Method for prediction of a channel coefficient
AssigneeInfineon Technologies AG
InventorsRobert Denk, Berndt Pilgram
Life of this patent
Estimating the term…
Abstract
The present invention provides a method for prediction of a channel coefficient for a propagation path in a mobile radio channel with multipath propagation, in which a canonical random variable analysis is used for prediction of the channel coefficient. The method is preferably used in an estimator for channel coefficients in a base station or in a mobile station in a mobile radio system.
Claims8
A method for prediction of a channel coefficient in a wireless communication device adapted to communicate over a propagation path in a mobile radio channel with multipath propagation, comprising:
- using a canonical random variable analysis on a signal having a known sequence to predict the channel coefficient in the wireless communication device, wherein using the canonical random variable analysis comprises:
- producing a past matrix P and a future matrix F from a known time series xk of estimated values of the channel coefficient, wherein xk is associated with the signal having the known sequence;
- solving the equation system:
- F=B·M+E
- M=A·P
- in order to obtain estimates  and {circumflex over (B)}, for matrices A and B, wherein F is the future matrix, P is the past matrix, M is a memory matrix representing an intermediate set of variables and a finite number of past states of the respective channel coefficient, E is an error matrix that contains prediction error variables and represents an error in a linear prediction of the future matrix;
- producing a past vector p from a new time series of known estimated values of the channel coefficient;
- and
- calculating a prediction vector {circumflex over (f)} for estimated values of the channel coefficient from the equations:
- {circumflex over (f)}={circumflex over (B)}·m
- m=·p,
- wherein m is an optimum dimension portion of the memory matrix identified by ascertaining canonical correlations.
A method for prediction of a channel coefficient in a wireless communication device adapted to communicate over a propagation path in a mobile radio channel with multipath propagation, comprising:
- using a canonical random variable analysis on a signal having a known sequence to predict the channel coefficient for the wireless communication device, wherein using the canonical random variable analysis to predict the channel coefficient comprises:
- evaluating the signal having the known sequence to identify a sequence of values xk associated therewith, wherein k comprises a discrete integer time variable;
- producing a past matrix P and a future matrix F from the identified sequence of values xk, and
- using the past matrix P and the future matrix F to generate a prediction vector {circumflex over (f)}, wherein a portion of the prediction vector {circumflex over (f)} represents the predicted channel coefficient.
A method of predicting a channel coefficient for estimating a channel impulse response in a wireless communication device, comprising:
- receiving a signal having a known sequence at the wireless communication device;
- and
- evaluating the signal using canonical random variable analysis to determine the channel coefficient, wherein using the canonical random variable analysis to evaluate the signal and determine the channel coefficient comprises:
- evaluating the signal having the known sequence to identify a sequence of values xk associated therewith, wherein k comprises a discrete integer time variable;
- producing a past matrix P and a future matrix F from the identified sequence of values xk;
- and
- using the past matrix P and the future matrix F to generate a prediction vector {circumflex over (f)}, wherein a portion of the prediction vector {circumflex over (f)} represents the predicted channel coefficient.
Dependencies are inferred from the claim text.