Roundoff Noise in Fixed-Point Recursive Digital Filters.
Dale John Mayer
Abstract
Dale John Mayer
Abstract
Digital filtering techniques are very appealing where complex signal processing is desired. A major problem with digital filters however is the effect of quantization errors due to finite precision arithmetic. This report is concerned with roundoff errors (due to truncation or rounding) occurring after each multiplication in linear shift-invariant fixed-point recursive digital filters. The accumulation of these errors appears as noise (error uncorrelated) or limit cycles(correlated errors) at the filter output thus limiting the dynamic range of the filter operating with a given wordlength. (Author)
OpenAlex reports 1 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
Digital filtering techniques are very appealing where complex signal processing is desired. A major problem with digital filters however is the effect of quantization errors due to finite precision arithmetic. This report is concerned with roundoff errors (due to truncation or rounding) occurring after each multiplication in linear shift-invariant fixed-point recursive digital filters. The accumulation of these errors appears as noise (error uncorrelated) or limit cycles(correlated errors) at the filter output thus limiting the dynamic range of the filter operating with a given wordlength. (Author)
Key concepts: Fixed-point arithmetic, Rounding, Digital filter, Quantization (signal processing), Mathematics, Round-off error, Algorithm, Saturation arithmetic