Application of genetic algorithms for the design of effective and comprehensive schedules for the processing of data packets in a pipeline system at specific time intervals

Abstract

The article suggests a solution to the problem of designing effective data processing schedules with the processing pipeline functioning at specific time intervals. The problem of creating a comprehensive schedule is divided into three subtasks, each being solved at a certain level of the hierarchy. The first level determines the composition of the data packets. The second level determines the composition of groups of data packets, which are processed at specified time intervals. The third level determines the schedule of processing of the data packets included in each group. To solve the problem of creating comprehensive schedules within this hierarchy, the hierarchical game theoretic approach was applied. The article describes a hierarchical game model used to determine local optimal solutions for each level of the hierarchy. It also demonstrates that genetic algorithms can be used to determine the optimal composition of groups of data packets processed during specified time intervals in order to minimise the downtime of the pipeline segments. We propose a method to encode the decision making process regarding the composition of the data packets groups. The study determined the functioning of genetic operators which are used to search for local optimal solutions regarding the composition of the groups of packets based on the initial decision. We also developed software which optimises the composition of data packets and the composition of groups of data packets, and creates the processing schedule for the packets in the groups. The study demonstrated that the use of genetic algorithms in order to optimise the composition of groups of data packets allows for a 45 % increase in the efficiency of using the time resource of the pipeline segments as compared to the initial solution.