Estimation of noisy quantized Gaussian AR time-series with randomly varying observation coefficient

Presents an estimation algorithm for the parameters of Gaussian auto-regressive AR processes from one-bit quantized observation sequences. The input signal to the quantizer is the AR signal corrupted by multiplicative white Gaussian noise. The estimation algorithm is computationally inexpensive as it involves counting the number of occurrences of particular patterns of zeros and ones in the observation sequence.<
>