Welcome to O N I S S

The Opportunistic Network Information Sharing Service

 

t h e    s e r v i c e
 
ONISS provides a survival-of-the-fittest model content sharing service that allows clients to download (legal) content from nearby clients, via an ad-hoc 802.11b service (Wave Relay). Users rate and request content and submit ratings to a server, which computes behavior-derived content suggestions, and distributes efficient content-source mappings. In the absence of the server, clients communicate directly. Downloaded content is maintained in a holding pen till the client user accepts or deletes the item.

The ONISS system will result in the rapid distribution of popular content, and the slow distribution and removal of unpopular content. The relative popularity of content, and other statistics, will be available from the central catalog maintained on the server. The service will be written in Python. The data interchange format will be XML Schemas.

t h e    d e s i g n    t e a m
 
Andy LaMora
Ryan Kon
Io Hanafusa
 (alamora)
(ryankon)
(iohana)


u s i n g    O N I S S
 
The novelty of ONISS is a user can find his/her laptop or PDA automatically populated with content and media that is popular in their specific location. This content will have been gleaned from passersby and nearby ONISS clients, and will include both material the user deliberatly sought, and content that was automatically provided. In the latter case, the content may fit a behavior profile for the user and thus seem likely to be wanted, or may merely be a "hot" item in that region.

Possible content for the ONISS system includes:
  • music media (links only, users must download with favorite download program, eg iTunes)
  • video media
  • browser bookmarks
  • documents (pdf, .doc, etc)
  • images
  • RSS news/media feeds
  • alerts (e.g. an NEB node dumps a message on every client in range)
  • personals

a b o u t    o u r    s o l u t i o n
 
Although ONISS could be successfully implemented as a strict peer-to-peer service, we propose using a centrally located application server to manage content ratings, and to efficiently provide clients with sources for desired content. That is, the server will perform the work of mapping one client's specific content requests to client sources on the network; individual clients need not compare content indexes, or download all content from every other client, when in "sight" of the server. This will also provide a convenient platform for viewing usage statistics. Additional features:

  In the absence of the server, clients will automatically default to peer-to-peer connections.

Client machines will designate one "incoming" folder, and one or many "shared" folders.

Content from any networked source will download to the "incoming" folder first. Content may be deleted directly from this folder, or removed via ONISS Cataloger (see link below).

To encourage a measure of randomness, content in "holding pens" will always be synchronized between in-range clients. Previously cataloged content will be excluded from synchronization; Client B will not download file X from Client A, if Client B has previously recieved and/or deleted file X.

The most popular content (objective rating to be determined) will also always be automatically copied between in-range clients

l e a r n    m o r e . . . 
 
More information : [ppt]
More information : [html]

Design Document (new browser) : [html]
Timeline (new browser) : [html]

Final Report (new browser) : [html]