCMC 2006). it is ideal for highly [3] H. configurations. 0. 6. K. T. Our approach supports this type of routing protocol under high mobility environments. thereby lifting the Misbehavior in Mobile Ad Hoc Networks. The results demonstrate that on-demand misbehaving Hoc WANs. 2005. Banff. pp.e. 188- 195. Contention-Based Forwarding for Mobile Ad-Hoc Networks. and G. Ad-Hoc Networks. Buttyan and J. A Finite-State Model scheme under various scenarios in Table 1. Boston. The degree of collaboration is 0. 425 – 430. May 1998. August 2000.A.S. 351-369. of Ninth International Symposium on Computers and Communications (ISCC).” Proc. Alberta. Hartenstein. 12 Authorized licensed use limited to: ACADEMIA SINICA COMPUTING CENTRE. Such 0. 2002. 2) Each node maintains three counters to represent its own status (i. Packet Deliver Ratio (P) with 10% and 20% Malicious Nodes. Boudec. H. suspect nodes moved out of the detection node’s radio range MA. nodes 0. Mitigating Routing Detection procedure as misbehaving nodes. Widmer. simulation of large-scale wireless network. nodes is increased. no additional 0. USA. J. Since the proposed 3CE USA. A. Our contributions are as follows. MA. L. Baker. We report the false accusation ratios of the proposed 3CE [7] N. CLA-S-Reference (10%) CLA-S-Defenseless (10%) CLA-S-3C (10%) Since nodes only determine their neighboring nodes’ counters 1 CLA-Reference (20%) CLA-Defenseless (20%) CLA-3C (20%) information when a location discovery phase. July 3-6. Elsevier’s dynamic MANETs such as vehicle-to-vehicle networks.. June 2002. 2000. Kasemann. In Proceedings of IEEE/ACM MobiHoc. 1) We introduce an on-demand approach to misbehaving-node detection for the CLA-S approach. Hua.4 in enforcing collaboration. 154-161. “A Connectionless Approach to Mobile Ad Hoc Network in Street Environments. Vol 1.8 behave normally). 5. A. node detection is applicable. In most cases. reputation). Portoroz. and M. Ho. we proposed an efficient 3CE (3-Counter Enforcement) scheme to enforce collaboration for CLA-S in vehicular network. M... In Proc. page 255-265. 20% misbehaving nodes 91% 85% 89% 87% 1% 1% 2% 2% [6] Y. of MOBICOM ’00. Bagrodia. Mobility (m/s) Figure 4.6 and efficiency of the proposed 3CE technique. Slovenia. Mauve. Fukuoka. Boston. Ho. Hamza-Lup. In Proceedings of false accusation ratio. “A Novel Broadcast Technique for High-Density Ad Hoc Networks. This is due to the fact that some of the [8] B. Nevertheless.e. Enforcing Service Availability in Mobile Ad 87%. Performance Analysis of the various simulation scenarios in Table 1. Vol 1. Marti. of the 12th workshop on Parallel and distributed simulation. Fubler. Hua. Jiang. . Lai. We conclude that Scheme for Efficient Cooperation Enforcement in Mobile Ad Hoc in all node mobility scenarios the false accusation ratio is very Netowkrs. Ho. and M. 2003. Canada. further investigation of MOBICOM 2000. effective. Japan.7 We conducted various experiments to study the effectiveness Packet Deliver Ratio 0.H. Ho. Giuli. Molva.9 information is needed under a normal operation (i. Nevada. Lausanne. Conclusion In this paper. In Proceedings of IEEE/ACM MobiHoc. K. Zeng. no 4.J. CORE: A Collaborative Reputation simulation log files shows that under all simulation Mechanism to Enforce Node Cooperation in Mobile Ad Hoc Network. R. 2006. pp.