Delineation of Optimized Single and Multichannel Approximate DA-Based Filter Design Using Influential Single MAC...

Area of Science:

• Digital Signal Processing
• VLSI Design
• Computer Engineering

Background:

• Multichannel Finite Impulse Response (FIR) filters are crucial in various signal processing applications.
• Efficient hardware implementation of these filters is challenging due to resource constraints.
• Existing designs often require significant computational resources and power.

Purpose of the Study:

• To propose a novel multichannel FIR filter design.
• To optimize resource utilization by employing a single multiply-add unit.
• To reduce latency, area, and power consumption in filter implementation.

Main Methods:

• Utilizing a Time Division Multiplex (TDM) approach for resource sharing.
• Employing approximate distributed arithmetic (DA) circuits for multiplier optimization.
• Implementing radix-8 and radix-4 Booth algorithms within the DA framework.
• Using an erratum mending unit for input stream truncation and partial product construction.
• Incorporating an approximate Wallace tree for partial product aggregation.

Main Results:

• Demonstrated significant reduction in latency, area, and power consumption.
• Achieved a high operating frequency of up to 429 MHz on a Xilinx Vertex device.
• Verified filter performance with a 16-tap multichannel realization.
• Confirmed cell-level performance using TSMC 180 nm CMOS technology and Cadence RC compiler.

Conclusions:

• The proposed TDM-based multichannel FIR filter design offers substantial hardware savings.
• The integration of approximate DA techniques effectively minimizes resource usage and enhances performance.
• The design is suitable for high-frequency applications requiring low power and area.