4. When all agents in community receive all the values, they choose the lowest execution time
Rj
sol T . The
agent As, who can afford this time take subtask qR, mark it as "working" and start its solution;
5. Agent As, requests the data required for the solution of task qR from all agents, which are owners of subset
of vertices connected with vertex qR by incoming arc;
6. If one of agents finishes decision of some subtask, it marks this subtask as "ready";
7. After solving every subtask agent use its computational complexity YR and time of decision Tsol to
determine the current performance of its CNs Sj, as Sj = YR / Tsol and sends the result values to all other
agents and to bulletin board;
8. Agent estimate characteristics of the community to find out if the task can be solved in specified time. If
community can not solve the task then agent mark it as "incomplete" and the creation of community restarts;
9. If there are any unsolved subtasks then go to step 1;
10. If the agent finishes solving of the last subtask in the user task it send the resulting data to the bulletin
board and then to the user.
The main benefit of proposed algorithm is adaptive process of computation.
4. When all agents in community receive all the values, they choose the lowest execution time
Rj
sol T . The
agent As, who can afford this time take subtask qR, mark it as "working" and start its solution;
5. Agent As, requests the data required for the solution of task qR from all agents, which are owners of subset
of vertices connected with vertex qR by incoming arc;
6. If one of agents finishes decision of some subtask, it marks this subtask as "ready";
7. After solving every subtask agent use its computational complexity YR and time of decision Tsol to
determine the current performance of its CNs Sj, as Sj = YR / Tsol and sends the result values to all other
agents and to bulletin board;
8. Agent estimate characteristics of the community to find out if the task can be solved in specified time. If
community can not solve the task then agent mark it as "incomplete" and the creation of community restarts;
9. If there are any unsolved subtasks then go to step 1;
10. If the agent finishes solving of the last subtask in the user task it send the resulting data to the bulletin
board and then to the user.
The main benefit of proposed algorithm is adaptive process of computation.
การแปล กรุณารอสักครู่..