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.

172 H.-P. Bischof, A. Kam<strong>in</strong>sky, and J. B<strong>in</strong>dertimes. Group key management paradigms which provide a shared symmetric key thatis shared among all group members, have been used throughout the wired network<strong>in</strong>gdoma<strong>in</strong> to secure broadcast and many-to-many communication environments; however,very few attempts have been made to adapt group key management <strong>in</strong>frastructuresto an ad hoc sett<strong>in</strong>g.Dom<strong>in</strong>ant group key management paradigms <strong>in</strong>clude the well-known CLIQUESproject [8], Kim et al [5], and several others. Each of these protocols is based on thegeneralized Diffie-Hellman problem, which requires every network member to contributeto the generation of the shared group key. Because they were developed forwired environments, many of these approaches require po<strong>in</strong>t-to-po<strong>in</strong>t and broadcastmediums, synchronous messag<strong>in</strong>g, and static network topologies. Unfortunately, thewireless, amorphous, transient, many-to-many nature of ad hoc networks precludesmany of the assumptions on which the above protocols were developed. We, therefore,<strong>in</strong>troduce a new approach to key management that can effectively functionwith<strong>in</strong> the constra<strong>in</strong>ts of an ad hoc network environment.Look<strong>in</strong>g ForwardThe ad hoc network environment we envision is transient, dynamic <strong>in</strong> structure andmembership, proximal, and broadcast-based. We also assume that network nodeswish to collaborate, that is, our primary goal is to ensure secure many-to-many communication.As a result, our paradigm is fully decentralized (i.e., it lacks server orotherwise specialized nodes), lightweight, and best-suited for small, spontaneousnetworks. The first protocol we present is not authenticated; the second is an extensionof the first that <strong>in</strong>cludes authentication mechanisms.The nucleus of our first protocol is a tuple-like entity, <strong>in</strong>spired <strong>by</strong> Gelerntner's tuples<strong>in</strong> [2], that is effectively a hash table shared among all members of the group.Each member of the group has an entry <strong>in</strong> the hash table, which <strong>in</strong>cludes that member'scontribution to the group key.The follow<strong>in</strong>g atomic operations may be performed on the tuple:• take() - removes the tuple from the space, such that no other group member maymodify its contents.• read() - reads the current contents of the tuple• write() - writes the tuple <strong>in</strong>to the space, overwrit<strong>in</strong>g the previous tupleAlthough the tuple spaces are often implemented as a centrally-based service, thetuples used <strong>in</strong> this context are fully distributed: each member hosts its own entry <strong>in</strong>the tuple. Nodes may host more than one entry if replication is desired <strong>in</strong> the <strong>in</strong>terestof availability.Group GenesisGroup genesis requires two or more parties to be present.1. Group members agree on a cyclic group, G, of order q, and a generator, α <strong>in</strong> G;each member then chooses a secret share, N i∈ G.2. The first member, M 1, <strong>in</strong>stantiates a Tuple Space and places a new tuple <strong>in</strong> thespace. The tuple <strong>in</strong>itially conta<strong>in</strong>s M i's contribution and the current card<strong>in</strong>al value.M ithen sends a broadcast message to the group stat<strong>in</strong>g that tuple has been created.

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

Saved successfully!

Ooh no, something went wrong!