We view the environment as a system consisting of a finite collection of
principals.
We define for a principal P
a local state
as a tuple , with the intuitive
interpretation:
, the set of formulas that P currently believes;
, the set of (sub-)messages P once said;
, the set of messages that P has seen so far;
, the set of keys P possesses.
It is closed if it
satisfies the following (mutually defined) closure
properties, each of which corresponds directly to an axiom:
(Saying parts of a joint message)
If a principal said a combination of messages at a certain time, then that
principal said each of the messages as well:
The reverse does not hold, since the presence of a joint message in
implies that both components were uttered (as a joint message)
at the same time;
(Saying contents of an encrypted message)
If a principal said an encrypted message and believes the key is
good, then that principal said the contents of the encrypted message as
well:
(Decryption)
If a principal P possesses a key K , and if P sees a message X
labeled with Q and encrypted with K , then P also sees X itself:
(Note that there is no closure property corresponding to
the axiom Good key ensures the utterer.)
The closure of a local state s is the least closed local
state s' such that , where the
ordering is obtained by component-wise lifting of the set ordering.
Note that taking the closure only adds elements to
the sets involved, and leaves an already closed local
state unchanged.
A global state is a mapping from principals to
local states (for each principal in the environment).
Global states are ordered by lifting the ordering on
local states.
The closure of a global state s , denoted by
, is defined in the obvious way.
The unqualified term ``state'' will, from now on, mean a
closed global state.
We consistently use the convention that for a state denoted by the
variable s its local state for principal P is
denoted by the tuple ,and likewise that for a state s'
its local state for principal P is
denoted by the tuple