Signal-to-noise ratio (SNR) estimator in wireless fading channels

[0001] The present invention relates generally to wide-band code division multiple access (WCDMA) systems, and more particularly to turbo decoding likelihood ratio estimation in WCDMA systems.

[0002] The generation and transmission of signals invariably involves the introduction of noise into signals, which degrades the quality of the signals and may prevent accurate decoding of the signals. A determination (or an accurate estimate) of the noise contained in the system is useful for enhancing the signal-to-noise ratio of a received signal. Wide-band code division multiple access (WCDMA) systems, such as base stations and mobile terminals, employ a turbo decoder, which uses signal noise power values to enhance accurate decoding of the degraded signals that the WCDMA systems receive. Furthermore, WCDMA systems control power transmission by measuring the received Signal-to-Noise power Ratio (SNR). Accordingly, an accurate estimate of SNR improves the power control and subsequently increases the system performance and capacity.

[0003] An appreciation of the present invention and its improvements can be obtained by reference to the accompanying drawings, which are briefly summarized below, to the following detailed description of illustrated embodiments of the invention, and to the appended claims.

[0007] In the following detailed description of exemplary embodiments of the invention, reference is made to the accompanied drawings, which form a part hereof, and which is shown by way of illustration, specific exemplary embodiments of which the invention may be practiced. These embodiments are described in sufficient detail to enable those skilled in the art to practice the invention, and it is to be understood that other embodiments may be utilized, and other changes may be made, without departing from the spirit or scope of the present invention. The following detailed description is, therefore, not to be taken in a limiting sense, and the scope of the present invention is defined only by the appended claims.

[0008] Throughout the specification and claims, the following terms take the meanings explicitly associated herein, unless the context clearly dictates otherwise. The meaning of “a,” “an,” and “the” includes plural reference, the meaning of “in” includes “in” and “on.” The term “connected” means a direct electrical connection between the items connected, without any intermediate devices. The term “coupled” means either a direct electrical connection between the items connected, or an indirect connection through one or more passive or active intermediary devices. The term “circuit” means either a single component or a multiplicity of components, either active and/or passive, that are coupled together to provide a desired function. The term “signal” means at least one current, voltage, or data signal. Referring to the drawings, like numbers indicate like parts throughout the views.

[0009] The present invention is directed towards a signal-to-noise estimator for estimating the Signal-to-Noise power Ratio (SNR) of mobile wireless systems in wireless fading channels. The estimator is applied to likelihood ratio estimation of turbo decoders in Third Generation Wideband Code Division Multiple Access (WCDMA) receivers in Base Stations and Mobile Terminals. In addition, the SNR estimate is used for transmission power control in WCDMA systems. Time-multiplexed samples of pilot symbols from a wireless receiver are supplied to the estimator to obtain individual estimates of signal power and/or noise power and/or signal-to-noise power ratio. The estimator uses a de-correlating filter in which the coefficients are known functions (such as eigen-vectors) of the auto-correlation matrix for a wireless multi-path fading channel. The decorrelating filter is used by the estimator to map an observation vector (from received pilot symbols) into a set of statistically independent samples (“sufficient statistics”). The energy of each vector component is computed individually and combined (weighted sum) to produce estimates of the signal and interference power estimates. The weighting factors depend on the type of estimators, such as maximum likelihood and sub-space methods. In various embodiments, the estimator can produce joint estimations of signal and noise power in wireless channels by using properties of signal and noise sub-spaces to decouple the observation space into orthogonal domains.

[0010] FIG. 1 is a diagram of a generalized WCDMA system comprising a Base Station and a Mobile Terminal, which depicts the usage of the Signal-to-Noise Ratio Estimator in the system. System 100 comprises at least one base station (BS) 110 and at least one mobile transmitter (MT). Base station 110 comprises the RF transceivers, antennas, and other electrical equipment that are located in each cell site.

[0011] Mobile station 120 is typically a portable wireless communication device, although other configurations are possible.

[0012] BS 110 and MT 120 may transfer voice and data signals between each other and other networks such as the public switched telephone network (PSTN) (not shown), the internet, and the like. BS 110 and MT 120 comprise functional blocks that perform similar functions (although the physical implementations may vary considerably). BS 110 and MT 120 comprise Turbo/Channel Decoder 130, SNR Estimator 140, Transmission Power Control 150, Transmitter 160, and Receiver 170. As described in more detail below, SNR Estimator 140 operates in response to a signal that is received by Receiver 170 and is used to adjust the power of Transmitter 160 and to improve the performance of the channel decoder 130.

[0013] FIG. 2 is a block diagram for a signal-to-noise estimator for estimating the signal-to-noise ratio in wireless fading channels in accordance with the present invention. Estimator 200 comprises decorrelating filter 210, a plurality of sufficient statistics generators 220, and weighted sum generator 203.

[0014] In general, decorrelating filter 210 receives an observation vector from N (total) received pilot symbols. (The operation of estimator 200 is generally described with respect to FIG. 2, while a more detailed description follows with respect to FIG. 3.) Decorrelating filter 210 projects the received observation vector into signal and noise subspaces. The coefficients of this filter (i.e., the projection vectors) are the N (total) eigen-vectors of the fading channel autocorrelation matrix. The plurality of sufficient statistics generators (120) produces a set of N (total) sufficient statistics, which are produced according to the square-magnitude of the outputs of decorrelating filter 210. Weighted sum generator 230 receives the generated sufficient statistics and generates signal and/or noise (“signal/noise”) power estimates in response. Specific sets (in accordance with various estimation algorithms) of weighting coefficients can be used to weight each of the generated N sufficient statistics for producing the signal/noise power estimate.

[0015] FIG. 3 is a flow diagram for estimating the signal-to-noise ratio in wireless fading channels in accordance with the present invention. In block 310, a matched filter receiver or equalizer (such as from a “finger” of a rake receiver) produces outputs for each periodic timeslot (“slot”). Each slot is typically 10–15 milliseconds, although other durations are possible. The outputs are de-multiplexed into data symbols, control symbols, and pilot symbols.

[0016] In block 320, a number of pilot symbols (in which P is the number of pilot symbols) are de-rotated by the known phase of the transmitted symbols. The de-rotated P pilot symbols obtained from a number of successive slots (in which L is the number of successive slots) are assembled in an “observation vector.” The observation vector is an array of length N, wherein N=LXP with L being the number of time-slots used for estimation and. P being the number of pilots per time-slot. The observation vector may be expressed as:
X=[x[1]], [x[1]], . . . [x[N]]

[0017] In block 330, a “projection vector” is obtained for projection of the observation vector onto orthogonal signal subspaces. The projection vector is an array, which is also (typically) of length N. The N projection vectors may be expressed as:
u_j=1, . . . , N
where the N projection vectors are projected onto N orthogonal signal subspaces.

[0018] In various embodiments of the invention, the projection vectors (which are used as coefficients of the de-correlating filter) are the N eigen-vectors associated with the auto-correlation matrix of the fading channel. The auto-correlation matrix can be assumed to be known beforehand for any given Doppler frequency. The Doppler frequency of a mobile transmitted signal is directly proportional to the speed of the mobile and typically introduces phase distortion into the transmitted signal.

[0019] Calculating the N projection vectors in real time is computationally intensive. The projection vectors can be obtained by various methods by which the computational load can be reduced and power can be conserved. For example, the projection vectors can be pre-calculated for different Doppler frequencies, or for example, obtained by nonlinear interpolation from an estimate of the Doppler frequency.

[0020] In the example where the projection vectors are pre-calculated, a lookup table of pre-calculated sets of coefficients (i.e., projection vectors) for different Doppler frequencies can be stored in the memory of the mobile. A set of coefficients can be selected from the table based on a real-time coarse estimate of the Doppler frequency. The coarse estimate can, for example, be divided into low Doppler (around 5 Hz), medium Doppler (around 50 Hz), or High Doppler (around 500 Hz). For cases where a Doppler estimate may be unavailable or unreliable, a fixed set of coefficients can be used regardless of the speed of the mobile. The coefficients are selected such that a relatively good performance is achieved over the entire range of mobile speed.

[0021] In the example where the projection vectors are obtained by nonlinear interpolation, the N projection vectors can be calculated by interpolation for one or more pre-calculated set(s) of coefficients (projection vectors) for particular Doppler frequencies. The particular Doppler frequencies are chosen such that adequate coefficients can be derived using the particular Doppler frequencies.

[0022] The interpolation equation can be expressed as: $u_i\left(f_1\right)=\sum _{j=1}^N{u}_j\left(f_0\right)e^{{\xi }_{\mathrm{ij}}}{f}_1^{{\theta }_{\mathrm{ij}}},$
where the projection vectors at an arbitrary frequency f₁ are obtained from u_i(f₀), in which i=1, . . . N, and in which exponents ξ_i and θ_ij are pre-calculated and stored off-line. The pre-calculated vectors are calculated for frequency f₀, which is typically a low frequency such as 0 Hz. The interpolation coefficients, θ_ij and ξ_ij are obtained by an interpolation algorithm such as a least-squares estimation technique.

[0023] In block 340, a de-correlating filter projects the obtained observation vector onto the j-th signal subspace according to the following equation for j=1, . . . , N: ${\tilde{x}}_j=\sum _{k=1}^N\mathrm{Re}\left(x_j\left[k\right]\right)\mathrm{Re}\left(u_j\left[k\right]\right)+\mathrm{Im}\left(x_j\left[k\right]\right)\mathrm{Im}\left(u_j\left[k\right]\right)$

[0024] In block 350, sufficient statistics T_j, for j=1, . . . , N are calculated. The sufficient statistics are calculated by computing energy of the observation vector in each signal subspace as:
T_j=Re({tilde over (x)}_j)²+Im({tilde over (x)}_j)²

[0025] In block 360, estimates of signal (ρ) and interference (σ²) power are computed. The estimates can be computed by forming a weighted sum of the sufficient statistics T_j as in the following equations $\hat{\rho }=\sum _{j=1}^N{\zeta }_j{T}_j$
and ${\hat{\sigma }}^2=\sum _{j=1}^N{\beta }_j{T}_j$
where ζ and β are weight coefficients given by, for example, maximum-likelihood and subspace estimators. In the following equations for the maximum-likelihood and subspace estimators, N_s is the size of signal subspace and γ_i's are the eigen-values (singular values) of the fading channel auto-correlation matrix (which may be calculated off-line or in real-time).

[0026] The weight coefficients obtained from a maximum-likelihood estimator can be produced according to the following equations. The signal coefficients of the maximum-likelihood estimator can be produced according to ${Ϛ}_j=\frac{{\gamma }_j-1}{N*\left(\mu -1\right)}$
and the noise coefficients of the maximum-likelihood estimator can be produced according to ${\beta }_j=\frac{\mu -{\gamma }_j}{N\left(\mu -1\right)}$
where $\mu =\frac{1}{N}\sum _{i=1}^N{\gamma }_j^2$

[0027] The weight coefficients obtained from a sub-space estimator can be produced according to the following equations. The signal coefficients of the maximum-likelihood estimator can be produced according to ${Ϛ}_j=\{\begin{array}{cc}\frac{1}{N}& 1\le j\le N_s\\ \frac{-N_s}{\left(N-N_s\right)*N}& j>N_s\end{array}$

[0028] The noise coefficients of the sub-space estimator can be produced according to ${\beta }_j=\{\begin{array}{cc}0& 1\le j\le N_s\\ \frac{-1}{\left(N-N_s\right)}& j>N_s\end{array}$

[0029] In block 370, the signal-to-noise power ratio estimate is then calculated by dividing the signal power estimate derived in block 360 by the noise power estimate computed in block 360.

[0030] Various embodiments of the invention are possible without departing from the spirit and scope of the invention. For example, weighting coefficients that are used to produce the signal/noise estimate may be produced by methods other than maximum-likelihood or subspace estimations. The above specification, examples and data provide a complete description of the manufacture and use of the composition of the invention. Since many embodiments of the invention can be made without departing from the spirit and scope of the invention, the invention resides in the claims hereinafter appended.