Flexible Real-Time Algorithms Adapting Themselves To Alternating Load Situations

. In many applications, approximate results are often sufficient to achieve an acceptable behavior of a real--time system. Imprecise calculations could be necessary due to a transient overload situation caused by extraordinary events. To handle this exceptional work as well as the normal tasks, it is possible to reduce temporarily the computation time of the normal jobs by applying a more intelligent and flexible form of algorithms. In this paper, algorithm structures are presented, which are suitable to implement the mentioned methods. The basic idea is to offer multiple algorithm alternatives of different complexities for solving a problem. Then one can decide with the knowledge of the deadline and of the computation time which of the alternatives can be executed in time. These decisions will happen at runtime. To prove the applicability of the proposed methods on a real system, the theoretical considerations are devolved to a practical example consisting of a vehicle convoy. In this..