Download APNN Description of the model.
Example Models for APNN-ToolBox
This Net is divided into two sections(Left-section (refered as l) and Right-section (refered as r)). Figure shows an algorithm based on message passing. The essential concept of the mutex algorithms in fig. is a token that at each reachable state is held by one of the sites. A site may go critical only while holding the token. The site without token may gain it on demand. The token is initially held by site_l in the local state avail_l. With the token in avail_l, the site_l is available to move immediately from pending_l by action a to critical_l. With action e the site_l then returns from critical_l to quiet_l and makes the token again available for l. Furthermore, l may receive a request for the token sent by the site_r along requested_l. Fairness of c guarantees that l eventually sends the token to granted_r and coincidently turns silent_l.The request sent by site_r along requested_l is due to an occurrence of action h. Hence site_r is at waiting_r until the token on granted_r arrives. Occurrence of action k then brings the site_r to critical_r. Site_l may meanwhile be pending again. As site_r is now the owner of the token, site_l is in silent_l and may send a request to r by occurrence of b. The two sites l and r are structurally symmetrical. The initial state, however, is not symmetrical, as site_l initially carries the token (at avail_l), whereas site_r is at silent_r. Site_l (and likewise site_r) is no sequential machine. Actions f and c (actions n and j in site_r) may very well occur concurrently. The two sites cooperate by message passing, with two types of messages (requested and granted) in each direction. Site_l has one fair action, c. The corresponding conflict place avail_l is not even read by site_r. Symmetrically, the conflict places of j is avail_r, not read by l. Site_l is fault tolerant with respect to actions a and b. Site r's iterated access to critical_r is not affected in case a or b maliciously remains enabled forever.
Download APNN Description of the model.