A NOVEL ROUTING STRATEGY FOR AD HOC NETWORKS WITH SELFISH NODES Shailender Gupta Chander Kumar Abstract— To secure the data in the ad hoc network the intermediate nodes with in a route are created on the basis of their currently known reputation (trust) index. However, a node which was earlier trust worthy may no longer to be so as it becomes selfish due to loss of power. This selfish behavior problem is quite common in ad hoc networks and the literature contains many strategies to deal with them. This paper presents a novel routing strategy to deal with such nodes. The strategy assumes that the selfish nodes are not malicious and obligated to speak truth about their current energy level. Index Terms— Adhoc Network, Selfish Nodes, Trust Based Routing

1 INTRODUCTION The initial routing protocols in MANETs such as AODV [1], DSR [2], TORA [3, 4] etc. were devoid of security features and assumed that the all the nodes, in the network will behave in the rational manner and extend the desired cooperation required for the multihop routing. Later it was found that it is utopia to assume that all the nodes would behave rationally and selflessly forward the packets for other nodes [5]. The reason for such behavior may be malicious or non malicious (for example, low level of residual power) [6, 7, 8]. Therefore, need was felt to secure the information being transmitted across the network. To accomplish this, security protocols such as ARAN [9], ARIDANE [10] SAODV [11] were proposed wherein the data is encrypted before transmission, so as to prevent them from attacks by the malicious node which may have entered into the network. These protocols suffered through two major drawbacks: first, encryption process is quite expensive computationally and secondly there is no fun of protecting the data if the route is not secured and the packet is dropped by any of the intermediate node due to one reason or the other. Keeping this in view the trust based protocols such as CONFIDANT [12], SORI [13], CORE [14], WATCH DOG and pathrator [15], OCEAN [16] were proposed. Also some researcher proposed the trust based version of standard protocols such as AODV [17]. Here in these protocols the behavior of each node in the network is observed by the other nodes of the network and a trust index is developed. This trust index developed by a node about other nodes of the network can either be due to its own observation ( first hand trust) or may be due to the opinion given by the other nodes (second hand trust). The trusted nodes, in such an environment, are used to forward the packets in a route. A node, before it is made the part of ————————————————

• Shailender Gupta is with YMCA University of Science and Technology, Faridabad, India. • Chander Kumar is with YMCA University of Science and Technology, Faridabad, India.

the routing path, has to be trusted one. The trust protocols rely upon the past behavior of the nodes [18]. However it is quite possible that a node which was earlier trustworthy may no longer continue to be so as it may become selfish due to its energy constraints [6]. Such a node is not basically malicious but is selfish and is unwilling to spend its residual energy for forwarding the data packets of other nodes. As the time passes by, there is an increase in the selfish nodes as more and more nodes loose their energy. The rise in the number of selfish nodes with the time can have the disruptive effect on the activities of the network and may bring the network activity to near halt. The literature provides many mechanisms to deal with such selfish nodes as discussed in the next section of this paper.

2 LITERATURE SURVEY The ad hoc networks provide ubiquitous connectivity without the need of fixed infrastructure. This makes them very suitable choice when the communication has to be provided temporarily such as in case of battle field, disaster hit area or to create a network between members of an interim group. Such a network is composed of mobile nodes which are powered by the battery. Therefore energy is a precious resource for all the nodes participating in the communication process and has to be used very carefully spent by every node who intends to stay alive in the network. The communication in the ad hoc network takes place using the concept of forwarding where a source node sends a packet to a far off destination node using intermediate relay nodes. This mechanism of transmission through relay node leads to the better connectivity and lower cost of power transmission than in case of direct transmission over large distance. Since the traffic in an ad hoc network is through the relay nodes hence it is desirable that every node participating in the network faithfully forwards the packets which it receives but are meant for some other node as destination. If such cooperation is received from every node in the network it would be an ideal situation. But like all other

aspects of real life here also the conditions are not ideal and there exists non cooperative nodes in the network. These nodes may have two reasons for their non cooperation: malicious attitude or selfish attitude. The malicious attitude of a node can be due to the opponent’s intervention in the network where it intends to sabotage the network activity. The selfish attitude may be due to the various reasons where a legitimate node in the network starts avoiding the forwarding activity due to its current low power status or it feels so over utilized in the forwarding activity and it fears that it will drain so much power that it will not have enough energy to send or receive its own packets in the future. The communication in the ad hoc networks suffers a lot because of the selfish attitude of the participating nodes and a lot of research work has been carried out to mitigate this type of attitude of the nodes and how to maintain the connectivity with such behavior persisting [19, 20]. The selfish behavior of the participating nodes leads to the increased retransmission and thereby the overall decrease in the throughput of the network. As the number of selfish nodes increase, the efficiency of the network keeps on decreasing which ultimately leads to the near or total shut down of the network. Keeping this aspect in consideration, the literature on ad hoc networks contains a lot of papers [21, 22, 23, 24, 25, 27, 28] which provide strategies or mechanisms to deal with or cope up with such nodes. The subsequent part of this section explains some of these strategies and their relative merits and demerits. The strategies used to tackle the selfish nodes can be categorized into two basic categories: Motivation / incentive based approach and detect and exclude approach. The motivation/ incentive approach tries to motivate the users of the ad hoc network to actively participate in the forwarding activities. Such a system involves certain amount of money transfer to the relay nodes, on behalf of source or destination, to motivate them to forward messages [21, 22, 23]. One of the motivation/ incentive based approach [24] is based on a virtual currency called nuglet. Every network node has an initial stock of nuglets. Either the source or the destination of each traffic connection use nuglets to pay the relay nodes for forwarding the traffic. The cost of a packet may depend on several parameters such as required total transmission power and the battery status of the intermediate nodes. Packets sent by or destined to nodes which do not have a sufficient amount of nuglets are discarded. The major drawback of this approach is the demand for trusted hardware to secure and maintain the record of the currency at central level. One such protocol, the ad hoc VCG [25]is based on the monetary transfer and discovers an energy-efficient path between the source and the destination. However, the number of messages that must be exchanged in order to find the route to the destination is quite high – in the order of O(n3), where n is the number of network nodes. Vikram Srinivasan et. al. [26] have proposed a random strategy to tackle the nodes which are not having enough battery. They introduced the concept of sympathy factor for each path on the basis of energy left in the interme-

diate nodes. If the sympathy factor is low for a certain path that means the relaying nodes do not have enough energy to participate in the route discovery phase. The path having maximum sympathy factor is selected and the data packets are routed from that path. Detect and exclude avoid selfish nodes from the routing paths. In these scheme two types of trust are developed first and second hand trust • First hand trust: The nodes personal observation about the neighboring node. • Second hand trust: The observation of neighboring node for the other neighbors. The Watchdog and Pathrater is a mechanism [19] based on detect and exclude principle to deal with the selfish nodes. It has been designed to optimize the forwarding mechanism in the Dynamic Source Routing protocol [2]. It has two components: Watchdog and Pathrater. The Watchdog is responsible for detecting selfish nodes that do not forward packets. For this purpose each node in the network buffers every transmitted packet for some time. During this interval, the node places its wireless interface into the promiscuous mode in order to overhear whether the next node has forwarded the packet or not. The Pathrater assigns different rating to the nodes based upon the feedback that it receives from the Watchdog. These ratings are then used to select routes consisting of nodes with the highest forwarding rate. The protocols CONFIDANT [12] adds trust manager and reputation index to the above mechanism. It maintains two lists to deal with the selfish nodes. The node which behaves rationally are kept in the friends list and the nodes which drops the packets or tampers it is kept in the black list by the other nodes. The list is exchanged by each node with its neighboring nodes. Based on these list trust of a particular node is calculated. Whenever the trust value for a particular node falls below a certain threshold the protocol stops forwarding packets of that node. In [27,28] Y. Zhang et. al. have describe a distributed intrusion detection system (IDS) for MANETs that consists of the local components ”data collection”, ”detection” and ”response” and of the global components “cooperative detection” and ”global response”. Their architecture is very promising in dealing with the selfish nodes but they neglect the aspect how their local data collection should find out on incidents like dropped packets, concealed links, etc. Another system is the”Collaborative Reputation Mechanism” or CORE [14, 29]. It is similar to the distributed IDS by Zhang et al. and consists of local observations that are combined and distributed to calculate a reputation value for each node. Based on this reputation, nodes are allowed to participate in the network or are excluded. In their work, the authors specify in detail how the different nodes should cooperate to combine the local reputation values to a global reputation and how they should react to negative reputations of nodes. For the actual detection of selfish nodes, they only refer to the work of Marti [15]. The protocols SORI [13] is also based on detect and exclude mechanism. It makes two record, the local evaluation record (First hand trust) and the overall evaluation

record based on the reputation index given by the nodes about their neighbors are very much similar to above mechanism. Each node in the network maintains tables of first and second hand trust of their neighboring nodes. Based on these tables the trust of a node is calculated and then action is taken against the selfish nodes The next section proposes a protocol that avoids such node from participating in the route discovery process that has their energy status as low or very low. This strategy thus prevents the energy deficient nodes from becoming selfish.

3 THE PROPOSAL The Communication in an ad hoc network is a multihop communication wherein a source node communicates with a distant node using intermediate nodes in order to save the power. Thus the major activity in an ad hoc network environment is to find a suitable route such that the delivery of the message is ensured beyond doubt. The route should be so chosen that all the nodes in the path are trustworthy, non malicious, unselfish and the hop count is minimised. The first receiver of the message to a distant node is some immediate neighbor of the source node. Therefore, it is necessary that every node in the ad hoc network must be aware of its immediate neighbors at every moment. To remain aware about its neighbor nodes, a node in the network keeps on broadcasting hello requests on the periodic basis and keeps on receiving the hello replies as well. Using these hello request and replies a node in the ad hoc network constructs and maintains a table of its neighbors known as neighbor table. Since the nodes in the ad hoc networks are mobile the neighbor table keeps on changing with time. Our proposal begins with the format for hello request packet as shown in Fig. 1. The hello request packet has three fields, namely packet Packet

Source

Power

Type

Address

Status

Fig. 1. Hello Request Packet

type, source address and power status. The packet type field denotes that it’s a hello request packet, source address field is the identifier of the node in the network which generated the hello request and power status field indicates the current status of the power of the node issuing the hello request. There is no destination address in this packet as hello request is a broadcast mechanism. Hello reply is a unicast mechanism wherein a node rePacket

Source

Destination

Power

Type

Address

Address

tus

Sta-

Fig. 2. Hello Reply Packet

sponds to the node from whom it has received a hello request. The format of hello reply packet is shown in the Fig. 2. It has four fields: packet type, source address, destination address and power status. The packet type field here is hello reply, source address field contains the identifica-

tion of the node from which the reply packet originated, destination address field is the identification of the node to which the packet has to be sent and the power status field provides the current power status of the sender node. The proposed routing mechanism in this paper has modified the conventional hello request and reply mechanism to include a new feature called power status. This feature keeps a node aware about the power status of neighboring nodes. Thus the neighbor table of a node, in the proposed routing protocol, will have an additional entry in form of power status of the neighboring node. The knowledge about the power status of the neighbors helps a node in avoiding a node which is very low in power and may drop the packet in selfish manner to save the energy. A node in the ad hoc network can have five power statuses: very low, low, medium, high, very high with their ordinal values as 0, 1, 2, 3, 4 respectively.

4 PROPOSED ROUTING ALGORITHMIC DETAILS Let N be the number of nodes in the ad hoc network under consideration, Let A be the source node and Z be the destination node, L[X] be the neighbor list of some arbitrary node X. Routing Algorithm Enum power status = {very low, low, medium, high, very high}; hop count= 0; current node=A path= ‘A’; do { If Z ∈ L [current node] /* If Z is the immediate neighbor */ { /*Route found*/ Path= Path + ’Z’; Current node= Z; /* destination reached*/ hop_ count ++; route=path; Return route and hop_count; Exit (0); /*Come out of the program*/ } else /*Z is not the immediate neighbour*/ { T=0 ; /* Number of neighbors whose Z is a immediate neighbor */ neighb_list=L[Current node]

while (neighb_list != empty) {Remove first element of neighb_list, say X. Send unicast message to X to enquire about number of its neighbors N, Power Status P and Ans to know if Z is a immediate neighbor of Current node; Read and Record (node_id, P, N, Ans); If (Ans=”Y”) T++; }; If (T==0) K=1; /* no one has information about the desti-

nation node Z*/ if (T==1)K=2; /* only one node has information about the destination node Z*/ if (T >=1)K=3; /* more then one node has information about the dhestination node Z*/ Switch (K) {Case 1: /* there is no neighbor node with Z as its immediate neighbor */ For each neighbor of current node Check (node id, P, N, ANS) If (P= medium || P= High || P= Very high) { Add node to the probable list; For every node i belonging to probable list, find suitability index Su[i]. Let X be node with highest suitability index. Su=ord(power status) – N/C; Choose X as the next node in route. hop_count ++; current node = X; path =path + ‘X’; } Break; Case 2: /* there is only one neighbor with Z as the immediate neighbor */ find record with (Ans= “Y”); Let this node be X; path=path +’X-Z’; Route = path; hop_count = hop_count + 2; Return route and hop_count; Exit (0); Break; Case 3: /* there is more than one neighbours with Z as the immediate neighbour” */ find records with (Ans= “Y”); Let these node be X1, X2, ……….Xk; For each node Xi to Xk find suitability index Su. Su=ord(power status) – N/C; Choose the node with highest suitability index. Let the node be Y. path=path +’Y-Z’; route = path; hop count=hop_count +2; Return route and hop_count;; Exit (0); Break; } } While (current node! = ‘Z’ || hop count <=16) /* ord (power status)gives ordinal value of power status for example if power_status = “ very low” ordinal value is 0 and for power status = “ very high” ord value is 4. Nodes with higher energy status are more suitable. The above algorithm uses an index based technique (suitability factor) to find a path from source to destination. The source node (A) initially checks its neighbor list L[A] to find out whether the destination node (Z) is a member

of list or not. If the destination node Z is a member of L [A] then the route is found and is A-Z. The destination confirms the route to the source with the request to initiate the data sending process. If Z is not a member of L[A] then A sends a route request packet to all its known neighbors using multiple unicast mechanism. The format for the route request packet is shown in Fig. 3. In response to the route request packet all the neighbors send a unicast route reply. The route reply packet sent by Packet Type

Source

Destination

Address

Address

(A)

(Z)

Neighboring Node Address

Fig. 3. Route Request packet

the neighbors includes the following information in the form of packet with format shown in Fig. 4. • Node id of the replier Node R • Number of neighbors in the neighborhood of R • Current Power status of R • An Ans (Y/N) to indicate Z is a member of L[R] when the node (who initiated route request) receives all the replies from its neighboring nodes it checks the Ans Packet

Source

Destination

Type

Address

Address

Ans

Neigh_Count

Final Dest

Fig. 4. Route Reply packet

field of each reply packet. The node then makes a list of neighboring nodes that have destination node (Z) as their immediate neighbor. The following case may be generated. • Only one node, say X, has Z as its immediate neighbor. In this case, the route stands found. The current path is extended by X-Z and is returned as a route. If the power status of X is very low or low then the data packet is attached with the special bit called tag bit to indicate to the intermediate node X to necessarily transmit the data. • If more than one node, say X1; X2;………Xk, have Z as their immediate neighbor then a suitability factor is calculated on the basis of • power status of Xi • Number of neighbors of Xi The formula for it is given below: Su=power status – N/C Su denotes the suitability factor N denotes the neigh_count C is a constant to be chosen by designer (set to 10 in our code) The suitability factor denotes that how much suitable a node is to forward the data packet. A node with higher power is more suitable than a node with lower power. A node with larger number of neighbors is more prone to interference; hence such a node is less desirable. C is a constt. which depends upon number of nodes in the net-

A simulator was designed in C++ in which an area of 35*35 sq. unit’s size was chosen. The nodes were distributed randomly in the given area and following performance metrics [30] results were recorded in an output file. 1. Average Power left per node 2. Average Throughput 3. Average no. of hop count for successful transmission 4. Average no. of retransmission required The following assumptions were made in measuring average power: 1. The node looses 2 units battery in transmitting a packet 2. The node looses 1.5 units of its battery power while receiving Initially each node was given 100 units of power. The simulator designed selects random source and destination every run. Each execution of program involved 20 runs. The average power left per node after all the 20 execution was recorded in an output file as shown in Fig 5 Throughput: Defined as total number of packets received by the destination node. The average throughput increas-

Average power left per node

Proposed Routing Protocol

Proposed Routing Protocol Average Throughput

5 SIMULATION RESULTS

throughput also get increased as shown in Fig. 7.

1.2 1 0.8

N=20

0.6

N=25

0.4

N=30

0.2 0 5

7

9

11

13

Transmission Range

Fig. 6. Average Throughput for different transmission radius

The average number of retransmission decreases as the transmission range increases as shown in the graph since as the transmission range increases the number ofneighProposed Routing Protocol Average Number of Hop count for successful Transmission

work • If no neighbor node has Z as its immediate neighbor then in this case a list of probable forwarding partners is created. Then for each node in the probable list suitability index is computed and the node with highest suitability index is chosen as the gateway.

100

6 5 4

N=20

3

N=25

2

N=30

1 0 5

7

9

11

13

80 Transmission Range

N=20

60

N=25

40

N=30

20

Fig. 7. Average number of intermediate hops for different transmission radius

0 5

7

9

11

13

Proposed Routing Protocol

Transmission Range

Retransmission

Avearage Number of

Fig.5. Battery status after 20 transmissions for different transmission radius

es as the transmission range increases due to the fact that with increase in transmission radius neighboring nodes get increased (see Fig. 6) resulting in higher probability for a data packet to reach to its destination within permissible hop count. The average number of intermediate nodes that a packet takes to reach to its destination for successful transmission is shown in Fig. 7. As seen in the graph the average number of hops are fewer at lower transmission range since the number of neighboring nodes is quite less as well as the average throughput is also very low. As the transmission range is increased the average numbers of hops for successful transmission increases since the number of neighboring nodes gets increased as well as the

4 3.5 3 2.5 2

N=20

1.5 1

N=30

N=25

0.5 0 5

7

9

11

13

Transmission Range

Fig. 8. Average Retransmission for different transmission radius

bor gets increased and the probability to reach to the destination also gets increased (see Fig. 8).

6.

CONCLUSION

The proposed mechanism assumes that the trusted nodes are well behaved and obligated to speak truth about their power status. The incorporation of power status in the hello request and reply helps in knowing the current power status of the neighboring nodes and excludes those with lower power. To decrease the interference a node with lesser number of neighbors is preffered over the node with more number of neighbors if other conditions are same. The proposed protocol can be implemented independently or can be augmented with other trust based protocols in order to take care of selfish nodes.

ACKNOWLEDGMENT The authors wish to thank Sh. Manish Gupta for implementing the routing protocol in C++.

REFERENCES [1]

C.E. Perkins and E.M. Royer. Ad hoc on demand Distance Vector routing, mobile computing systems and applications, 1999. Proceedings. WMCSA ’99. Second IEEE Workshop on, 1999, p90 - p100. [2] D. Johnson, D. A. Maltz, Dynamic source routing in ad hoc wireless networks, in Mobile Computing (T. Imielinski and H. Korth, eds.), Kluwer Acad. Publ., 1996. [3] V. D. Park and M. S. Corson. A highly adaptive distributed routing algorithm for mobile wireless networks, INFOCOM ’97, Sixteenth Annual Joint Conference of the IEEE Computer and Communications Societies. Driving the Information Revolution, Proceedings IEEE, Volume: 3, 1997 Page(s): 1405 -1413 vol.3. [4] V. Park, and S. Corson, Temporally-Ordered Routing Algorithm (TORA) Version 1 Functional Specification. IETF Internet draft, 1997. [5] K. Mandalas, D. Flitzanis, G. F. Marias, P. Georgiadis"A Survey of Several Cooperation Enforcement Schemes for MANETs" International Symposium on Signal Processing and Information Technology IEEE 2005. [6] L. Blazevic, L. Buttyan, S. Capkun, S. Giordano, J.P. Hubaux, J. Y. Le Boudec,”Self-Organization in Mobile Ad-Hoc Networks: the Approach of Terminodes,” IEEE Communications Magazine, Vol. 39, No. 6, June 2001. [7] L. Buttyán, J.-P. Hubaux, “Nuglets: a Virtual Currency to Stimulate Cooperation in Self-Organized Mobile Ad Hoc Networks,” Technical report No. DSC/2001/001, Swiss Federal Institution of Technology, Lausanne, January 2001. http://icawww.epfl.ch/hubaux/ [8] L. Buttyán, J.-P. Hubaux, “Stimulating Cooperation in Self-Organizing Mobile Ad Hoc Networks,” Technical Report No. DSC/2001/046, Swiss Federal Institution of Technology, Lausanne, 31 August 2001. http://icawww.epfl.ch/hubaux/ [9] B. Dahill et al., “A Secure Protocol for Ad Hoc Networks,“ IEEE ICNP, 2002. [10] Y. Hu, A. Perrig, and D. Johnson, “Ariadne: A Secure On-demand Routing Protocol for Ad Hoc Networks,“ ACM MOBICOM, 2002. [11] M. Zapata, and N. Asokan, “Securing Ad Hoc Routing Protocols,” ACM WiSe, 2002. [12] Buchegger S, Le Boudec JY. Performance analysis of the CONFIDANT protocol. In Proceedings of 3rd ACM International Symposium, on Mobile Ad Hoc Networking and Computing, June 2002.

[13] He Q, Wu D, Khosla P. SORI: a secure and objective reputation- based incentive scheme for ad-hoc networks. In Proceedings of IEEE WCNC2004, March 2004. [14] Michiardi P, Molva R. CORE: A collaborative reputation mechanism to enforce node cooperation in mobile ad hoc networks. In Proceedings of 6th IFIP Communication and Multimedia Security Conference, September 2002. [15] S. Marti, T. Giuli, K. Lai, and M. Baker, “Mitigating Routing Misbehavior in Mobile Ad Hoc Networks,” Proc. Sixth Ann. Int’l Conf. Mobile Computing and Networking (MobiCom), pp. 255-265, 2000. [16] Bansal S, Baker M. Observation-based cooperation enforcement in adhoc networks. Techical Report, Stanford University, 2003. [17] Jared Cordasco and Susanne wetze “Cryptographic Versus Trust-based Methods for MANET Routing Security” Electronic Notes in Theoretical Computer Science www.elsevier.nl/locate/entcs. [18] A. Pirzada and C. McDonald. Establishing trust in pure ad-hoc networks. In Twenty-Seventh Australasian Computer Science Conference (ACSC2004), 2004. [19] A. A. Pirzada, C. McDonald and A. Datta, “ Performance Comparison of Trust Based Reactive Routing Protocols” IEEE transaction on mobile computing, Vol. 5, No. 6, June 2006, pp 695-710. [20] Shin Yokoyama†, Yoshikazu Nakane Osamu Takahashi, Eiichi Miyamoto, "Evaluation of the Impact of Selfish Nodes in Ad Hoc Networks and Detection and Countermeasure Methods", Proceedings of the 7th International Conference on Mobile Data Management (MDM'06) [21] L. Buttyan and J. Hubaux, “Stimulating Cooperation in Self-Organizing Mobile Ad Hoc Networks”, ACM/Kluwer Mobile Networks and Applications, Vol. 8, No. 5, October 2003. [22] L. Buttyan and J. Hubaux, “Enforcing Service Availability in Mobile Ad-Hoc WANs”, in Proceedings of IEEE/ACM Workshop on Mobile Ad Hoc Networking and Computing (MobiHOC), Boston, August 2000. [23] N. Ben Salem, L. Buttyan, J.P. Hubaux, M. Jakobsson, “A Charging and Rewarding Scheme for Packet Forwarding in Multi-hop Cellular Networks”, in Proc. ACM MobiHoc 03, pp. 13–24, 2003. [24] L. Buttyan and J. Hubaux, “Nuglets: a virtual currency to stimulate cooperation in self-organized ad hoc networks”, in Technical Report EPFL, DSC, 2001. [25] L. Anderegg, S. Eidenbenz, “Ad hoc-VCG: A Truthful and CostEfficient Routing Protocol for Mobile Ad hoc Networks with Selfish Agents”, Proc. ACM Mobicom, pp. 245–259, 2003. [26] Vikram Srinivasan, Pavan Nuggehalli, Carla F. Chiasserini, Ramesh R. Rao, “Energy Efficiency of Ad Hoc Wireless Networks with Selfish Users”, --------------------------[27] Yongguang Zhang and Wenke Lee. Intrusion detection in wireless adhoc networks. In Mobile Computing and Networking, pages 275–283, 2000. also available as http: //citeseer.nj.nec.com/zhang00intrusion.html. [28] Yongguang Zhang, Wenke Lee, and Yi-An Huang. Intrusion Detection Techniques for Mobile Wireless Networks. to appear in ACM Wireless Networks (WINET), 9, 2003. also available as http://www.wins.hrl.com/people/ygz/papers/winet03. pdf. [29] Pietro Michiardi and Refik Molva. Prevention of Denial of Service attacks and Selfishness in Mobile Ad Hoc Networks. http://www.eurecom.fr/ michiard/pub/michiardi adhoc dos.ps. [30] S. Corson, and J. Macker, Mobile Ad hoc Networking: “Routing Protocol Performance Issues and Evaluation Consideration” rfc 2501, jan 2003.