On-Line Scheduling Multiprocessor Tasks in the Non-predictive Environment

The paper considers the problem known in the literature as the multiprocessor task scheduling. It appears in the context of embedded systems and devices working in critical civil and military infrastructure objects, Internet of Things, Internet of Vehicles and so on, with an augmented level of security. Enlarged reliability is achieved through hardware and software redundancy. A number of independent devices (microprocessors) which execute the same programs for identical data, provide independent results to set the final decision by “voting”. In the paper we assume that the set of tasks for processing is uncertain while tasks arrive randomly. We provide some on-line scheduling algorithms with the theoretical evaluation of their quality.