Automotive and industrial embedded systems are increasingly dependent on real-time capabilities. TSN aims to offer flexibility of the traffic by providing Ethernet with hard and soft real-time capabilities which allows for integration with other protocols in legacy systems. TSN requires the network to be fully synchronized to achieve high performance. However, there are cases where legacy systems are not able to synchronize with TSN. These systems might nonetheless be able to synchronize with each other through their legacy synchronization mechanisms. In this thesis, we have investigated effects in terms of jitter and clock drift in endpoints by synchronizing them with each other and passing communication through an unsynchronized intermediary TSN switch. Our results revealed that with the introduction of TSN, jitter was reduced, while clock drift between endpoints and the TSN switch was introduced. The results show that negative clock drift leads to packets missing their scheduled TSN windows and positive drift leads to packets being dropped in the switch buffer queues. We proposed two solutions in order to manage the experienced clock drift. In one solution we statically changed the switch cycle, and in the other, we let the receiver node dynamically update the sending period in the sender node. In the static solution, the clock drift was reduced from negative eight microseconds per second to two nanoseconds per second. In the dynamic solution, a packet error rate of one per 100 seconds was reduced to zero errors in 19 hours.
| Identifer | oai:union.ndltd.org:UPSALLA1/oai:DiVA.org:mdh-58563 |
| Date | January 2022 |
| Creators | Johansson, Andreas Johansson |
| Publisher | Mälardalens universitet, Akademin för innovation, design och teknik |
| Source Sets | DiVA Archive at Upsalla University |
| Language | English |
| Detected Language | English |
| Type | Student thesis, info:eu-repo/semantics/bachelorThesis, text |
| Format | application/pdf |
| Rights | info:eu-repo/semantics/openAccess |
Page generated in 0.002 seconds