|
چکیده
|
The continuous evolution of Cyber-Physical Systems (CPS), particularly those empowered by the Internet of Things (IoT), has led to the generation of massive volumes of data and a growing demand for intelligent, real-time decision-making. Within such systems, efficient software-based task scheduling is vital to ensure timely responsiveness, optimal resource utilization, and sustained Quality of Service (QoS). Heterogeneous multi-server architectures offer a promising platform by enabling parallel processing and adaptive workload distribution. However, the inherent heterogeneity of computational nodes, coupled with the stringent temporal requirements of time-sensitive applications, imposes substantial challenges on software-level scheduling mechanisms. To address these challenges, this paper introduces a profit-aware and adaptive scheduling algorithm specifically designed for CPS environments comprising multiple input queues and heterogeneous servers. The proposed algorithm utilizes a greedy heuristic and a utility-based task modeling approach to dynamically allocate resources in accordance with current system state and task deadlines. Tasks that are completed within their deadline thresholds contribute positively to system profit, whereas delayed tasks incur penalties or are rejected. Extensive simulation experiments demonstrate that the proposed algorithm significantly enhances system responsiveness and improves overall system utility. In comparative evaluations against baseline and well-established scheduling strategies—including Random, Round Robin, MaxWeight, and MaxWeight with Discounted UCB—the proposed method achieves up to 30% reduction in average response time, 31% increase in total profit, and 50% improvement in deadline satisfaction rate. These results highlight the effectiveness of the proposed software-level scheduling approach in enhancing the operational efficiency of CPS in time-sensitive contexts.
|