Middleware and Architecture for Advanced Applications of Cyber-physical Systems
MetadataShow full item record
In this thesis, we address issues related to middleware, architecture and applications of cyber-physical systems. The first problem we address is the cross-layer design of cyber-physical systems to cope with interactions between the cyber layer and the physical layer in a dynamic environment. We propose a bi-directional middleware that allows the optimal utilization of the common resources for the benefit of either or both the layers in order to obtain overall system performance. The case study of network connectivity preservation in a vehicular formation illustrates how this approach can be applied to a particular situation where the network connectivity drives the application layer. Next we address another aspect of cross-layer impact: the problem that arises when network performance, in this case delay performance, affects control system performance. We propose a two-pronged approach involving a flexible adaptive model identification algorithm with outlier rejection, which in turn uses an adaptive system model to detect and reject outliers, thus shielding the estimation algorithm and thereby improving reliability. We experimentally demonstrate that the outlier rejection approach which intercepts and filters the data, combined with simultaneous model adaptation, can result in improved performance of Model Predictive Control in the vehicular testbed. Then we turn to two advanced applications of cyber-physical systems. First, we address the problem of security of cyber-physical systems. We consider the context of an intelligent transportation system in which a malicious sensor node manipulates the position data of one of the autonomous cars to deviate from a safe trajectory and collide with other cars. In order to secure the safety of such systems where sensor measurements are compromised, we employ the procedure of “dynamic watermarking”. This procedure enables an honest node in the control loop to detect the existence of a malicious node within the feedback loop. We demonstrate in the testbed that dynamic watermarking can indeed protect cars against collisions even in the presence of sensor attacks. The second application of cyber-physical systems that we consider is cyber-manufacturing which is an origami-type laser-based custom manufacturing machine employing folding and cutting of sheet material to manufacture 3D objects. We have developed such a system for use in a laser-based autonomous custom manufacturing machine equipped with real-time sensing and control. The basic elements in the architecture are a laser processing machine, a sensing system to estimate the state of the workpiece, a control system determining control inputs for a laser system based on the estimated data, a robotic arm manipulating the workpiece in the work space, and middleware supporting the communication among the systems. We demonstrate automated 3D laser cutting and bending to fabricate a 3D product as an experimental result. Lastly, we address the problem of traffic management of an unmanned aerial system. In an effort to improve the performance of the traffic management for unmanned aircrafts, we propose a probability-based collision resolution algorithm. The proposed algorithm analyzes the planned trajectories to calculate their collision probabilities, and modifies individual drone starting times to reduce the probability of collision, while attempting to preserve high performance. Our simulation results demonstrate that the proposed algorithm improves the performance of the drone traffic management by guaranteeing high safety with low modification of the starting times.
Unmanned aerial system traffic management
Ko, Woo Hyun (2017). Middleware and Architecture for Advanced Applications of Cyber-physical Systems. Doctoral dissertation, Texas A & M University. Available electronically from