Scholarlink

<p>Round Robin (RR) CPU scheduling is renowned for its simplicity and equitable allocation of computational resources among processes. However, the efficiency of RR scheduling is critically dependent on the selection of the time quantum. A fixed time quantum can lead to suboptimal performance, causing excessive context switching which increases overhead and waiting times for parallel processes. The given paper introduces a temporal convolutional network (TCN) model that dynamically predicts the optimal time slice for each process within a RR scheduling environment. The model is trained to estimate the ideal time quantum that minimizes the process completion time as compared to baseline linear regression model. The model is implemented within a Field-Programmable Gate Array (FPGA). Our modified RR algorithm integrates the FPGA-based predictor to assign individualized time slices dynamically, adapting to the specific characteristics of each process. In addition to this, the proposed TCN model attains the highest accuracy (95.67%) thus surpassing existing studies in terms of performance.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Temporal convolutional network (TCN) model for optimizing round robin scheduling in FPGA- based systems

  • 期刊论文

  • 发表日期：

    2025年7月2日

查看全文
  • B. Abirami,
  • V. Vasudevan

摘要

Round Robin (RR) CPU scheduling is renowned for its simplicity and equitable allocation of computational resources among processes. However, the efficiency of RR scheduling is critically dependent on the selection of the time quantum. A fixed time quantum can lead to suboptimal performance, causing excessive context switching which increases overhead and waiting times for parallel processes. The given paper introduces a temporal convolutional network (TCN) model that dynamically predicts the optimal time slice for each process within a RR scheduling environment. The model is trained to estimate the ideal time quantum that minimizes the process completion time as compared to baseline linear regression model. The model is implemented within a Field-Programmable Gate Array (FPGA). Our modified RR algorithm integrates the FPGA-based predictor to assign individualized time slices dynamically, adapting to the specific characteristics of each process. In addition to this, the proposed TCN model attains the highest accuracy (95.67%) thus surpassing existing studies in terms of performance.