Scholarlink

<p>Electrocardiogram (ECG) signal detection is a critical function in modern biomedical monitoring systems, requiring both high accuracy and efficient hardware utilization. Classical VLSI implementations tend to struggle with power consumption, hardware complexity, and delay requirements, hindering their scalability in portable or embedded health devices. To overcome these limits, the present study presents a Transpose Form Retimed Delayed Least Mean Square (LMS) filter-based VLSI architecture that is both area- and power-efficient, augmented with the Improved Progressive Cross sea lion Graph Contextual Convolutional Attention Network (Imp-PCSL-G2CAN). Imp-PCSL-G2CAN is a new hybrid model that combines Progressive Graph Convolutional Network (PGCN) with Cross-Contextual Attention Mechanism (CCAM). This framework facilitates efficient spatial–temporal extraction of ECG signal features with low computational complexity. The network parameters are optimized through the Improved Sea Lion Optimization (ISLO) algorithm to achieve accurate convergence and firm performance. Furthermore, the proposed model consumes only 10,167 LUTs and draws 0.4124W of power, making it the perfect choice for biomedical device systems that require less power. This layout maintains lower computational latency and improved detection performance while drastically reducing hardware resource utilization.</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

Low-Power VLSI Architecture for ECG Signal Detection Using Transpose Form Retimed Delayed LMS Filter with Improved Progressive Cross Sea Lion Graph Contextual Convolutional Attention Network

  • 期刊论文

  • 发表日期：

    2026年1月29日

查看全文
  • Loganathan Mohana Kannan,
  • Krishna Prakash Arunachalam,
  • Rudramani Bhutia,
  • Elangovan Muniyandy

摘要

Electrocardiogram (ECG) signal detection is a critical function in modern biomedical monitoring systems, requiring both high accuracy and efficient hardware utilization. Classical VLSI implementations tend to struggle with power consumption, hardware complexity, and delay requirements, hindering their scalability in portable or embedded health devices. To overcome these limits, the present study presents a Transpose Form Retimed Delayed Least Mean Square (LMS) filter-based VLSI architecture that is both area- and power-efficient, augmented with the Improved Progressive Cross sea lion Graph Contextual Convolutional Attention Network (Imp-PCSL-G2CAN). Imp-PCSL-G2CAN is a new hybrid model that combines Progressive Graph Convolutional Network (PGCN) with Cross-Contextual Attention Mechanism (CCAM). This framework facilitates efficient spatial–temporal extraction of ECG signal features with low computational complexity. The network parameters are optimized through the Improved Sea Lion Optimization (ISLO) algorithm to achieve accurate convergence and firm performance. Furthermore, the proposed model consumes only 10,167 LUTs and draws 0.4124W of power, making it the perfect choice for biomedical device systems that require less power. This layout maintains lower computational latency and improved detection performance while drastically reducing hardware resource utilization.