Real-Time systems must often handle several independent periodic macro-tasks, each one represented by a general tasks graph, including communications and precedence constraints. The implementation of such applications on a distributed system communicating via a bus, requires tasks assignment and scheduling, as well as the taking into account of the communication delays. As periodicity implies macro-tasks deadlines, the problem of finding a feasible schedule is critical. This paper addresses this NP-hard problem resolution, by using a genetic algorithm, under offline and non-preemptive scheduling assumptions. This algorithm performances are evaluated on a large simulation set, and compared to classical list-based algorithms, a simulated annealing algorithm and a specific clustering algorithm.