source: fedd/abac-src/ttg/README @ 93d4442

ABAC Demo Distribution README
$Date: 2003/07/16 16:26:48 $

documents - contains papers and presentations relating to ABAC.

rbtm - contains java source for a credential management system which is used by
       the by the ABAC packages. This code base was originally developed at 
       Stanford University with modifications made at Network Associates 
       Laboratories. For more information, see the automatically generated 
       documentation (javadoc) in the docs directory.

        * IntersectionProofNode currently only supports backward searching.  
          Forward searching is need for a complete implementation of 
          distributed discovery.

ttg -+  contains java source code for the the ABAC packages. This code base 
     |  was developed at Network Associates Laboratories. 
     | 
     +-- docs - automatically generated javadoc html goes here.
     |
     +-- test - contains launch script and *.txt configuration files.
     |    |     Configuration files are grouped into different scenarios and
     |    |     may also contains scripts "go*" or test to launch a test.
     |    |
     |    +-- simple - simple Swedish and EPub (e-commerce) scenarios
     |    |
     |    +-- failed - examples of failure cases
     |    |
     |    +-- intersect - simple policies extended to include intersections
     |    |
     |    +-- linked - intersection policies extended to included linked
     |    |            roles with and without linking goals.
     |    |
     |    +-- heuristic_strategy - two scenarios: sweden1 and usdod, each 
     |    |                        are designed to exercise the heuristic,
     |    |                        stingy strategy (see below)
     |    |
     |    +-- rtml - examples similar to linked above which incorporate RTML
     |               credential stores and distributed discovery of RTML 
     |               credentials using distributed disovery
     | 
     +-- demo - tomcat servlet source code and configuration and distributed
     |          discovery servlet which allows access to the discovery service.
     |
     +-- INSTALL - installation notes for building and installing the demo

atnvis -+ contains the visualization package in atnvis.jar (also relies on 
        | grappa.jar from AT&T Research  which is available at 
        | http://www.research.att.com/~john/Grappa/ )
        |
        +- demo - the demo configuration (*.prop) files generated from the 
        |         TestEngine class.
        |
        +- graphviz - the grappa java packages rely on GraphViz. Windows and 
        |             Linux binaries are included for convenience. Grappa and
        |             GraphViz are licensed identically. Please see the README
        |             file in this directory for more information.
        |
        +- strategy_comparison - visualization results for using various 
                                 strategies in the same scenario (USSH vs 
                                 SAdmir).


Strategy Configuration 
======================

The test engine supplied with this package can be easily configured with text 
files. The [Strategy] section has several configurable parameters; for example,
add this on a single line "type=default".  Note the left hand side parameter 
should be all lower case.

    type - set the type of strategy to use: "default" or "stingy", where the
           the default is a breadth first search. Stingy refer to the 
           history-based, heuristic strategy. The remaining parameter types
           are only useful when the "stingy" strategy is selected.

    weightFile - a file name for loading and storing the weights used in the 
                 stingy strategy's algorithm.

    init - a single-precision floating point number for the initial value of
           new weights.
    
    sat - a single-precision floating point number to which a role's weight 
          is set if the role is contained in a target that is satisfied at the 
          end of a negotiation. 
  
    training - "true" or "false" to tell the stingy strategy whether to 
               save the set of weights generated by the current negotiation