Example Models for APNN-ToolBox
This Net is divided into two sections(Left-section (refered as l) and Right-section (refered as r)). This algorithm for mutual exclusion, as shown in fig. 13.6, is extended from ordered crosstalk algorithm. The ordered occurrence of crosstalk_l and crosstalk_r implies that finished_l and finished_r never carry a token at the same time. Figure 13.6 refines finished_l into crit_l, action n and terminated_l. Finished_r is refined accordingly. Local states and actions are renamed according to their new role in 13.6, and further components (quiet_l, m, pend1_l,quiet_r, p, pend1_r) are added, providing the elements as required in fig. 13.1, the system for fig. 13.6 operatesin rounds. Wanting to go to critical, site_l sends a request to site_r by action a and remains inpend2_l until critical in both cases by occurrence of action b or action j, respectively. Site_r likewise may send a request to l by action g, then r remains in pend2_r, until site_l reacts with a token on granted_l , or requested_l. Site_r becomes critical in the first case by action h. The second case occurs in the situation of crosstalk, where both sites strive at their respective critical state in the same round. Site_r has to wait in this case until l leaves crit_l and sends a token to terminated_l. The two sites l and r are structurally not symmetrical: l precedes r in case of crosstalk. Site_l is no sequential machine, as n may occur concurrently to e and f. In site_r, the action q may likewise occur concurrently to c and d. Fairness of action a guarantees that site_l in state pend_l will eventually become critical. The corresponding conflict place, local_l, is not read by site_r. Symmetrically, the conflict place local_r of the fair action g of site_r is not read by site_l. Site_l is fault tolerant only with respect to action a. Due to the round-based nature of the algorithm, each step of site_r to crit_r must explicitly be granted by l. Vice versa, each step of l to crit_l must be granted by r.
Download APNN Description of the model.