580 California St., Suite 400
San Francisco, CA, 94104
Key research themes
1. How can hardware-efficient designs minimize arithmetic complexity in FIR filter implementations?
This research area focuses on reducing the computational cost and hardware resource requirements of FIR filters, particularly targeting adders and multipliers in digital signal processing implementations. Efficient FIR filter designs are crucial for low-power, high-speed DSP applications, especially in resource-constrained embedded systems.
Key finding: Introduced the frequency-response masking (FRM) technique to design two-channel FIR filter banks that significantly reduce the number of adders and multipliers needed compared to conventional implementations, especially for... Read more
Key finding: Developed integer linear programming (ILP) methods for simultaneous FIR filter design and multiplierless hardware implementation that optimize the number of adders to meet given frequency specifications. Demonstrated that... Read more
Key finding: Proposed a floating point fixed-width multiplier utilizing a column bypassing technique which reduces switching activities in partial product operands. This design lowers power consumption by 10% compared to conventional... Read more
Key finding: Introduced a reduced precision redundancy (RPR) unit integrated with fixed-width multipliers to compensate truncation errors caused by partial product elimination. This design achieves a trade-off by reducing power... Read more
Key finding: Designed and implemented a low-power FIR filter replacing traditional multipliers with an 8-bit multiplier-accumulator (MAC) unit using a Wallace tree multiplier to reduce partial products and latch-based D flip-flops for... Read more
2. What advances in window-based FIR filter design improve spectral characteristics for signal processing applications?
This research concentrates on enhancing the spectral performance of FIR filters through novel window functions or hybrid windows. Improvements target better main lobe width, side lobe attenuation, ripple ratios, and overall frequency response fidelity, which are critical in applications like audio processing, ECG denoising, and communications.
Key finding: Proposed 20 hybrid window functions by combining the two-parameter Cosh window with various one-parameter existing windows. Simulation results in FIR filter design showed nine hybrid windows improving key filter... Read more
Key finding: Developed a new dynamic two-variable window function combining Hamming, Blackman-Harris, Chebwin, and Kaiser windows to provide adjustable spectral characteristics. MATLAB simulations demonstrated improved ripple ratio, main... Read more
Key finding: Evaluated various window functions (Blackman, Hamming, Hanning) for FIR filter design applied in M-ary PSK communication systems. Empirical results showed the Blackman window produced FIR filters with superior bit error rate... Read more
Key finding: Applied multiple FIR filter windows including Kaiser, Chebyshev, and Hamming for ECG signal noise reduction, comparing spectral removal efficiency. The study confirmed that window choice significantly impacts filtering... Read more
Key finding: Systematically analyzed 49 combinations of band stop and high pass FIR filters using seven different window functions to remove baseline wander and power line interference in ECG signals. Identified the best performing window... Read more
3. How do FIR filter designs enable effective sample rate conversion and flexible system identification in DSP?
This theme addresses FIR filter usage in critical signal processing operations such as sample rate conversion (SRC) and system identification. FIR filters' linear phase properties and stability make them favorable for these tasks, and advances include varied architectures (polyphase, Farrow, CIC), optimization techniques, and adaptive algorithms for estimating unknown systems.
Key finding: Provided a unified tutorial and classification of FIR-based SRC techniques, including classical upsampling/filtering/downsampling, polyphase, Farrow, CIC, and Newton filter structures. The survey highlighted implementation... Read more
Key finding: Demonstrated the use of adaptive FIR filters with the Least Mean Square (LMS) algorithm implemented on an ATMEGA16 processor for system identification. By iteratively adjusting filter weights to minimize error between... Read more
Key finding: Presented practical MATLAB experiments demonstrating FIR filtering including convolution, step response analysis, and signal filtering by both sample-by-sample and block processing methods. The exercises illustrated... Read more