Gesamtheitlicher Ansatz von Kommunikations- und Steuersystemen in Cyber Physical Systems unter Ausnutzung von adaptiven Kommunikationsnetzwerken (eNC2)
Zusammenfassung der Projektergebnisse
The project eNC2 was advocating for joint design of communication and control (C&C) in Cyber Physical Systems (CPS) exploiting adaptive networking concepts to enable global (and not local) CPS. In this regard, we considered networked control of UAVs, which poses a complicated flight state estimation and control problem. We first focused on the problem of flight state estimation which corresponds to a sensor fusion problem using extended Kalman filters. We investigated accurate and reliable estimation for the position, velocity, attitude, and the 3D wind in real-time for fixed-wing UAVs by fusing noisy observations from multiple sensors. In this context, we also developed a general framework and testbed for testing flight state estimation algorithms forming the basis for the study of joint design of C&C systems of UAVs based on real measurements and experiments. The estimator has been evaluated based on a complete set of recorded sensor data. The data has been analyzed in two testbeds, i.e, in MATLAB and on a commercial glider avionics hardware. We have observed that the nonlinear estimator shows excellent tracking and convergence behavior even if subject to strong perturbations. The result has been achieved without the use of a magnetometer. This is of great practical significance since the magnetometer is the weak part as it is significantly perturbed by other magnetic fields. We analyzed the observability based on the singular value decomposition (SVD) of a simplified 2D system, which shows that the system state is observable as soon as the true airspeed is changing. This condition is true in any application due to the random nature of the air mass, as shown by the numerical observability analysis based on the SVD of the complete system. We studied the fundamental limits of control systems attached to communication network scenarios and dealt with massive number of controllable entities. To demonstrate the combined C&C approach, we implemented our own testbed to study the role of agile hierarchical decision making by means of coexisting inverted pendulums. We relied on multiple controllers closer to the CPS, since the decision-making entity can be placed in different locations on-the-fly, as trade-off between delay and accuracy/reliability. Those controllers can be in the middle of the network, or even directly on the base station. Network Coding (NC) was implemented on an inverted pendulum model to verify the impact of NC on information transmission in Networked Control Systems. Implementations under different packet loss ratios were simulated to compare the performance of the original system with the system using NC and Erasure Correcting Codes. The biggest benefit of using NC is that a network node is no longer required to gather all data packets one-by-one. One suitable approach to overcome the effects of delays, packet losses, and bit-flips in the communication channels is placing multiple replicas of dedicated controllers and higher control nodes of several CPS’ (inverted pendulums in this case), to make sure that the actuators get the required input on time. With multiple controllers several signal-selection models were implemented and analyzed to find an appropriate solution for controlling coexisting CPS. Different models were tested, showing the ”Latest Signal” and ”Similar Signal” models outperformed the others in several scenarios. From the test results, we also found that bit-flips are always the killer of the system, which should be taken seriously. The packet loss seems to be less important than bit-flip and latency. With higher latency, the packet loss tolerance also drops severely. The demonstrator “The Classical Pendulum Meets the Network” was shown at the 3rd IEEE 5G Summit Dresden 2018.
Projektbezogene Publikationen (Auswahl)
-
Reducing latency in virtual machines: Enabling tactile internet for human-machine co-working,” IEEE Journal on Selected Areas in Communications, vol. 37, no. 5, pp. 1098–1116, 2019
Z. Xiang, F. Gabriel, E. Urbano, G. T. Nguyen, M. Reisslein, and F. H. Fitzek
-
“Towards a better understanding of live migration performance with docker containers,” in European Wireless 2019; 25th European Wireless Conference. VDE, 2019, pp. 1–6
R. Torre, E. Urbano, H. Salah, G. T. Nguyen, and F. H. Fitzek
-
“Jointly virtual and physical prototype design and testing of a sensor fusion platform for UAVs,” IEEE International Conference on Unmanned Aircraft Systems (ICUAS), Sep. 2020
P. Huang, H. Meyr, and G. Fettweis
-
“Observability analysis of flight state estimation for UAVs and experimental validation,” IEEE International Conference on Robotics and Automation (ICRA), Jun. 2020
P. Huang, H. Meyr, M. D¨ orpinghaus, and G. Fettweis
