Approximate Computing on FPGAs for Edge Computing Applications: A Case Study of Adaptive Filters

Omidi, Reza

doi:10.22075/mseee.2025.37914.1215

Approximate Computing on FPGAs for Edge Computing Applications: A Case Study of Adaptive Filters

Document Type : Research Article

Author

Reza Omidi

Department of Electrical Engineering, University of Zanjan, Iran.

10.22075/mseee.2025.37914.1215

Abstract

The exponential growth of data and the paradigm shift towards edge computing have necessitated innovative approaches to energy-efficient computation, especially for resource-constrained IoT devices. Approximate computing, a paradigm that exploits the inherent tolerance of many applications to imprecision, has been extensively explored in the context of ASICs to reduce power consumption and area overhead. However, its potential in FPGA-based devices, which offer flexibility and rapid prototyping capabilities, remains largely untapped. This paper investigates the feasibility and performance implications of approximate computing techniques for FPGAs, with a focus on the implementation of FIR and adaptive filters as illustrative case studies. Specifically, we propose novel approximate multipliers based on the Reverse Carry Propagation (RCP) adders, which are evaluated through their integration into adaptive and finite impulse response filters. Simulation results demonstrate significant improvements in operating frequency, as well as substantial reductions in hardware area for FIR filters. While the area reduction for adaptive filters is less pronounced, the proposed multipliers still exhibit acceptable performance. Our findings highlight the potential of approximate multipliers for low-power computing systems, particularly in edge computing applications.

Keywords

Approximate computing

Edge computing

FPGA

adaptive filter

multiplier

adder

inexact arithmetic

Subjects

Computer

[1] H. J. Damsgaard, A. Ometov, and J. Nurmi, “Approximation opportunities in edge computing hardware: A systematic literature review,” *ACM Comput. Surv.*, vol. 55, no. 12, Art. no. 252, 2023, doi: 10.1145/3572772.

[2] S. S. Sahoo, S. Ullah, S. Bhattacharjee, and A. Kumar, “AxOCS: Scaling FPGA-based approximate operators using configuration supersampling,” *IEEE Trans. Circuits Syst. I, Reg. Papers*, vol. PP, pp. 1–14, 2024, doi: 10.1109/TCSI.2024.3385333.

[3] S. Ullah and A. Kumar, *Approximate Arithmetic Circuit Architectures for FPGA-Based Systems*, 2023.

[4] V. Chowthri, A. Uma, and P. Kalpana, “Design and implementation of low complexity LMS adaptive filter,” *Int. J. Nonlinear Anal. Appl.*, vol. 12, no. Special Issue, pp. 1827–1833, 2021, doi: 10.22075/ijnaa.2021.5893.

[5] V. B. A. Rasel, A. Uma, and P. Kalpana, “Design of window based FIR filter for electrocardiogram noise removal,” *Int. J. Nonlinear Anal. Appl.*, vol. 12, no. Special Issue, pp. 1519–1528, 2021, doi: 10.22075/ijnaa.2021.5804.

[6] D. Esposito, G. D. Meo, D. D. Caro, N. Petra, E. Napoli, and A. G. M. Strollo, “On the use of approximate multipliers in LMS adaptive filters,” in *Proc. 2018 IEEE Int. Symp. Circuits Syst. (ISCAS)*, May 27–30, 2018, pp. 1–5, doi: 10.1109/ISCAS.2018.8351089.

[7] I. Qiqieh, R. Shafik, G. Tarawneh, D. Sokolov, and A. Yakovlev, “Energy-efficient approximate multiplier design using bit significance-driven logic compression,” 2017, pp. 7–12.

[8] M. Chakraborty, A. Dhar, and M. Lee, “A trigonometric formulation of the LMS algorithm for realization on pipelined CORDIC,” *IEEE Trans. Circuits Syst. II, Exp. Briefs*, vol. 52, pp. 530–534, 2005, doi: 10.1109/TCSII.2005.850784.

[9] D. Allred, H. Yoo, V. Krishnan, W. Huang, and D. Anderson, “LMS adaptive filters using distributed arithmetic for high throughput,” *IEEE Trans. Circuits Syst. I, Reg. Papers*, vol. 52, pp. 1327–1337, 2005, doi: 10.1109/TCSI.2005.851731.

[10] M. T. Khan, S. R. Ahamed, and F. Brewer, “Low complexity and critical path based VLSI architecture for LMS adaptive filter using distributed arithmetic,” in *Proc. 2017 30th Int. Conf. VLSI Design and 2017 16th Int. Conf. Embedded Syst. (VLSID)*, Jan. 7–11, 2017, pp. 127–132, doi: 10.1109/VLSID.2017.16.

[11] P. K. Meher and S. Y. Park, “Critical-path analysis and low-complexity implementation of the LMS adaptive algorithm,” *IEEE Trans. Circuits Syst. I, Reg. Papers*, vol. 61, pp. 778–788, 2014, doi: 10.1109/TCSI.2013.2284173.

[12] S. Choudhary, P. Mukherjee, M. Chakraborty, and S. S. Rath, “A SPT treatment to the realization of the Sign-LMS based adaptive filters,” *IEEE Trans. Circuits Syst. I, Reg. Papers*, vol. 59, no. 9, pp. 2025–2033, 2012, doi: 10.1109/TCSI.2012.2185300.

[13] M. Hassan, F. Awwad, M. Atef, and O. Hasan, “Approximate computing-based processing of MEA signals on FPGA,” *Electronics*, vol. 12, no. 4, Art. no. 848, 2023. [Online]. Available: https://www.mdpi.com/2079-9292/12/4/848