13.07.2015 Views

Page 2 Lecture Notes in Computer Science 2865 Edited by G. Goos ...

Page 2 Lecture Notes in Computer Science 2865 Edited by G. Goos ...

Page 2 Lecture Notes in Computer Science 2865 Edited by G. Goos ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

148 J. Zhen and S. Sr<strong>in</strong>ivasthe time of f<strong>in</strong>d<strong>in</strong>g a route with i hops be<strong>in</strong>g verified is Ti =3t∗i+t∗(Y −i).Theexpectation E i of the time consumed <strong>in</strong> f<strong>in</strong>d<strong>in</strong>g a route will be ∑ i=Yi=0 P i ∗ T i .Forexample, with X = 10% and Y = 6 we can have follow<strong>in</strong>g table:Table 1. Verification Probabilitiesi 0 1 2 3 4 5 6P i 0.531 0.354 0.010 0.015 0.001 0.000 0.000T i 6t 8t 10t 12t 14t 16t 18tFrom the table above, we can calculate E i =7.2t. Such that when neighborverification probability is 10%, the average extra time for f<strong>in</strong>d<strong>in</strong>g a new routewould be (7.203t − 6t)/6t = 20%. The fact that this overhead will be sharedamong the process<strong>in</strong>g of M RREQ’s makes it not significant.6 Discussion and ConclusionsAdd<strong>in</strong>g more mechanisms always needs more protection. The question is: willthis mechanism pose new security risks to the rout<strong>in</strong>g protocol? Can verificationrequests (VEF REQ) be replayed to some other area <strong>in</strong> the network? We arguethat this is not possible because each verification request needs the knowledge ofthe shared key between two nodes. The verification packets between two nodescannot be applied to other pairs.Can RTT REP packets be forged <strong>by</strong> illegitimate nodes to fool the node tocalculate wrong RTT threshold? This will not occur because if we have authentictrustable neighbors from <strong>in</strong>itial stage of the network and each round of calculationof RTT threshold is based on the replies from trusted neighbors, the authenticationof RTT REPs will be guaranteed and the impact from a particularnode will be balanced <strong>by</strong> other nodes after RTT averag<strong>in</strong>g.Can RTT REQs be replayed to some other area <strong>in</strong> the network? When receiv<strong>in</strong>ga RTT REQ replayed <strong>by</strong> an illegitimate node, the node may add theillegitimate node to its neighbor list and send back a RTT REP. S<strong>in</strong>ce the costof send<strong>in</strong>g a RTT REP is just trivial this will not impact the send<strong>in</strong>g node much.The replayed RTT REP will be discarded simply because the source node is not<strong>in</strong> the trusted neighbor list. The nodes <strong>in</strong> trusted neighbor list are added afterthe verification process.What if a neighbor moves away while the node still has not sent anotherRTT REQ s<strong>in</strong>ce there is an <strong>in</strong>terval? In this case, regular connectivity ma<strong>in</strong>tenancewill discover the leav<strong>in</strong>g of the neighbor and purge it out of its neighborlist.In summary, we proposed a solution to prevent two important types of replayattacks, namely, the wormhole attack and the RREQ flood<strong>in</strong>g attack, on theAODV rout<strong>in</strong>g protocol. Our technique is based on strengthen<strong>in</strong>g the neighborauthentication mechanism <strong>by</strong> a simple extension to the protocol. Analysis of the

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!