swarm-modeling
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

P2P Modeling


From: Owen Densmore
Subject: P2P Modeling
Date: Wed, 21 Feb 2001 16:17:55 -0800

Short Form: I'm looking for agent models (Swarm, RePast, Ascape) looking
at bottom-up, decentralized, Small Worlds networks.  Could anyone provide
a pointer or two?

Long Form:

I'm looking into modeling aspects P2P (Peer to Peer) systems.  These
systems are local-knowledge systems which do not use typical hierarchic
controls and infrastructure. See: http://openp2p.com/

The most famous example is file sharing among peers. Napster is most
well known, and has a centralized directory services while others like
Gnutella and FreeNet are truely decentralized.

Other examples are serverless web ..uh..servers (i.e. getting web pages
not from servers but from a hopefully Small Worlds collection of
friends).  Similar designs are emerging for mail/messaging/chat.
Another example is getting computational services from "the Grid" like
SETI and others do.

These systems, I think (.. suspect .. hope) are very amendable to agent
modeling because of the general insistance of local knowledge only ..
little or no reliance on central services such as naming and the like.
    Note: Peers tend to be fairly transient and often disadvantaged.
    DHCP, for example, renders most Peers "nameless" due to transient
    name to address binding.  They are often behind NAT gateways
    (Network Address Translation which lets several hosts use a single
    address .. but at the cost of being again without identity) or within
    limited ISP domains not supporting full internet connectivity
    (multicast, for example, is typically not available) or worse,
    behind a corporate firewall.

So in one sense, the P2P movement is forming a new network within
the existing one; one where all hosts are independent agents.  Thus
the interest in modeling them via an agent modeling system like Swarm,
RePast, Ascape and/or others.   (I'm limiting myself to Java for
platform independence .. another Peer value!)

Well, so after all that, I'd like to have pointers to existing models
in this area, or to papers discussing such models.  Mainly, I'd like to
see network oriented things, especially Small Worlds networks.  I'd
also be interested in models for "trust" and "reputation" in such
worlds.  From what I can tell, however, these have not been modeled on
agent simulators.

Thanks!!

Owen

Attachment: owen.vcf
Description: Card for Owen Densmore


reply via email to

[Prev in Thread] Current Thread [Next in Thread]