CSC News
Zhishan Guo Receives RTAS Outstanding Paper Award
Department of Computer Science Associate Professor Zhishan Guo has co-authored a paper, “Jointly Ensuring Timing Disparity and End-to-End Latency Constraints in Hybrid DAGs,” that was recently selected as an Outstanding Paper at the 31st IEEE Real-Time and Embedded Technology and Applications Symposium (RTAS 2025). The conference, part of CPS-IoT Week (Cyber-Physical Systems and the Internet-of-Things), took place May 6-9, 2025 in Irvine, California.
Learn more about Guo’s research and the Cyber-Physical Systems Focus group at NC State.
Note to editors: the study abstract follows.
Authors: Jinghao Sun (Dalian University of Technology), Xisheng Li (Dalian University of Technology), Mingyang Gong (Dalian University of Technology), Nan Guan (City University of Hong Kong), Zhishan Guo (North Carolina State University), Mingsong Chen (East China Normal University), Jun Zhao (Dalian University of Technology), Qingxu Deng (Northeastern University)
Abstract: Autonomous machines often encounter complex timing constraints, such as those concerning end-to-end timing guarantees and real-time data fusion, etc. Tasks are often event-triggered or time-triggered at varying rates and exhibit data dependencies in between. Maintaining the real-time performance of autonomous machines becomes a highly challenging endeavor. In this paper, we formulate the workload of an autonomous machine as a hybrid Directed Acyclic Graph (DAG), which contains both time-trigger tasks and event-trigger tasks, with a distinct focus on the task of ensuring timing consistency in data fusion and adherence to end-to-end constraints within the DAG model. We design a concise mechanism to select suitable data received by a node and transmit them to successor nodes. This ensures both the timing disparity—as reflected by the differences in timestamps of the data used for fusion—and the end-to-end latency from the sensor to the controller is confined within a certain boundary. The proposed method is proven to be optimal as it always selects suitable data to guarantee the timing correctness of an autonomous machine as far as it (inherently) has the capacity. Experimental results show that our method can significantly improve the success rate of guaranteeing both timing consistency and end-to-end constraints of the autonomous machine.
Return To News Homepage