Linear prediction models signals as all-pole IIR systems predicting samples from prior ones

Evidence

In LP modelling, we can imagine that the signal to be analysed is generated by an IIR system (Figure 3.11), where in the numerator of Equation (3.32) the coefficient b0 = 1 and all the other coefficients bn = 0 for n ≥1. This is called an all-pole type IIR system. LP analysis yields optimal values in the least mean square sense for the denominator polynomial coefficients ap, so that the resulting all-pole IIR filter system is the best one for predicting a new signal sample as a linear combination from the previous samples. — Ville Pulkki, Matti Karjalainen, Communication Acoustics- An…, 430

Categorization

• DEFINE: Explains the core conceptual move of LP modelling: treating a signal as if produced by an all-pole IIR system.
• corroborated: 3 independent passages support this claim — Ville Pulkki, Matti Karjalainen, Communication Acoustics- An…, 430