ESTIMATING OPTIMAL LINEAR REGRESSION FILTER LENGTH FOR CHANNEL ESTIMATION

摘要

Techniques for obtaining smoothed channel response estimates in a communications receiver are disclosed. An example method begins with obtaining ordered channel measurement samples for a current estimation interval and continues with the estimation of a short-term slope value for each of several sequential observation time intervals. A distortion metric is calculated for each of a plurality of candidate filter lengths, where each candidate filter length is an integer multiple of the uniform length. The distortion metric is based on the short-term slope values for observation time intervals falling within the candidate filter length, and characterizes the extent to which the candidate filter length will introduce tracking bias in subsequent signal processing. A filter length is then selected for use in demodulating a received signal from at least among the candidate filter lengths, based on the distortion metric values.