OMNET++ simulation model for integrated modular avionics

Abstract

As the number and variety of electronic devices in aircraft continue to grow, the traditional federated architecture needs to be revised to meet these vehicles' size, weight, and power (SWaP) constraints. Integrated Modular Avionics (IMA) architecture has emerged as a promising solution for the SWaP problems. The IMA architecture optimizes the utilization of size, weight, and power by centralizing multiple application tasks onto a single hardware platform. When developing an IMA system, it is crucial to consider the relevant standards. ARINC 653 and ARINC 664 P7 (also known as AFDX) are two prominent standards that have garnered considerable attention and recognition within this context. However, these standards have numerous configuration parameters and offer various design options for engineers and designers. Therefore, determining the system configuration for optimal network performance is complex. In this regard, performing a significant portion of the IMA system design process in a simulation environment can efficiently conserve limited resources, including time and finances. This thesis proposes a simulation model of the IMA system to solve these issues. It is neither logical nor necessary to simulate all the rules defined by ARINC 653 and AFDX standards to measure the network performance of applications. Therefore, the first step of the thesis is to develop a system model of ARINC 653 concepts and AFDX devices to identify the necessary components used to measure communication performance. It is necessary to design components such as partition, partition manager, process, and process manager to manage avionic tasks according to ARINC 653 standard. The role of the partition manager component is to handle the initiation and termination operations of the required partition components based on the Major Time Frame (MTF). On the other hand, the partition component encompasses sub-components, including the process manager and process. Furthermore, the partition should relay the initiation and termination requests it receives to the process manager. As for the process manager, it executes the operations of stopping or starting process according to the received requests. In addition to managing avionic tasks, it is also necessary to have communication between the partitions to measure the system's communication performance. Therefore, it is essential to develop components that perform the sampling and queuing communication modes defined by the ARINC 653 standard. Two devices must be modeled for the AFDX standard: the End System (ES) and the Switch. The ES serves as the network device for communication between processes in the network. When a process wants to send a message, it writes the message to the sender communication ports within the ES. After writing the message, the device applies techniques such as BAG, data packet size compatibility, and redundancy management mechanisms to the packet before sending it to the network. On the receiving side, the ES receives the message from the physical link and performs operations like integrity checking and redundancy management. Then, the packet is written to the appropriate receiver communication port. When the receiver process is activated, it can read the message from the port. The switch device connects the ES in the network and performs filtering, policing, and switching tasks. While filtering ensures that packets comply with data packet size limitations, policing checks adherence to BAG rules. Also, switching determines the appropriate output ports for incoming packets. The devices and structures developed in the system model must be converted to the simulation model in a simulation environment. OMNET++ offers better scalability and extensibility compared to other simulation environments. Additionally, we can benefit from an active community, open-source code, and frameworks like INET that provide a detailed implementation of the OSI layer. That is why OMNET++ has been chosen as the environment for implementing a simulation model of the developed system model. In addition to the partition, partition manager, process, and process manager components specified in the system model, two additional components have also been developed for the simulation model of the ARINC 653 concepts: port channel and network transmitter driver. The port channel connects the receiver process to the receiver communication port of the ES, while the network transmitter driver is responsible for writing the data packets sent by the sender process to the correct sender communication port within the ES. For the AFDX device, a previously developed model [1] has been utilized. The same model from the paper has been used for the switch device. However, additional enhancements are required for the ES. In the existing model, the ES was designed solely for performing measurements at the device level, so no sender and receiver communication ports were designed. In addition to communication ports, the design does not include the demux component, which is responsible for writing messages to the appropriate receiver communication ports. The newly developed simulation model can handle packet reception and transmission operations by adding these two components to the previous simulation model. The developed simulation model is tested to determine their capability to handle packet reception and transmission tasks successfully. For this purpose, a scenario was created in the network model, consisting of two IMA modules referred to as sender and receiver. The sender module generates data packets and transmits them to the receiver module. The message integrity of the received data by the process of the receiver module is observed on the simulation console, confirming the successful execution of packet reception and transmission operations. It is essential to test not only the packet reception and transmission operations but also the timing of component executions to ensure proper functioning. A more advanced scenario from a previous study has been utilized [2]. Initially, the theoretical delays of the Virtual Link (VL) in the scenario are calculated using the application delay formula that includes software-based overheads. Then, the application delays are obtained through simulation, and it is observed that they converge with the theoretical values. Furthermore, the sender application in VL 1 of the scenario sends two data packets without any time gaps. However, since the design of the receiver and sender applications did not consider this, the current BAG value of the VL cannot prevent packet loss. The expectation is that reducing packet loss can be achieved by increasing the BAG value to match the period of the receiver application. This relation has been confirmed through testing with various BAG values. The consistency between the theoretical and simulated application delays and the expected relationship between BAG and packet loss indicates that the components are executed at the correct timing. The system design has been accurately transferred to the simulation environment, allowing for performance measurement in various IMA scenarios. Future developments can explore advanced technologies, such as using devices that support Time-Triggered Ethernet (TTEthernet) and Time-Sensitive Networking (TSN) instead of AFDX-compliant devices. Additionally, the ARINC 653 standard model has been designed to model a single processor system, but it can be further enhanced to enable parallel execution of applications and conduct complex measurements.