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.

Resist<strong>in</strong>g Malicious Packet Dropp<strong>in</strong>g <strong>in</strong> Wireless Ad Hoc Networks 157is that one node does not need to rely upon the <strong>in</strong>formation from other nodes to detectmalicious ones. The disadvantage is that it may generate significant network overhead.The network overhead can be reduced if probe messages piggyback normal packets.To simplify the problem, we divided the prob<strong>in</strong>g technique <strong>in</strong>to three algorithms: 1)the prob<strong>in</strong>g path selection algorithm; 2) the prob<strong>in</strong>g algorithm; and 3) the diagnosisalgorithm. These are described below.4.1 Prob<strong>in</strong>g Path Selection AlgorithmThe prob<strong>in</strong>g paths are selected solely from the rout<strong>in</strong>g cache ma<strong>in</strong>ta<strong>in</strong>ed <strong>by</strong> a mobilenode. There are usually many redundant paths <strong>in</strong> the rout<strong>in</strong>g cache. Although prob<strong>in</strong>gover each of them may allow for validat<strong>in</strong>g all the known paths, it will also producesignificant network overhead. The ideal strategy shall select a m<strong>in</strong>imum number ofpaths but allows for monitor<strong>in</strong>g the forward<strong>in</strong>g behavior of as many nodes <strong>in</strong> the rout<strong>in</strong>gcache as possible. The prob<strong>in</strong>g path selection algorithm returns a set of paths with thefollow<strong>in</strong>g properties.1) For any two paths p i and p j , p i p j . S<strong>in</strong>ce prob<strong>in</strong>g over a path can alwaysdisclose the forward<strong>in</strong>g behavior of those nodes <strong>in</strong> any of its subsets, any path which isa subset of another path will be elim<strong>in</strong>ated.2) For any two paths p i and p j , if the second farthest node <strong>in</strong> p i is an <strong>in</strong>termediatenode of p j , the farthest node of p i will be removed. For example, given two pathsp 1 = A → B → C → D and p 2 = A → E → F → C → G → H, node D willbe removed from p 1 . With D <strong>in</strong> p 1 , A can monitor the forward<strong>in</strong>g function of C.S<strong>in</strong>cesuch monitor<strong>in</strong>g can be achieved <strong>by</strong> prob<strong>in</strong>g over p 2 , there is no need to keep D <strong>in</strong> p 1 .C will still be kept <strong>in</strong> p 1 ,s<strong>in</strong>ceA needs to monitor B <strong>by</strong> send<strong>in</strong>g a probe message to C.3) The length of any path (<strong>in</strong> terms of number of hops) is greater than 1. S<strong>in</strong>ce weare <strong>in</strong>terested <strong>in</strong> monitor<strong>in</strong>g the forward<strong>in</strong>g function of mobile nodes, prob<strong>in</strong>g over aone hop path offers no <strong>in</strong>formation. A probe message is sent to a neighbor node onlywhen a node subsequent to it doesn’t respond to the probe message and the prob<strong>in</strong>gnode needs to know if the neighbor node is BAD or has moved out of its transmissionrange.4.2 The Prob<strong>in</strong>g AlgorithmWith a set of selected prob<strong>in</strong>g paths, the prob<strong>in</strong>g algorithm will probe over each ofthem. Given a prob<strong>in</strong>g path, there are at least two ways of prob<strong>in</strong>g. One way is to probefrom the farthest node to the nearest. The other way is to probe from the nearest node tothe farthest. Each has its own advantages and disadvantages. Prob<strong>in</strong>g from far to near isbetter if the prob<strong>in</strong>g path is GOOD s<strong>in</strong>ce it takes only one probe message and proves thegoodness of all the <strong>in</strong>termediate nodes. But it may take more probe messages if a BADnode is located near the prob<strong>in</strong>g node. This method can be applied to a network wherewe have the confidence that the majority of the nodes <strong>in</strong> the network are GOOD. Theadvantage of prob<strong>in</strong>g from near to far is that it generates smaller number of prob<strong>in</strong>gmessages to detect a BAD node located near the prob<strong>in</strong>g node. Another advantage isthat we have the prior knowledge of the states of all the <strong>in</strong>termediate nodes along thepath to the probed node except its immediate predecessor node. The disadvantage is

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

Saved successfully!

Ooh no, something went wrong!