Read Aloud the Text Content

This audio was created by Woord's Text to Speech service by content creators from all around the world.

Text Content or SSML code:

Hello everyone, my name is Muhammed Fatih ELGİN. Today I want to talk about my master thesis work; a novel approach to reducing worst case response time in avionics full duplex switched ethernet networks. I work with Ece hoca and Klaus hoca for my thesis work. Next; Avionic is a term that is a mix of aviation and electronics. It refers to the electronic systems used on aircrafts. As you can see from the picture, there are lots of different tasks that needs to be done by the avionic systems. For example, the auto pilot navigation system can help the pilot by driving autonomously at any point. To do that, auto pilot needs performance data such as speed, torque, altitude or the position of the plane. Each of these information is created by another system and each system requires different information to be able to work. This requires a structure that can support large volume of data to be transmitted between devices. Over time, there are lots of different protocols that are used for this purpose. Next; Here, you can see two common protocols. The first one is arinc-429 protocol. İt is a point-to-point protocol where only one transmitter sends messages and one receiver receives this message. By multiplying the wires here, we can have multiple receivers. Its transmission speed is very limited, at most 100-kilo bit per second. The other commonly used protocol is MIL-STD-fifteen fifty-three. It has a data bus and a bus controller. When bus controller allows, the remote terminals can send data to the bus. Each device that is connected to the bus can receive these messages. There is duplicate two channels because of redundancy. It has up to 1 mega bit per second data rate so it is better than Arinc 429 standard but this data rate is affected by the length of the bus as well as other things. Also, when the bus controller fails, the bus fails as well. Next; As the technology increases, we want more things from machines. For example, 10 years ago, we have analog indicator in cars, everything is manual. No autonomous driving or no automatic braking system. The situation was similar in aircrafts. Nowadays, all these actions are done automatically by the devices. they have more calculation capabilities, more functionality. To do that, each device needs more information about the aircraft status. This brings us to the challenges in Avionic Communications. The most important one is scalability. As we talk before, the technology advances rapidly. In each day, the functionalities that we want from aircraft increases. We need a structure that is scalable to keep pace with advancement. We can add another device easily without changing anything, which also means the complexity actually. Another important challenge is data transmission speed. The aircrafts need to react any factor from the environment as quickly as possible. For an aircraft, the decision time should be in milliseconds. Furthermore, since the volume of data that is transmitted between the avionic devices increased, we need to increase the speed as well. Last challenge is the reliability. We need to make sure that the message is transmitted to the destination no matter what. Next; Avionic full duplex ethernet networks present a solution for these challenges. It is a special version of the ethernet technology with some key modifications to be used in avionics. Ethernet is a very stable technology. It is widely used and there are lots of people that are knowledgeable about it so the transition to this protocol can be very smooth. It has hundred mega bit per second data rate. It is scalable and flexible as the classic ethernet technology. There are 3 main components of this network. The end systems are the avionic devices that sends and receive messages. The switches are the transfer and the links are physical medium between these two. The network is redundant because as you can see from the picture, there are two duplicate networks. End systems send packets to both of these networks. The destination end systems drop the packets comes later. Next; AFDX is a protocol that is firstly proposed by airbus. It is now an IEEE standard. It is implemented in aircrafts. As you can see from the picture, which shows Airbus A380 AFDX network, the network can be very complex and AFDX can handle it. Next; Before going into detail, I want to talk about some terminologies. First one is Virtual Links. They are statically defined connections between a single source end system and one or more destination end systems. The routing in switches is done according to these virtual links. Next; Next one is Bandwidth Allocation Gap, BAG parameters. It is the minimum delay between two consecutive frames of the associated virtual link. As you can see from the picture, the end system put the message to the transmit queue after at least BAG time. Here, the F22 message is delayed because of the BAG values. Next; Now, I want to talk about the Response Time. It is the time that a packet is transmitted from the source to the destination end systems. Bas; It includes all the delays from the end systems as well as switches. Next; Here, you can see an example queueing delays. Bas; Firstly, the VL1 packets comes to the switch and starts transmitting to the other switch. Bas; While transmitting, the VL3 packets comes. Since the line is busy at the moment, the VL3 packet needs to wait until VL1 is completely transmitted. This is one of the sources of the delay. Next; The Worst-Case Response Time is maximum time that a packet is transmitted between the source and destination end systems. It is an important metric while designing the AFDX network. The important delay contributors are given here. The transmission delay is the time that a packet is transmitted on the line. It depends only the line rate so we can’t optimize it. The technological delays are caused by switching fabric while forwarding the packet so we can’t optimize it as well without changing the switches. The Propagation delay is the time it takes for the packet to travel from the sender to receiver. It depends on the length of the line so we can’t change it. The last cause is the queueing delay. It is the delay when the line is busy and another packet comes as I showed before. This is the delay that we will try to minimize. Next; My thesis work is continuation of another work. In that work, the exact worst case response time is calculated. From that work, we find out that the burst of packets and queueing delay is the main cause the worst-case response time. You can see from the picture that the packets m1 and m2 is transmitted back-to-back. When more than one such end systems with long bursts are present, we observe longer response times. In this research, we are introducing Offset Scheduling Mechanism. It is a concept that is implemented in another protocol, named CAN. It assigns a unique time gap between messages from the same end system and tries to prevent the queueing delay by reducing burst of packets. As you can see from the picture below, by introducing and efficient offset, the burst of packets can be minimized. Next; At the moment, we are developing a worst-case response time calculation method when offset scheduling is utilized. In the previous work, this method was developed but since there is limited burst of packets and offset scheduling mechanism, that method cannot be used directly here. After that, we will develop an efficient offset assignment algorithm. Similar algorithms were developed for different protocols. However, since the dynamics of protocols is different, the already developed algorithms may not work as efficiently as possible. Last Step will be the simulation and verification step. We are going to use real time scenarios and simulation tools such as OMNeT++ to show that our worst-case response time calculation method works. We will also show that using offset scheduling can decrease the worst-case response time. Next; In conclusion, we believe that the worst-case response time can be optimized using offset scheduling. The key contributions of this research are introducing offset scheduling mechanism in AFDX, Worst case Response Time Calculation Method with offset scheduling and efficient offset assignment algorithm. From the hand written calculations, we observed that the average and worst-case response time can be improved. As we finish our work, we expect to see better simulation results as well. Next; Here are my references. Next; Thank you for your attention.