High level data flow scheduling method with time constraints

In order to resolve problems inherit in high level data flow scheduling with time constrains,a method dynamically combining a genetic algorithm(GA) and an ant algorithm(AA) was developed.Encoding methods,crossovers,mutations,and the fitness function of the GA were evaluated,as well as probability selections and pheromone update rules for the AA.To determine the optimal opportunity for a switch from GA to AA,two critical problems had to be resolved: the first was a means to dynamically determine termination conditions for the GA;the second was a method for using the scheduling results of the GA to generate the initial pheromone distribution of the AA.Experimental results showed that,on average,the resources needed in the proposed method were 5.2% less than with GA alone and 4.9% less than with AA alone.The run time using the proposed method was 44% less than with GA and 31% less than with AA.