Interconnection Optimization for Multi-cluster Avionics Networks

Abstract

The increasing complexity and heterogeneity of avionics networks make resource optimization a challenging task. In contrast to many previous approaches pursuing the optimization of traffic-source mapping and backbone network analysis, our work presented herein mainly focuses on the optimization of interconnection devices for multi-cluster avionics networks. In this paper, we introduce an optimized interconnection device, integrating novel frame packing strategies and schedulability analysis to enhance the communications between an AFDXlike backbone network and various peripheral sensor/actuator networks in terms of resource savings. The performance analysis conducted on a representative avionics communication architecture highlights the efficiency of our proposal to save resources particularly consumed bandwidth. These latter is considered as an important feature for avionics applications to guarantee easy incremental design during the long lifetime of an aircraft.

Specification and Performance Indicators of AeroRing—A Multiple-Ring Ethernet Network for Avionics Embedded Systems

“…The second class consists of either solutions working on the current ADCN architecture as in [ 21 ] or defining new architecture solutions as in [ 22 ]. In [ 21 ], a new CAN-AFDX Remote Data Concentrator (RDC) device was proposed to enhance the network bandwidth use, while meeting the timing constraints.…”

“…The second class consists of either solutions working on the current ADCN architecture as in [ 21 ] or defining new architecture solutions as in [ 22 ]. In [ 21 ], a new CAN-AFDX Remote Data Concentrator (RDC) device was proposed to enhance the network bandwidth use, while meeting the timing constraints. Two major functions were adopted to reach this goal: (i) frame packing of the upstream flows coming from the CAN sensors, to reduce communication overheads, and consequently decreasing the AFDX bandwidth use; (ii) hierarchical Traffic Shaping (HTS) of downstream flows destined to actuators on CAN buses, to guarantee isolation between upstream and downstream flows on each I/O CAN bus, and consequently favoring the frame packing process.…”

Scheduling of rate-constrained and time-triggered traffic in multi-cluster TTEthernet systems

“…In our use case, this normal distribution is selected randomly from one of the following: U(0.9ms, 1 …”

“…There is an ongoing trend to real-time Ethernet in many domains such as avionics [1], factory automation [2] and automotive networks [3]. Due to increasing functionality along with corresponding resource requirements, multi-cluster realtime Ethernet networks are envisioned.…”

“…1 This work has been supported by the EU project DREAMS (No. 610640) A major technical challenge is the mapping of the application subsystems with their computational components and communication activities to such a multi-cluster platform, while satisfying real-time and reliability constraints.…”