Design of DIMA scheduling algorithm based on network partition integrating model

Distributed Integrated Modular Avionics (DIMA) system can be seen as a distributed integration of sub-function areas with integrated modular structure and switched connection on airborne environment. During the process of integration, DIMA pays more attention to the consistency performance of mission-critical and safety-critical guarantee capability among integrated core processor (ICP) and universal avionic network (UAN). However, the separated scheduling design of ICP and UAN causes the problem of consistency guarantee ability interrupt. Therefore, we propose a scheduling algorithm based on network partition integrating model to realize holistic scheduling design. In this paper, we firstly present network partition integrating model, which is based on spatial partition and bandwidth partition. The network partition integrating model is an extension of network partition technology and provides the infrastructure of time scheduling design. Based on network partition integrating model, we propose DIMA scheduling design. ICP is designed as two-level scheduling. In first scheduling level, each partition is activated one by one by using weighted round robin mode. Second scheduling level uses time-triggered scheduling design. As an intermediate node of DIMA network, the scheduling method of TT tasks and RC plus BE tasks is separately designed in UAN. According to DIMA scheduling design, we also proposed corresponding DIMA scheduling algorithm. The scheduling algorithm consists of end-system scheduling algorithm and switch scheduling algorithm. For the DIMA system, there are three steps to configure time scheduling table: 1) configure the end-system within IMA; 2) configure the ports of switches within IMA; 3) configure the ports of switches of UAN. This scheduling algorithm solves the mismatch problem of separated scheduling about ICP and UAN, realizes the holistic schedule of ICP and UAN and satisfies the real time performance of avionics tasks. Finally, the simulation results of DIMA network show the schedulablity and determinism of the DIMA scheduling algorithm.