In general, method of solving of incoming task in decentralized DCS can be described as follows. User of
the system forms task he want to solve, sets requirement for the time to solve the task and determines a virtual
payment for solving, which is represented by some virtual points (it can be money if we use private computers
or just priority of task if we use corporative CNs). After that user sends all the data to a BB. Proactive agents
of the DCS are monitoring BBs and look for new tasks. If some agent finds new tasks, it use task’s
computational complexity and the number of virtual points to evaluate profit of the performing of each task.
When agent finds the most "beneficial" task, it joins the community to solve it: it takes some part of
computations required to solve the task. Every time new agent joins community, it evaluated time of task
solving, and at some point of time, the community becomes powerful enough to solve the task in time and the
community stops its expansion. When new agent joins the community, it take some subtask and begin to solve
it. While solving the subtasks agents are monitoring their CNs’ parameters. If variations of parameters cause
exceeding of user-specified time of task solving, the community begin accepting new agents. When
community solves the task, the agent, that produce the result, send it to the user, and then make the
community to dissolve itself [14]
In general, method of solving of incoming task in decentralized DCS can be described as follows. User of
the system forms task he want to solve, sets requirement for the time to solve the task and determines a virtual
payment for solving, which is represented by some virtual points (it can be money if we use private computers
or just priority of task if we use corporative CNs). After that user sends all the data to a BB. Proactive agents
of the DCS are monitoring BBs and look for new tasks. If some agent finds new tasks, it use task’s
computational complexity and the number of virtual points to evaluate profit of the performing of each task.
When agent finds the most "beneficial" task, it joins the community to solve it: it takes some part of
computations required to solve the task. Every time new agent joins community, it evaluated time of task
solving, and at some point of time, the community becomes powerful enough to solve the task in time and the
community stops its expansion. When new agent joins the community, it take some subtask and begin to solve
it. While solving the subtasks agents are monitoring their CNs’ parameters. If variations of parameters cause
exceeding of user-specified time of task solving, the community begin accepting new agents. When
community solves the task, the agent, that produce the result, send it to the user, and then make the
community to dissolve itself [14]
การแปล กรุณารอสักครู่..