Collaborative Resource Constraint Scheduling with a Fractional Shared Resource

We consider a collaborative scheduling problem motivated by mining in remote off-grid areas. In our model, jobs are assigned to processors who each have their own machine for executing them. As each job needs a certain amount of a resource shared between the processors, a coordination mechanism between the processors is needed. We present a framework which collaboratively computes a schedule while exchanging only limited information between the processors and a central resource manager. Our computational experiments show that our negotiated approach outperforms a one-shot solution approach by a wide margin and produces fairer solutions than a centralised genetic algorithm that can make use of the private information of each processor. Depending on the number of processors, the solution quality found by the mechanism presented in this paper is competitive with or even better than that of the centralised genetic algorithm.