The Design of a Real-Time Adaptive Filter Development System
Abstract
This paper describes a hardware test platform designed to implement adaptive lattice filters in real time. To achieve real time processing speeds, algorithm complexity was accomplished by custom designing the computation engines with respect to the lattice data flow. Execution speeds of the computation engines were dramatically increased by a memory architecture that supports efficient addressing and by providing a floating point ALU with numerous data paths and efficient implementation of division. Performance is further enhanced by pipelining multiple computation engines. In addition, the architecture is flexible enough to support other filter structures and to allow observation of filter variables as they adapt. With this system, various Adaptive Filter algorithms are being tested in real-time implementation of Adaptive Line Enhancers (ALEs) and Adaptive Noise Cancelers (ANCs) in order to characterize their performance and behavior, especially long term stability and the ability to track non-stationary signals.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 01, 1991
- Accession Number
- ADA237134
Entities
People
- Michael D. Juniper
- Richard C. North
- Terence R. Albert
- Walter H. Ku