Anuraj V.
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Anuraj V. has a total of 4 co-authors. Here are the top co-authors by number of shared publications:
- Dhandapani Vaithiyanathan
National Institute of Technology Delhi, Delhi 110036, India
5 shared publications - Sanjeev Ramkumar
Department of Electronics and Communication Engineering National Institute of Technology Delhi Delhi Delhi India
2 shared publications - Neeraj Sinha
National Institute of Technology Raipur
1 shared publication Nitin Kumar
1 shared publication
Latest Publications
- 1 September 2026
- Published by Wiley in International Journal of Numerical Modelling: Electronic Networks, Devices and Fields
- Vol. 39 (5)
Application‐Specific Vedic Multiplier Architectures for Area and Power Efficient VLSI Implementation
Multiplication and squaring are fundamental operations in digital signal processing (DSP), image processing, and embedded computing, where hardware efficiency directly affects speed, area, and energy consumption. Conventional multipliers often introduce unnecessary complexity when operands follow specific patterns. This paper presents a systematic study of selected Vedic mathematics sutras, namely Urdhva‐Tiryakbhyam, Ekanyunena Purvena, Ekadhikena Purvena, Sopaantyadvayamantyam, Yaavadunam, and Nikhilam Navatashcaramam Dashatah, focusing on their suitability for practical computational patterns. The study examines multiplication by constants, operands near base values, and range constrained squaring relevant to DSP applications such as signal energy computation, RMS estimation, filtering, image processing, and neural networks. Corresponding general‐purpose, operand‐pattern‐optimized, fixed‐coefficient, and range‐constrained architectures are implemented and evaluated. FPGA implementations require 6‐117 LUTs with 5.144–11.012 ns delay, while ASIC implementations achieve 1408.491–19 043.73 μm 2 area and 0.565877 × 10 5 to 9.65297 × 10 5 W power. The results show that no single sutra is universally optimal; hardware efficiency depends on operand characteristics and application requirements. The proposed framework therefore enables systematic selection of Vedic multiplication architectures based on computational patterns and hardware constraints.
- 1 September 2026
- Published by Wiley in International Journal of Numerical Modelling: Electronic Networks, Devices and Fields
- Vol. 39 (5)
Application‐Specific Vedic Multiplier Architectures for Area and Power Efficient VLSI Implementation
Multiplication and squaring are fundamental operations in digital signal processing (DSP), image processing, and embedded computing, where hardware efficiency directly affects speed, area, and energy consumption. Conventional multipliers often introduce unnecessary complexity when operands follow specific patterns. This paper presents a systematic study of selected Vedic mathematics sutras, namely Urdhva‐Tiryakbhyam, Ekanyunena Purvena, Ekadhikena Purvena, Sopaantyadvayamantyam, Yaavadunam, and Nikhilam Navatashcaramam Dashatah, focusing on their suitability for practical computational patterns. The study examines multiplication by constants, operands near base values, and range constrained squaring relevant to DSP applications such as signal energy computation, RMS estimation, filtering, image processing, and neural networks. Corresponding general‐purpose, operand‐pattern‐optimized, fixed‐coefficient, and range‐constrained architectures are implemented and evaluated. FPGA implementations require 6‐117 LUTs with 5.144–11.012 ns delay, while ASIC implementations achieve 1408.491–19 043.73 μm 2 area and 0.565877 × 10 5 to 9.65297 × 10 5 W power. The results show that no single sutra is universally optimal; hardware efficiency depends on operand characteristics and application requirements. The proposed framework therefore enables systematic selection of Vedic multiplication architectures based on computational patterns and hardware constraints.
- 22 July 2026
- Published by IOP Publishing in Physica Scripta
- Vol. 101 (29) , 295003
Compact low-power hardware design for real-time ECG signal pre-processing based on the Pan–Tompkins algorithm
In this work, Electrocardiogram (ECG) QRS complex detection is presented using the Pan–Tompkins algorithm with emphasis on hardware efficient real-time signal pre-processing. Two architectures are proposed, an finite state machine (FSM)-based reusable multiply and accumulate (MAC) architecture optimized for low-power and area-efficient ASIC implementation and a parallel MAC architecture targeting high-speed FPGA applications. Both architectures implement the complete Pan–Tompkins processing chain including band-pass filtering, differentiation, squaring, moving window integration (MWI) along with hardware-based thresholding and QRS detection using signed 18-bit fixed-point arithmetic with 9 fractional bits, 360 Hz ECG samples format. Synthesized on 90 nm and 45 nm GPDK libraries, the FSM-based reusable MAC architecture achieves reduction in ASIC area, power and power-delay product (PDP) by over 60%, while FPGA results show a 35% PDP improvement with reduced resource utilization and delay. Although a complete hardware-based QRS detection chain is implemented and evaluated for performance metrics, achieving clinically reliable detection accuracy using fixed-point hardware thresholding remains challenging. Therefore, to achieve clinically reliable detection accuracy, a hybrid evaluation framework is adopted in which hardware-generated MWI output is used for QRS detection in software using thresholding, refractory logic and search-back mechanisms. By using this hybrid approach evaluation on the MIT–BIH arrhythmia database achieves an average QRS detection accuracy of 98.26%.
- 22 July 2026
- Published by IOP Publishing in Physica Scripta
- Vol. 101 (29) , 295003
Compact low-power hardware design for real-time ECG signal pre-processing based on the Pan–Tompkins algorithm
In this work, Electrocardiogram (ECG) QRS complex detection is presented using the Pan–Tompkins algorithm with emphasis on hardware efficient real-time signal pre-processing. Two architectures are proposed, an finite state machine (FSM)-based reusable multiply and accumulate (MAC) architecture optimized for low-power and area-efficient ASIC implementation and a parallel MAC architecture targeting high-speed FPGA applications. Both architectures implement the complete Pan–Tompkins processing chain including band-pass filtering, differentiation, squaring, moving window integration (MWI) along with hardware-based thresholding and QRS detection using signed 18-bit fixed-point arithmetic with 9 fractional bits, 360 Hz ECG samples format. Synthesized on 90 nm and 45 nm GPDK libraries, the FSM-based reusable MAC architecture achieves reduction in ASIC area, power and power-delay product (PDP) by over 60%, while FPGA results show a 35% PDP improvement with reduced resource utilization and delay. Although a complete hardware-based QRS detection chain is implemented and evaluated for performance metrics, achieving clinically reliable detection accuracy using fixed-point hardware thresholding remains challenging. Therefore, to achieve clinically reliable detection accuracy, a hybrid evaluation framework is adopted in which hardware-generated MWI output is used for QRS detection in software using thresholding, refractory logic and search-back mechanisms. By using this hybrid approach evaluation on the MIT–BIH arrhythmia database achieves an average QRS detection accuracy of 98.26%.
- 1 May 2026
- Published by IOP Publishing in Engineering Research Express
- Vol. 8 (10) , 105303
Gate-level error modeling of approximate 4:2 compressors for fault-resilient computing
This paper presents a systematic framework for the design of approximate multipliers (AMs) that explicitly explores the trade-off between computational accuracy and hardware efficiency at the standard cell level. The proposed methodology employs a symbolic regression model, Python symbolic regression, to estimate error characteristics arising from logic approximation, enabling informed selection of cell configurations during multiplier construction. The predictive model achieves an MAE of 2.33%, an root mean squared error of 5.07%, and explains 74% of the observed variance ( R 2 = 0.74). Leveraging this model, a family of ACOMP is developed, maintaining an identical circuit topology across configurations while offering improved error resilience and reduced hardware cost. The proposed designs are validated through FPGA prototyping and ASIC-level synthesis, and evaluated in terms of area, delay, area delay product, and power delay product. In this article, symbolic regression is inherently integrated into the design methodology to uncover an explicit and interpretable relationship between gate-level design parameters and error behavior, thereby establishing a principled foundation for data-driven approximate circuit design. By enabling early-stage error estimation, the proposed prediction equation enhances state-of-the-art methodologies for approximate logic design in error-tolerant systems. To the best of our knowledge, this is the first study to establish a direct, predictive relationship between error behavior and gate count, providing a practical guideline for the efficient design of AMs in error-tolerant computing systems.
- 1 May 2026
- Published by IOP Publishing in Engineering Research Express
- Vol. 8 (10) , 105303
Gate-level error modeling of approximate 4:2 compressors for fault-resilient computing
This paper presents a systematic framework for the design of approximate multipliers (AMs) that explicitly explores the trade-off between computational accuracy and hardware efficiency at the standard cell level. The proposed methodology employs a symbolic regression model, Python symbolic regression, to estimate error characteristics arising from logic approximation, enabling informed selection of cell configurations during multiplier construction. The predictive model achieves an MAE of 2.33%, an root mean squared error of 5.07%, and explains 74% of the observed variance ( R 2 = 0.74). Leveraging this model, a family of ACOMP is developed, maintaining an identical circuit topology across configurations while offering improved error resilience and reduced hardware cost. The proposed designs are validated through FPGA prototyping and ASIC-level synthesis, and evaluated in terms of area, delay, area delay product, and power delay product. In this article, symbolic regression is inherently integrated into the design methodology to uncover an explicit and interpretable relationship between gate-level design parameters and error behavior, thereby establishing a principled foundation for data-driven approximate circuit design. By enabling early-stage error estimation, the proposed prediction equation enhances state-of-the-art methodologies for approximate logic design in error-tolerant systems. To the best of our knowledge, this is the first study to establish a direct, predictive relationship between error behavior and gate count, providing a practical guideline for the efficient design of AMs in error-tolerant computing systems.
- 26 February 2026
- Published by Elsevier in Integration
- Vol. 109
Error prediction in approximate multiplier using regression and feature expansion method
- 26 February 2026
- Published by Elsevier in Integration
- Vol. 109
Error prediction in approximate multiplier using regression and feature expansion method