A multi-objectives scheduling algorithm based on cuckoo optimization for task allocation problem at compile time in heterogeneous systems

• A novel algorithm for task scheduling in heterogeneous systems is proposed.
• We use an extended Cuckoo Optimization Algorithm (COA) to solve the problem.
• Defining an efficient immigration function to escape from local optimums.
• The results show the proposed algorithm superiority over the previous algorithms.

To handle scheduling of tasks on heterogeneous systems, an algorithm is proposed to reduce execution time while allowing for maximum parallelization. The algorithm is based on multi-objective scheduling cuckoo optimization algorithm (MOSCOA). In this algorithm, each cuckoo represents a scheduling solution in which the ordering of tasks and processors allocated to them are considered. In addition, the operators of cuckoo optimization algorithm means laying and immigration are defined so that it is usable for scheduling scenario of the directed acyclic graph of the problem. This algorithm adapts cuckoo optimization algorithm operators to create proper scheduling in each stage. This ensures avoiding local optima while allowing for global search within the problem space for accelerating the finding of a global optimum and delivering a relatively optimized scheduling with the least number of repetitions. Moving toward global optima is done through a target immigration operator in this algorithm and schedules in each repetition are pushed toward optimized schedules to secure global optima. The results of MOSCOA implementation on a large number of random graphs and real-world application graphs with a wide range characteristics show MOSCOA superiority over the previous task scheduling algorithms.