KR20030054995A - Fading channel estimation apparatus - Google Patents

Abstract

PURPOSE: A device for estimating a fading channel is provided to estimate coefficients of a time varying channel by using a VFF(Variable Forgetting Factor) Kalman filter, thereby implementing fast and stable estimation performance and automatically calculating an optimal forgetting factor. CONSTITUTION: A one-step predictor(1) estimates channel coefficients based on inputted signals. A VFF solver(2) obtains an optimal VFF at current point by using the inputted signals and error information. A Kalman gain operator(3) operates a gain of a Kalman filter based on signals fed back through a delayer(5). A Ricatti equation solver(4) updates an error distribution matrix, and feeds back the updated error distribution matrix through the delayer(5).

Description

Fading channel estimator {FADING CHANNEL ESTIMATION APPARATUS}

The present invention relates to a technique of W-CDMA uplink channel estimation, which is defined as an issue of implementation in 3GPP. In particular, the variable forgetting factor (VFF) is used to estimate a rapidly changing fading channel. The present invention relates to a fading channel estimator incorporating "

When the transceiver is moving, the time-varying channel, the signal transmission path model at both ends, is the main cause of data distortion in the communication system. Recently, research results of channel estimator using LMS algorithm have been published. However, if the channel to be estimated suddenly changes due to the slow convergence speed of the LMS algorithm, the performance of the estimator deteriorates and the performance of the receiver decreases.

In order to solve this problem, there is a demand for a channel estimator with a fast convergence speed such as a Kalman filter.

However, since the conventional Kalman filter has a fixed oblivion factor, when it is applied to a channel estimator, there is a problem that the optimal oblivion factor must be calculated in advance according to the channel environment.

Accordingly, an object of the present invention is to provide a fading channel estimating apparatus for estimating a coefficient of a time-varying channel using a variable forgetting factor which is a forgetting factor calculated according to a channel environment.

1 is a block diagram of a fading channel estimating apparatus according to the present invention.

2 is a Kalman filtering calculation table according to the present invention.

*** Description of the symbols for the main parts of the drawings ***

1: one-step predictor 2: VFF solver

3: Kalman Gain Operator 4: Rikaty Equation Solver

5: delay

1 is a block diagram showing an embodiment of a fading channel estimating apparatus according to the present invention, and as shown therein, a one step estimator 1 for estimating channel coefficients based on an input signal; A VFF solver (2) for finding an optimal variable forgetting factor at the present time by using an input signal and error information; A Kalman gain operator 3 for calculating the gain of the Kalman filter based on the signal fed back through the retarder 5; A Ricatti equation solver 4 for updating the error variance matrix; It is composed of a retarder (5) for delaying the output signal of the Ricardi equation solver (4) for a predetermined time, and will be described in detail with reference to Figure 2 attached to the operation of the present invention configured as described above.

The input signal string S (k) is buffered by the order of the desired filter and supplied to the one-step predictor 1 and the VFF solver 2. In general, the input signal sequence S (k) is the chip level pilot bits of the Dedicated Physical Control Channel (DPCCH) distorted by the channel.

In the case of DPCCH, the spreading factor is fixed at 256, and the pilot bit is 3 Since it can have 8 symbol values, it can be up to 768 to 2048 bits per slot if taken into account before despreading the chip level bits available for channel estimation. Where state vector Is the channel coefficient when modeling the channel to be estimated by FIR filter.

The initial value input to the one-step predictor 1 is generally zero vector ( ), And the channel coefficient is updated while receiving the pilot bits by the order of the filter for each slot. This is the same as the structure of a general adaptive filter. In addition, in the case of WCDMA, since the receiver already knows a pattern of pilot bits corresponding to the reference signal, a clear channel estimation can be performed using a coherent receiver structure.

The proposed structure of the present invention newly calculates the forgetting factor lambda value of the Kalman filter inputted with the VFF solver 2 set as a constant value by using the input signal information and the error information at every update. In this way, the optimal variable forgetting factor lambda (k) at the present time is obtained.

To this end, it is necessary to know the variance value (R (k)) of the noise signal, which is a statistical characteristic of the input signal, which is also required in the existing Kalman filter, so that any receiver capable of implementing the existing Kalman filter may be implemented without additional input signal information. This is possible.

However, since it is difficult to know this in an actual system, the variance of the error signal e (k) is calculated at every update, and the value tuned to an appropriate constant is treated as the variance of the noise signal.

After all, once the input signal is provided, it is stored in the observation vector S (k), where the information of the channel vector a (k | k) initialized to zero vector is used, and the information of Kalman gain calculated by the Riccati equation is used. It is to estimate the coefficient of the channel to be used in the next step. The estimated coefficient vector is used to recover the distorted input signal in the channel along with the time delay information (given in the MultipathSearcher) of the channel in the rake receiver.

In another embodiment of the present invention, a block for calculating the variable forgetting factor may be added in addition to the Rikaty equation solver 4.

As described in detail above, the present invention estimates the coefficient of the time-varying channel using the variable forgetful factor factor Kalman filter, thereby achieving stable and fast estimation performance even when the channel environment changes rapidly. Compared to the Kalman filter, the optimal forgetting factor can be calculated automatically, and the performance degradation due to channel estimation error of the receiver can be reduced.

Claims (2)

A one-step predictor 1 for estimating channel coefficients based on the input signal; A VFF solver (2) for finding an optimal variable forgetting factor at the present time by using an input signal and error information; A Kalman gain operator 3 for calculating the gain of the Kalman filter based on the signal fed back through the retarder 5; Fading channel estimator characterized in that it comprises a Ricardi equation solver (4) for updating the error variance matrix and feeding it back through the delay (5). The VFF solver (2) calculates a new value of the forgetting factor lambda of the Kalman filter, which is previously set as a constant value and is newly calculated by using the input signal information and the error information for each update. A fading channel estimator, characterized in that configured to obtain a variable forgetting factor lambda (k).