Method and apparatus providing unbiased signal-to-noise ratio estimation and its application to discontinuous transmission detection

[0001] This invention relates generally to Code Division, Multiple Access (CDMA) mobile communication systems and, more specifically, relates to systems where a mobile station, such as a cellular telephone, is required to measure the signal-to-noise ratio (SNR or (Eb/Nt)) of a forward link channel (a channel transmitted from a base site or base station to the mobile station), and also to those systems that use discontinuous transmission (DTX).

[0002] A current CDMA standard, IS-2000-2, Physical Layer Standard for cdma2000 Spread Spectrum Systems, release 0, July 1999, allows the base station to disable a forward link channel transmission on a frame-by-frame basis. This is referred to herein as discontinuous transmission (DTX). However, information as to whether a particular frame was transmitted or was not transmitted is not known to the Mobile Station. In IS-2000-2 the following channels can be discontinuous: the Forward Dedicated Control Channel (F-DCCH), the Forward Supplemental Channel 1 (F-SCH1), and the Forward Supplemental Channel 2 (F-SCH2).

[0003] Several algorithms executed by the mobile station require information as to whether a frame was actually transmitted. In the context of this patent application this will be referred to as a DTX detection problem. Three examples that illustrate the DTX detection problem follow.

[0004] In a first example, when the forward link power control is enabled on the F-DCCH or the F-SCHn (n1,2), the outer power control loop updates the current setpoint (operating point) based on the frame error information (CRC). However, if the frame was not transmitted the mobile station should not update the current setpoint. In order to accomplish this the mobile station needs to be able to detect whether actual frame transmission took place.

[0005] In a second example, the IS2000 Release A standard (IS-2000-2, Physical Layer Standard for cdma2000 Spread Spectrum Systems, release A, Jun. 9, 2000) specifies a forward link power control mode on F-DCCH (QIB (Quality Indicator Bit) power control). This procedure requires the detection of the F-DCCH frame transmission (DTX detection).

[0006] The IS2000 Release A standard also allows the power control subchannel on the forward link to be transmitted on the F-FCH or the F-DCCH, based on a parameter FPC_PRI_CHAN. When the power control subchannel is transmitted on the F-DCCH, the mobile station supervision procedures require DTX detection.

[0007] Prior to this invention, a satisfactory technique for solving the DTX detection problem was not available.

[0008] The foregoing and other problems are overcome, and other advantages are realized, in accordance with the presently preferred embodiments of these teachings.

[0009] This invention provides both a method and an apparatus to determine the DTX status of a frame (i.e., whether the frame is transmitted/not transmitted) based on the estimation of (Eb/Nt). An enhanced (Eb/Nt) estimator is obtained by noting that the signal part (Eb) and the noise part (Nt) of the (Eb/Nt) estimate contain certain equivalent terms. By subtracting these equivalent terms from the signal estimate, an improved (Eb/Nt) estimator is obtained, also referred to herein as an unbiased (Eb/Nt) estimator. The unbiased (Eb/Nt) estimator is more accurate than the original (Eb/Nt) estimate due to reduced variance. As but one important application of the unbiased (Eb/Nt) estimator, DTX detection of frame transmission can be accomplished.

[0010] A method is disclosed to estimate a signal-to-noise ratio (SNR) of a signal. The method includes sampling the signal, correlating and channel estimating the sampled signal, symbol combining the correlated and channel estimated signal, such as by using a maximal-ratio combining technique, estimating pilot channel and noise power to obtain a combined pilot power and noise variance estimate and inputting the symbol combined signal and the pilot power and noise variance estimate to an (Eb/Nt) estimator. The method further includes, in the (Eb/Nt) estimator, subtracting a noise term, multiplied by a constant, from the combined signal power and computing a ratio of the unbiased signal power to the noise power to obtain an unbiased (Eb/Nt) estimate.

[0011] Having thus obtained the unbiased (Eb/Nt) estimate, and in accordance with a non-limiting further embodiment of this invention, the method integrates the unbiased (Eb/Nt) estimate for a frame duration and compares the integrated unbiased (Eb/Nt) value to a predetermined threshold value to make a decision as to whether the frame was a transmitted frame or an untransmitted (DTX) frame.

[0012] Apparatus for accomplishing the method is also disclosed, as is a CDMA mobile station that operates in accordance with the method.

[0013] The foregoing and other aspects of these teachings are made more evident in the following Detailed Description of the Preferred Embodiments, when read in conjunction with the attached Drawing Figures, wherein:

[0014] FIG. 1 is a simplified block diagram of a CDMA system having a base station and a mobile station that are suitable for practicing this invention;

[0015] FIG. 2 is a block diagram that shows in greater detail the baseband section and the Eb/Nt estimator of FIG. 1, and also a DTX detection algorithm;

[0016] FIG. 3 is a block diagram of the Eb/Nt estimator of FIGS. 1 and 2;

[0017] FIG. 4 is a block diagram of the DTX detection algorithm of FIG. 2; and

[0018] FIG. 5 is a logic flow diagram in accordance with a method of this invention.

[0019] It is noted at the outset that while this invention is described in the context of a CDMA system, and more specifically as a cdma2000 Spread Spectrum system, the method and apparatus of this invention are not restricted to only these types of systems. Thus, while the invention will be described in the context of a CDMA system as a presently preferred embodiment, the teachings of this invention provide a general approach to realizing an improved signal-to-noise ratio estimation through the use of an unbiased (Eb/Nt) estimator, and further provide a general approach to using the signal-to-noise ratio estimate for, as one non-limiting example, DTX detection.

[0020] FIG. 1 is a simplified block diagram of a CDMA system 1 that includes a base station (BS) 2 (typically one of many) that is capable of transmitting a forward link channel, such as a F-DCCH, a F-SCH1 and a F-SCH2, to a mobile station (MS) 3. The MS 3 includes an antenna 3A coupled to a transceiver 3B. An output of the receiver portion of the transceiver 3B is connected to a baseband (BB) section 3C that includes the improved, unbiased (Eb/Nt) estimator 4 in accordance with this invention. A controller, such as a digital signal processor (DSP) 3D, has an input connected to an output of the unbiased (Eb/Nt) estimator 4 of the baseband section 3C, and is enabled to make, as a non-limiting example, a DTX determination there from. The DTX determination can then be employed by the controller when executing an algorithm, such as one of the power control-related algorithms discussed above.

[0021] A detailed analysis of the calculations performed by the unbiased (Eb/Nt) estimator is provided. Reference is also made to FIG. 2, which shows in further detail a portion of the baseband section 3C, the Eb/Nt estimator 4, shown in greater detail in FIG. 3, and a DTX detection algorithm 5, shown in greater detail in FIG. 4. The DTX detection algorithm may be assumed, for the purposes of this invention, to form a part of the controller 3D.

[0022] Briefly, FIG. 2 shows a portion of a CDMA receiver. An analog received signal (RX) is sampled (10) and processed in multiple correlator blocks (fingers 12). The correlator result is combined in a symbol combiner (16) using, preferably, a maximal-ratio combining technique. The output of each finger is also processed in a pilot and noise power estimation block (18). The combined pilot power and noise variance estimate, together with the combiner (16) output, are the inputs to the (Eb/Nt) estimation block 4 that operates as described in detail below.

[0023] FIG. 3 illustrates the calculation of the unbiased estimate of (Eb/Nt) by the (Eb/Nt) estimation block 4 using Equation (13), as described below. The unbiased (Eb/Nt) estimator 4 operates by subtracting the noise term, multiplied by a constant, from the combined signal power, and computes the ratio of the unbiased signal power to the noise power.

[0024] FIG. 4 illustrates the DTX detection algorithm 5 based on the unbiased (Eb/Nt) estimation. Prior to the hypothesis testing the unbiased (Eb/Nt) estimate is integrated over M samples in block 5A so that the total integration length equals the frame duration. The DTX decision is based on a comparison of the integrated unbiased (Eb/Nt) value to a predetermined threshold.

[0025] Discussing FIGS. 2, 3 and 4 now in further detail, consider a received CDMA signal transmitted over an AWGN channel that is A/D converted by A/D converter 10 and sampled at one sample/chip at the input of a correlator (finger) 12, actually a multi-finger correlator as in RAKE receiver. A pilot symbol at the output of the finger 12 is given by:

[0026] where k is the combined A/D and AGC gain, E_cp is pilot E_c/I_or, I_oc is the AWGN channel noise and N_p is the pilot symbol correlation length. The combined A/D and AGC gain k is $\begin{array}{cc}{k}^2=\frac{}{I_{\mathrm{or}}+I_{\mathrm{oc}}},& \left(2\right)\end{array}$

[0027] where is a constant.

[0028] Assume the pilot symbol is further filtered with a filter of unity gain and a noise reduction factor M_p prior to phase correction. This yields at the outputs of channel estimators 14:

[0029] Assume further the presence of a quadrature phase shift key (QPSK) modulated traffic channel. The correlator 12 output of the traffic signal is:

[0030] where N_t is the traffic symbol correlation length. The data bit is ignored from this point forward, since for this analysis one is interested only in the amplitude/power of the signal, and the sign of the data bit is of no interest.

[0031] At the symbol combiner 16 the complex traffic signal is multiplied by the complex conjugate of the pilot symbol. The result is a complex traffic symbol, which is demultiplexed to obtain a scalar stream of traffic symbols. The first traffic symbol (real part of the demodulated complex QPSK symbol) is $\begin{array}{cc}\begin{array}{c}\mathrm{Re}\left\{P_f^{*}T\right\}=\mathrm{Re}\{\left({\mathrm{kN}}_p\sqrt{2E_{\mathrm{cp}}}+n_{p_I}-j{n}_{p_Q}\right)\\ \left({\mathrm{kN}}_t\sqrt{E_{\mathrm{ct}}}+n_{t_I}+j\left({\mathrm{kN}}_t\sqrt{E_{\mathrm{ct}}}+n_{t_Q}\right)\right)\}\\ =k^2{N}_p{N}_t\sqrt{2E_{\mathrm{cp}}{E}_{\mathrm{ct}}}+{\mathrm{kN}}_p\sqrt{2E_{\mathrm{cp}}}{n}_{t_I}+\\ {\mathrm{kN}}_t\sqrt{E_{\mathrm{ct}}}\left(n_{p_I}+n_{p_Q}\right)+n_{p_I}{n}_{t_I}+n_{p_Q}{n}_{t_Q},\end{array}& \left(5\right)\end{array}$

[0032] and similarly for the following symbol (imaginary part of the complex traffic symbol).

[0033] The energy of the traffic symbol is obtained by taking the expected value of the square of Equation (5). By noting that the cross terms involving noise vanish due to an expectation operation, Equation (5) yields

[0034] The energy of the complex traffic symbol is given by:

[0035] The energy of the pilot symbol is obtained similarly in the pilot and noise power estimation block 18, from Equation (3), as:

[0036] There are several possible techniques to estimate the noise variance. A straightforward method calculates the square of the difference between the current and the previous pilot symbol. This yields a result that equals twice the variance of the pilot symbol:

[0037] The (Eb/Nt) estimation block 4 operates using Equations (7), (8) and (9) as follows: $\begin{array}{cc}\frac{E_b}{N_t}=\frac{{\left(pt\right)}^2}{p^2{}^2}=\frac{4k^4\left(N_p^2{N}_t^2{E}_{\mathrm{cp}}{E}_{\mathrm{ct}}+\left(N_p^2{N}_t{E}_{\mathrm{cp}}+\left(N_p/M_p\right)N_t^2{E}_{\mathrm{ct}}\right)I_{\mathrm{oc}}+\left(N_p/M_p\right)N_t{I}_{\mathrm{oc}}^2\right)}{8k^4\left(N_p^3{E}_{\mathrm{cp}}{I}_{\mathrm{oc}}+\left(N_p^2/M_p\right)I_{\mathrm{oc}}^2\right)}.& \left(10\right)\end{array}$

[0038] It can be seen that the first term in the numerator of Equation (10) is the actual signal, and that the three remaining terms are due to the noise (I_oc) which bias the actual signal. In most practical systems, typically (N_pN_t) and (M_p>>1). This implies that the second term in the numerator of Equation (10) is the dominant bias term.

[0039] Multiplying the denominator of Equation (10) by (N_t/(2N_p)) yields $\begin{array}{cc}\frac{N_t}{2N_p}\left(p^2{}^2\right)=4k^4\left(N_p^2{N}_t{E}_{\mathrm{cp}}{I}_{\mathrm{oc}}+\left(N_t{N}_p/M_p\right)I_{\mathrm{oc}}^2\right).& \left(11\right)\end{array}$

[0040] It can be seen by an inspection of Equations (10) and (11) that the quantity in Equation (11) equals the sum of the second and the fourth term of the numerator of Equation (10). Therefore, by subtracting the quantity shown in Equation (11) from the numerator of Equation (10), as depicted in FIG. 3, one reduces the bias of the (Eb/Nt) estimate. This yields $\begin{array}{cc}\begin{array}{c}{\left(\frac{E_b}{N_t}\right)}_{\mathrm{unbias}}=\frac{{\left(pt\right)}^2-\left(\frac{N_t}{2N_p}\right)\left(p^2{}^2\right)}{p^2{}^2}\\ =\frac{4k^4\left(N_p^2{N}_t^2{E}_{\mathrm{cp}}{E}_{\mathrm{ct}}+\left(N_p/M_p\right)N_t^2{E}_{\mathrm{ct}}{I}_{\mathrm{oc}}\right)}{8k^4\left(N_p^3{E}_{\mathrm{cp}}{I}_{\mathrm{oc}}+\left(N_p^2/M_p\right)I_{\mathrm{oc}}^2\right)}.\end{array}& \left(12\right)\end{array}$

[0041] Equation (12), implemented with blocks 4A, 4B, 4C and 4D in FIG. 3, shows the basis of the calculation of the unbiased estimate of (Eb/Nt). It can be seen that the dominant bias term is removed from the signal estimate. In practical systems the remaining bias is small compared to the actual signal, and can essentially be ignored.

[0042] The unbiased estimate of (Eb/Nt) shown in Equation (12) may then be used by the DTX detection algorithm 5 to identify if the frame transmission has taken place. This is accomplished by estimating the (Eb/Nt) over the frame duration B (block 5A of FIG. 4), and comparing the result to a predefined threshold (T) in block 5B. If the (Eb/Nt) estimate exceeds the threshold a decision is made that the frame was transmitted. Otherwise the frame is declared to have not been transmitted. The optimal decision threshold can be derived using simulations, together with laboratory and/or field testing. The value of (T) is typically constant for a fixed parameter configuration. That is, for a certain set of parameters such as data rate (bits/sec) and channel type (e.g., F-FCH, F-SCH1), a predetermined threshold value can be determined and tabulated.

[0043] By using Equation (12) instead of Equation (10) in the threshold comparison of block 5B (hypothesis testing), the miss and false alarm rates of the detection process are considerably reduced, thereby yielding improved performance.

[0044] In the case where a RAKE receiver is used, so that a finger is assigned to each of the multi-paths, two embodiments to obtain the unbiased estimate for a total (Eb/Nt) are disclosed. The first embodiment subtracts the combined bias from the total signal power, which is calculated by using maximal-ratio combining: $\begin{array}{cc}{\left(\frac{E_b}{N_t}\right)}_{\mathrm{unbias}}=\frac{{\left(\underset{i=1}{\overset{L}{}}{\left(pt\right)}_i\right)}^2-\left(\frac{N_t}{2N_p}\right)\left(\underset{i=1}{\overset{L}{}}{p}_i^2{}_i^2\right)}{\left(\underset{i=1}{\overset{L}{}}{p}_i^2{}_i^2\right)},& \left(13\right)\end{array}$

[0045] where L denotes the number of assigned fingers (see FIG. 2). The second embodiment subtracts the bias from each finger separately prior to combining: $\begin{array}{cc}{\left(\frac{E_b}{N_t}\right)}_{\mathrm{unbias}}=\frac{\underset{i=1}{\overset{L}{}}\left({\left(pt\right)}_i^2-\left(\frac{N_t}{2N_p}\right)p_i^2{}_i^2\right)}{\left(\underset{i=1}{\overset{L}{}}{p}_i^2{}_i^2\right)}.& \left(14\right)\end{array}$

[0046] For practical purposes the first embodiment is preferred, as it uses the output of the symbol combiner 16, where it is assumed that data is combined by using a maximal-ratio combining technique.

[0047] Referring to FIG. 5, a method is disclosed to estimate a signal-to-noise ratio (SNR) of a signal. The method includes sampling the signal at block A, correlating and channel estimating the sampled signal at block B, symbol combining the correlated and channel estimated signal, such as by using a maximal-ratio combining technique, at block C, estimating pilot channel and noise power to obtain a combined pilot power and noise variance estimate (block D) and inputting the symbol combined signal and the pilot power and noise variance estimate to an (Eb/Nt) estimator at block E. The method further includes, in the (Eb/Nt) estimator, subtracting a noise term, multiplied by a constant, from the combined signal power (block F) and computing a ratio of the unbiased signal power to the noise power to obtain an unbiased (Eb/Nt) estimate (block G).

[0048] Having thus obtained the unbiased (Eb/Nt) estimate, and in accordance with a non-limiting further embodiment of this invention, the method integrates the unbiased (Eb/Nt) estimate for a frame duration (block H) and compares the integrated unbiased (Eb/Nt) value to a predetermined threshold value to make a decision as to whether the frame was a transmitted frame or an untransmitted (DTX) frame (block I). Mobile station 3 power control processing may then be based on the DTX decision.

[0049] The method can be performed in software, hardware, or in a combination of software and hardware.

[0050] Also disclosed above has been a CDMA mobile station 3 operable with a base station 2 that is capable of transmitting a signal on a frame-by-frame basis in accordance with a DTX technique. The mobile station 3 includes a receiver that is operable during a frame time and, coupled to the receiver, the unbiased (Eb/Nt) estimator 4. A power controller, such as the controller 3D, is coupled to the unbiased (Eb/Nt) estimator 4 for making a power control decision, such as whether or not to update a setpoint, in accordance with at least a DTX status of a frame (i.e., was the frame transmitted or not transmitted by the base station 2). The DTX status is determined based on a comparison of an output of the unbiased (Eb/Nt) estimator to the threshold (T) value.

[0051] As was noted above, the improved technique for estimating the (Eb/Nt) in an unbiased manner, in accordance with this invention, is not limited for use with CDMA systems, and furthermore the unbiased (Eb/Nt) estimation is not limited for use in performing DTX detection, nor is it limited for use during mobile station 3 power control processing.

[0052] The foregoing description has provided by way of exemplary and non-limiting examples a full and informative description of the best method and apparatus presently contemplated by the inventor for carrying out the invention. However, various modifications and adaptations may become apparent to those skilled in the relevant arts in view of the foregoing description, when read in conjunction with the accompanying drawings and the appended claims. However, all such and similar modifications of the teachings of this invention will still fall within the scope of this invention. Further, while the method and apparatus described herein are provided with a certain degree of specificity, the present invention could be implemented with either greater or lesser specificity, depending on the needs of the user. Further, some of the features of the present invention could be used to advantage without the corresponding use of other features. As such, the foregoing description should be considered as merely illustrative of the principles of the present invention, and not in limitation thereof, as this invention is defined by the claims, which follow.