The details of static and adaptive fault tolerance policies are embedded into reusable agent plans using Hierarchical Task Network (HTN) formalism .
In adaptive fault tolerance policies, the replication degree and strategy are defined at runtime.
The aim of the leader agent is to be fault-tolerant against any crash failures; therefore, it has the Fault Tolerance plan and Adaptive Fault Tolerance plan in which adaptive replication is performed.
In order to apply an adaptive fault tolerance policy, the environment must be monitored to collect data and then data must be analyzed to adapt agent systems.
The quantity that defines a level of agent's criticality is sent within the content of a FIPA-ACL message to the leader agent(s) applying the adaptive fault tolerance policy.
In this performance study, the leader agents use an Adaptive Fault Tolerance plan (illustrated in Fig.
When the leader agent receives the message including its criticality from the adaptive replication manager, it executes the Adaptive Fault Tolerance plan and gets its criticality value as a provision (outlined in Fig.
Algorithm of the Adaptive Fault Tolerance plan PLAN (S_T): ADAPTIVE FAULT-TOLERANCE (T_1< T_2<T_3)[LAMBDA](S_T, S_T(Provision), T_1,T_1(Provision)) [LAMBDA](T_2, T_2(Outcome), S_T,S_T(Outcome))[LAMBDA](T_3, T_3(Outcome), S_T,S_T(Outcome)) [LAMBDA](T_1, T_1 (Outcome_1),T_2, T_2(Provision)][LAMBDA](T_1, T_1 (Outcome _2),T_3, T_3 (Provision)) 1.