WO2007028279A1

WO2007028279A1 - Detection method for ack/nack signals and detector thereof

Info

Publication number: WO2007028279A1
Application number: PCT/CN2005/001434
Authority: WO
Inventors: Oskar Mauritz; Branislav Popovic
Original assignee: Huawei Technologies Co., Ltd.
Priority date: 2005-09-08
Filing date: 2005-09-08
Publication date: 2007-03-15
Also published as: EP1922825A1; US20080155371A1; US7770086B2; CN101258696A; DE602005017520D1; EP1922825B1; CN101258696B; ATE447802T1; EP1922825A4

Abstract

A detection method for ACK/NACK and detector thereof, wherein the detection method comprises, calculating a threshold between ACK and DTX, which includes two values, the first being proportional to a noise standard deviation after despreading the received signal, and the second depending on an estimated received signal for ACK; calculating a threshold between NACK and DTX, which includes two values, the first being proportional to a noise standard deviation after despreading the received signal, and the second depending on an estimated received signal for NACK; selecting the one with maximum absolute value from the two calculated values of the threshold between ACK and DTX ;selecting the one with maximum absolute value from the two calculated values of the threshold between NACK and DTX; comparing the decision variable with the two selected thresholds and making decision. The detector comprises a means for generating thresholds, wherein the means is used for selecting the one with maximum absolute value from the two calculated values of the threshold between ACK and DTX, and the threshold between NACK and DTX, respectively. The invention improves performance of ACK/NACK decision.

Description

DETECTION METHOD FOR ACK/NACK SIGNALS AND

DETECTOR THEREOF

Field of the invention

The present invention relates to the field of digital radio communication, and in particular to a selective threshold ACK/NACK detector and method thereof.

Background of the invention

The evolution of the third generation mobile communication systems includes higher data rates and packet oriented modes. High Speed Downlink Packet Access (HSDPA) is a new feature in WCDMA that improves throughput in the system and increases the maximum data rate for a single user. HSDPA is a packet data transmission system, where the base station schedules and transmits data packets to different Mobile Stations (MSs).

An important component to prevent from losing data packets in HSDPA is the Hybrid Automatic Repeat Request (Hybrid ARQ). The data packets are preceded by indicators that inform the receiving MS about transmission time and other characteristics of the transmission. For each packet that the MS receives, the MS transmits an acknowledgement (ACK) signal if the packet has been received correctly and a negative acknowledgement (NACK) signal if the packet was received but not correctly. It may happen that the MS does not detect the indicator signal from the base station. In that case the MS is not able to receive the data packet, and will apply discontinuous transmission (DTX), i.e. the MS will not transmit any signal at all. In other words, the MS only transmits ACK/NACK signals when it has received a packet, otherwise there is no signal transmitted.

The transmitted signals, ACK and NACK, are antipodal signals. In parallel with the ACK/NACK signal there is a pilot signal that can be used for channel estimation. The network specifies power offsets AP _a and AP_n for ACK/NACK transmissions, respectively.

These power offsets are relative to the power for the pilot signal. The channel estimation and the known power offsets give the estimated received signal powers for ACK and NACK in case there was an ACK or NACK transmission. Thus the received signal powers and amplitudes of ACK and NACK transmissions can be estimated by the receiver.

When the base station tries to receive the ACK/NACK, there are three different possibilities: ACK, NACK, or DTX has been transmitted. Table 1 lists the target performance requirements on the physical layer that have been outlined in the 3GPP for the ACK/NACK detection. The requirements are given in the form of probabilities of erroneous detection that the physical layer should not exceed. The ability of the physical layer to fulfill the requirements depends on the ACK/NACK detector as well as the power offsets ΔP_a and AP_n for ACK and NACK transmission, respectively, which are specified by the network. In table 1, the notation P(DTX|ACK) represents the probability of detecting DTX signal when ACK signal is transmitted. It is similar for P(ACK| NACK) and P(ACK|DTX).

Table 1 Target performance requirements on ACK/NACK reception

Two ACK/NACK detectors have been proposed, the Constant False Alarm Rate (CFAR) detector, and a dynamic threshold detector. Wherein CFAR detector was presented in "Simulation conditions for HS-DPCCH (ACK/NACK) detection performance," R4-030928, 3GPP, Nov. 2003, and "Energy requirements for UL ACK/NACK signaling under different sets of constraints", Rl -02-0420, 3GPP, Feb. 2002. The dynamic threshold detector was presented in "On the decision threshold for detecting ACK/NACK messages," Rl -02-0823, 3GPP, May 2002.

The CFAR detector ensures a constant erroneous detection probability of ACK and NACK when no signal has been transmitted i.e. DTX condition, regardless of the noise power. This is achieved by having an adaptive detection threshold proportional to the noise standard deviation. For a CFAR detector of ACK/NACK, two such adaptive thresholds are needed: T_a, which is negative, between "ACK" and "DTX", in case of negative signs for ACK and T_n, which is positive, between "DTX" and "NACK", in case of positive signs for NACK. It is shown in Figure Ia. A decision variable z can be obtained by means of accumulating outputs of Rake combiner. If a decision variable z is less than the threshold T_a, the decision is ACK ; if a decision variable is greater than threshold T_n, the decision is NACK; if a decision variable is between the two thresholds, T_a and T_n, the decision is DTX.

The threshold T_n is formed by multiplying the noise standard deviation after despreading σ_w with the norm of column vector given by channel estimates, h , and the coefficient O₀ that is set to fulfill the requirement on P(ACK|DTX), while T_n = -T_a . The column vector of channel estimates is the set of weights used in a maximum ratio combiner. Other sets of weights are also possible, e.g. weights corresponding to equal gain combining. T_a is given by

T_a = ~cc_oσ_w h (1)

where P(ACK|DTX) is 0.01 for O₀ =1.65.

The channel is here defined as the product of the amplitude of the transmitted pilot signal, i.e. the square root of the pilot transmission power and the complex- valued radio channel, hi a fading channel, the CFAR detector can result in an unnecessarily high ACK power requirement, in order to ensure that the ACK signal is sufficiently above the average noise power. Such high required ACK power can be reduced by the following dynamic threshold detector proposed by Philips, using information about the instantaneous propagation channel conditions, which are reflected in the estimated signal amplitude.

The dynamic threshold detector has a threshold T_a that is proportional to the product of the noise standard deviation and the estimated signal amplitude after Rake combining

2

JAPJh

T_n = -ασ_uJΔP_Λ h (2)

The constant a is selected such that in average P(DTX|ACK) fulfils the target performance requirement. The value of α depends on the propagation channel, speed, antenna diversity, as well as ΔP_a.

Analogously, the detector has a threshold T_n that is proportional to the product of the noise standard deviation and the estimated signal amplitude after Rake combining:

2

T ^■ n. = ασ,.,_Λ/ΔP 1Ih₁ (3)

For both CFAR detector and dynamic threshold detector, in case of positive signs for ACK and negative signs for NACK, threshold T_a is positive and threshold T_n is negative. If a decision variable z is less than the threshold T_n, the decision is NACK; if a decision variable is greater than threshold T_a, the decision is ACK; if a decision variable is between the two thresholds, T_n and T_a, the decision is DTX. This is shown in Fig. Ib.

There are several drawbacks with this dynamic threshold detector:

1) It is impossible to calculate a for all possible channels, speeds, etc.

2) The required power for NACK can become higher than the required power for ACK. The reason is that: the required P(ACK|NACK) is much lower than the required

P(NACK| ACK) because if NACK is detected as ACK the physical layer will not detect an erroneous packet, and instead, the detection will be made in higher layers and thus cause longer delays for the retransmission; if the threshold between DTX and ACK, T_a, is close to the origin, the NACK power must be high to keep P(ACK|NACK) below the tight requirement.

3) The threshold depends on the product of the noise standard deviation and the estimated signal amplitude. This implies that the output of the detector depends not only

Il2 2 on the ratio _Λ/APj|h|| /σ,,, but on the absolute values of J AP and σ_w.

The ACK/NACK signals are important in an ARQ scheme. To achieve a reliable ACK/NACK detection in HSDPA the transmission power is typically relatively high. If the ACK/NACK signals need to be repeated to achieve sufficient detection performance, the interference increases and the maximum data rate is decreased. An efficient detector is needed to minimize the required transmission power for ACK and NACK signals and to maximize the data rate. Summary of the invention

The invention presents a selective threshold ACK/NACK detector and method thereof to improve performance of ACK/NACK decision.

The scheme is implemented as follows. The detection method for ACK/NACK, comprises, a. Calculating a decision variable by integration of a despread received signal; b. calculating a threshold between ACK and DTX, which includes two values, the first being proportional to a noise standard deviation after despreading the received signal, and the second depending on an estimated received signal amplitude for ACK, calculating a threshold between NACK and DTX, which includes two values, the first being proportional to a noise standard deviation after despreading the received signal, and the second depending on an estimated received signal amplitude for NACK, c. Selecting the one with maximum absolute value from the two calculated values of the threshold between ACK and DTX ; Selecting the one with maximum absolute value from the two calculated values of the threshold between NACK and DTX; d. Comparing the decision variable with the two selected threshold and making decision.

The ACK/NACK detector comprises, a Rake combiner (26) for despreading and performing channel estimation of a received signal to generate a single value, which includes at least a despreader (20) and a channel estimator (25), an integrator (21) for integrating the single value of one slot and accumulating the integrated results of each of N slots to obtain a decision variable, a comparator (22) for comparing the decision variable from the integrator (21) with a threshold between ACK and DTX and a threshold between NACK and DTX given from a means for generating thresholds (30), respectively, a noise estimator (24) for performing estimation of despread signal from the despreader (20) to obtain a noise standard deviation after despreading, wherein the means for generating thresholds (30), is used for calculating a threshold between

ACK and DTX that includes two values, the first being proportional to a noise standard deviation from the noise estimator (24), and the second depending on an estimated received signal amplitude for ACK, calculating a threshold between NACK and DTX that includes two values, the first being proportional to a noise standard from the noise estimator (24), and the second depending on an estimated received signal amplitude for NACK,

Selecting the one with maximum absolute value from the two calculated values of the threshold between ACK and DTX ;

Selecting the one with maximum absolute value from the two calculated values of the threshold between NACK and DTX.

The Rake combiner (26) may be a maximum ratio combiner.

With the selective threshold detector and method thereof, the required transmission power for ACK and NACK signals are decreased while the target performance requirements on the physical layer outlined in the 3GPP are met. Since the transmission power for ACK and NACK signals are decreased, interference is decreased and data rate is improved.

Brief Description of the Drawings

Figure 1 shows decision regions for the CFAR detector of the prior art. Figure 2 shows the detection method for ACK/NACK of the invention. Figure 3 shows a selective threshold detector for ACK/NACK of the invention.

Embodiments of the Invention

The CFAR detector ensures a constant probability of false alarm. However, the probability of missed detection, P(DTX|ACK), varies with the instantaneous

SNR(signal-to-noise ratio). Since knowledge about the instantaneous SNR is not used in the transmitter, the ACK power may be unnecessarily high to achieve the required probability of missed detection.

The dynamic threshold detector utilizes the information about the expected received signal power obtained from the known power offsets and the channel estimation. However, this detector has several problems as stated in foregoing disclosure. To ensure that P(ACK|NACK) is not higher than its strict requirement, the invention introduces a noise-level dependent threshold. In the relatively few cases when the absolute value of the dynamic threshold is lower than the absolute value of the noise-level dependent threshold, the latter threshold is selected. This limitation of the thresholds efficiently reduces P(ACK|NACK) and P(NACK|ACK) with only a small impact on P(DTX|ACK) and P(DTX|NACK).

The invention uses all the available information in the HSDPA air-interface that might help in detection of ACK/NACK and at the same time combine it with the information about the noise level in the receiver. Namely, since the channel coefficients and the ACK/NACK power offsets are known to the receiver, the expected received signal amplitudes of ACK and NACK are known in the receiver.

Since in HSDPA the ACK sign is negative and NACK sign is positive, the invention selects the threshold between ACK and DTX from the minimum of two values, the first being proportional to the noise standard deviation after despreading, and the second being proportional to the estimated received signal amplitude. The maximum of two values, the first being proportional to the noise standard deviation after despreading, and the second being proportional to the estimated received signal amplitude, is selected as the threshold between NACK and DTX. Contrarily, in case of in a system with reversed signs for ACK and NACK, i.e. positive for ACK and negative for NACK, the threshold between ACK and DTX should be selected as the maximum of two thresholds, and the threshold between NACK and DTX should be selected as the minimum of two thresholds. The decision criterion is same as which in Fig. Ib.

The invention will be described in more detail with reference to the accompanying drawing and an embodiment.

In HSDPA the ACK/NACK signal is a 1-bit message repetition coded to 10 bits giving the code words

"ACK": 1111111111 and "NACK": 0000000000.

After the usual mapping 0 -» +1 and 1 -> -1 , it is clear that ACK and NACK are antipodal signals, where ACK has negative and NACK positive sign. Hence, the decision variable z should be obtained by integration of despread ACK/NACK bits. The ACK/NACK transmissions are performed in the first time slot of three in. a subframe, and should be repeated in N consecutive subframes where N is a parameter set by the network.

Referring to the Fig.2, the selective threshold ACK/NACK detection method comprises, calculating the decision variable z by integration of despread ACK/NACK values received; estimating a noise standard deviation after despreading the received signals; generating a weight function by means of channel estimates and the noise standard deviation; wherein the weight function includes a first weight function for ACK and a second NACK weight function for NACK; generating the first threshold between ACK and DTX, T_aj, using the noise standard deviation and norm of column vectors given by channel estimates, and the second threshold between ACK and DTX, T_a>2, using the power offset for the ACK transmission and the first weight function; generating the first threshold between NACK and DTX, T_{n ιh} using the noise standard deviation and norm of column vectors given by channel estimates, and the second threshold between NACK and DTX, T_n^, using the power offset for the NACK transmission and the second weight function; selecting minimum of two values, first threshold T_a,i and the second threshold T_aj, and at the same time selecting maximum of two values, first threshold T_n, i and the second threshold T_n,?, comparing the decision variable with the selected T_a and T_n; if the decision variable is less than the threshold T_a, the decision is ACK; if the decision variable is greater than threshold T_n, the decision is NACK; if the decision variable is between the two threshold, T_a and T_n, the decision is DTX.

Regarding the decision variable, it can be calculated as follows: Step 101, a single ACK/NACK value, zf° , is obtained at the output of the Rake maximum ratio combiner and can be described as

where L is the number of Rake fingers, and _?," is the despread value of the /th finger at the ith. symbol of the nth transmission of the ACK/NACK and h"_; is the corresponding channel estimate. The symbol denotes complex conjugate. G{-} takes the real part or the imaginary part depending on whether the ACK/NACK signal is mapped on the in-phase (I) or quadrature phase (Q) component of the uplink signal.

Step 102, the integration of ACK/NACK value over one slot can be described as z^w =∑;>,^(π) . (5)

Step 103, the real- valued decision variable, z, is then obtained from accumulating the integrated results of each of the N slots, i.e.

where h and s are column vectors with 10/JV elements given by the channel estimates h"_} and the despread symbols s"_t , respectively.

The noise standard deviation after despreading, σ_w, are given as follows: Step 104, the noise standard deviation after despreading, σ_w, is obtained from the following relation: The despread values, s, can be expressed as

where h is the vector of actual channel coefficients, AP takes the values AP _a and

AP_n defined by the network for ACK and NACK, respectively, w is a vector of complex noise samples with standard deviation σ_w . x is the transmitted symbol, which takes the value -1, 0, and 1 for ACK, DTX, and NACK, respectively. Finally, γ is either one or the imaginary unity depending on whether the ACK/NACK signal is mapped on I or Q. The noise standard deviation σ_w can be obtained e.g. from the despread values, s.

In applications the channel estimation is not ideal. The expected received signal amplitude after Rake maximum ratio combining is therefore VAPh E(h | h) instead of VAPh h as for ideal channel estimation. Often the channel estimation error can be considered to be zero-mean and Gaussian. Then the channel estimate h is given by h = h + ε , (9) where ε has variance

for element k.

Regarding to the first weight function and the second weight function, it can be generated as follows.

Step 105, weight function g is given by σ_n gdO = h^Λ£(h | h) , (10) σ_n +σ. and

^² = ^|lb (11)

(T₁ ² = σ₀ ² + ΔPVar(Re{h"h}|h) , (12) where ||-| denotes the vector norm: h = h^h . When AP is the power offset for the

ACK signal received, the weight function g corresponds to the first weight function; similarly, the weight function g corresponds to the second weight function when AP is the power offset for the NACK signal received. The power offset is known in the receiver.

In the limit of ideal channel estimation, g(h)=0.5]|ϊi||².

The conditional probability distribution of h given h depends on the distribution of h. In mobile communication systems the multipath components are in many cases Rayleigh fading. Then it can be shown that (h| h) is Gaussian and its mean E(h| h)

E(h I ή) = Ah, (13) where A is a diagonal matrix and its Mi diagonal element is given by

where hk is the Mh element of h, |-| denotes absolute value, and σ_S ² _k is the variance of the channel estimate error for Rake finger k. The variance Of (R^hI h) is given by

where B is the covariance matrix of (h|h), b is a column vector with its Mi element given by

and C is the matrix of correlation coefficients. Its element on row / and column k, [C],_fo is the correlation coefficient between (/z_r|h) and (/z_A|h). The correlation coefficient p between the random variables X and 7 is defined by P -. , ^C0Y(X'^f> . (16)

Typically, [C]^ is close to one if the indices i and k refer to the same path in the same slot and zero if the indices refer to different paths. The channel estimation error variances σ_S ² _k can e.g., be calculated from knowledge of the noise power and the channel estimation filter coefficients. The mean pilot signal power received in Rake finger k,

\² -σ_S ² _k over a time that is much longer than the correlation time of the channel.

Step 106, regarding the threshold between ACK and DTX, T_a, and the threshold between NACK and DTX, T_n, they can be given as following equations using the foregoing obtained results:

T. = max) a M Jh ,<J*Pn 8n<n) (18) where the coefficient a_n is typically equal to a_a- a_a should be smaller than the coefficient α₀ in equation (1) to allow T_a to be both larger and smaller than in the CFAR detector.

Referring to Fig 3, a selective ACK/NACK threshold detector is shown. The detector comprises a Rake combiner 26 including a despreader 20, a multiplier 31 and channel estimator 25; an integrator 21, a comparator 22, a noise estimator 24 and a means for generating thresholds 30. The means for generating thresholds 30 includes a first threshold generator 28, a first weight function generator 27, a second threshold generator 23 and a second weight function generator 29.

Rake combiner 26 receives the signals from MSs, and then gives a single value z{, integrator 21 accumulates the single value z_f from Rake combiner 26 to obtain values over one slot, and then accumulates the obtained values of each of the N slots to obtain a decision variable z. In other words, integrator 21 calculates the decision variable z according to equations 4, 5, 6.

Noise estimator 24 estimates noise standard deviation after despreading, σ_w, based on the outputs of the despreader 20 and then sends to the first threshold generator 28, the second threshold generator 23, the first weight function generator 27 and the second weight function generator 29, respectively. In view of the foregoing description, the noise estimation can be based on the output of the despreader (20), but there are also other options.

Channel estimator 25 generates the norm of column vectors given by channel estimates to the first threshold generator 28, and the second threshold generator 23, respectively. At the same time channel estimator 25 gives channel estimates to the first weight function generator 27 and the second weight function generator 29, respectively.

The power offset for the ACK signal, AP_n, and the power offset for the NACK signal, AP_n, are sent to the first weight function generator 27 and the second weight function generator 29, respectively. The variance of the channel estimate for Rake finger k, σ_ε ² _k , and the mean received pilot signal power in Rake finger k, E{\h_k\²}, are sent to both weight function generators, 27,29.

According to equations 10 to 16, the first weight function generator 27 generates a first weight function, g_α(h) , using the noise standard deviation after despreading σ_w from noise estimator 24, the norm of column vectors given by channel estimates from channel estimator 25, and the known power offset for ACK, and then gives to the first threshold generator 28. Similarly, the second weight function generator 29 generates the second weight function, g_n (h) , and gives to the second threshold generator 23.

According to equation 17, the first threshold generator 28 generates the first threshold T_αj using the norm of column vectors given by channel estimates from channel estimator 25, the noise standard deviation after despreading σ_w from noise estimator 24, and generates the second threshold T_α,2 using the first weight function from the first weight function generator 27 and a known power offset for ACK, then selects the minimal one from the T_α,i and T_Qι2 and sends to comparator 22. Similarly, according to equation 18, the second threshold generator 23 generates first threshold T_n j using the norm of column vectors given by channel estimates from channel estimator 25, the noise standard deviation after despreading σ_w from noise estimator 24, and generates the second threshold T_n>2 using the second weight function from the second weight function generator 29 and a known power offset for NACK in the receiver, then selects the maximal one from the T_nj and T_n>2 and sends to comparator 22.

Comparator 22 compares the decision variable from integrator 21 with the two thresholds, T_a and T_n, from threshold generators 28 and 23, respectively. If a decision variable is less than the threshold T_a, comparator 22 gives a decision of ACK; if a decision variable is greater than threshold T_n, comparator 22 gives a decision of NACK; if a decision variable is between the two thresholds, T₀ and T_n, a decision of DTX is given by comparator 22.

The detector has been evaluated by simulations. The required Ec/No for the ACK/NACK bits to fulfill the target performance requirements P(ACK|DTX)<0.01,

P(DTX|ACK) ≤O.Ol and P(ACK|NACK)≤0.0001 have been calculated. Ec is defined as the total received energy per chip on each diversity antenna. The simulations have been performed for different propagation channels.

Table 2 Required Ec/No for P(ACK|DTX) <0.01, P(DTX |ACK)<0.01, and

P(ACK|NACK)<0.0001

The simulation results are shown in Table 2 for ideal channel estimation. The coefficient a has been optimized for simulation case A and is kept constant for all cases since it is impossible to optimize it for all propagation channels, speeds etc. All coefficients used to determine T_a for the three detectors are given in the table.

Let the maximum required power be defined as the larger of the required powers for either ACK or NACK transmissions. Then it is clear from Table 2 that the new selective threshold detector has lower maximum required power than prior art, both the CFAR detector and the dynamic threshold detector, in all simulated cases.

The invention for ACK/NACK detection is not restricted to HSDPA but can be applied to any system using ACK/NACK signalling with antipodal signals, known channel estimates and power offsets of ACK/NACK to the pilot signal power.

Claims

1. A detection method for ACK/NACK signals, comprising, a. Calculating a decision variable by integration of a despread received signal; b. Calculating a threshold between ACK and DTX, which includes two values, the first being proportional to a noise standard deviation after despreading the received signal, and the second depending on an estimated received signal for ACK, calculating a threshold between NACK and DTX, which includes two values, the first being proportional to a noise standard deviation after despreading the received signal, and the second depending on an estimated received signal for NACK, c. Selecting the one with maximum absolute value from the two calculated values of the threshold between ACK and DTX ;

Selecting the one with maximum absolute value from the two calculated values of the threshold between NACK and DTX; d. Comparing the decision variable with the two selected thresholds and making decision.

2. A detection method according to Claim 1, in step b, the second values depending on an estimated received signal for ACK is calculated by weighting the square root of a power offset of ACK with a first weight function; the second values depending on an estimated received signal for NACK is calculated by weighting the square root of a power offset of NACK with a second weight function.

3. A detection method according to Claim 2, wherein weighting the square root of the power offset of ACK with a first weight function is given as:

, where AP _a represents the power offset for the ACK and g_α(h) represents the first weight function, when the ACK signal is mapped to negative sign; weighting the square root of the power offset of NACK with a second weight function is given as:

^] AP_n g_n (h) , where AP_n represents the power offset for the NACK and g_n (ή) represents the second weight function, when the NACK signal is mapped to positive sign.

4. A detection method according to Claims 2 or 3, the weight function is given as:

h)

where σ_n ₂ σ₀ ² + ΔPVar(Re{h^ff h}|h) ; wherein σ_w denotes the

noise standard deviation after despreading, |.|j denotes the vector norm: h = h h , h

are column vectors given by the channel estimates, and E(h.h) represents mean of the

conditional probability distribution of h given h in case of Rayleigh fading, the EQx h) is calculated as: E(h | h) = Ah, where A is a diagonal matrix and its Mi diagonal element is given by

wherein h_k is the Mh element of h, |-| denotes absolute value, and σ_St is the variance of the channel estimate error for Rake finger k,

is the mean pilot signal power received by Rake finger Jc, and when AP= AP _a, g(h)= g_a (h) ; when AP= AP_n, g(h)= g_H(h) .

5. A detection method according to Claim 1, in step b, for the threshold between

ACK and DTX, when the ACK signal is mapped to negative sign, the first value being proportional to a noise standard deviation after despreading is given as: - a_aσ_v for the threshold between NACK and DTX, when the NACK signal is mapped to positive sign, the first value being proportional to a noise standard deviation after despreading is given as: a_nσ_w h wherein the coefficients a_a and a_n meet a target performance requirement.

6. An ACK/NACK detector comprising, a Rake combiner (26) for despreading and performing channel estimation of a received signal to generate a single value, which includes at least a despreader (20) and a channel estimator (25), an integrator (21) for integrating the single value of one slot and accumulating the integrated results of each of N slots to obtain a decision variable, a comparator (22) for comparing the decision variable from the integrator (21) with a threshold between ACK and DTX and a threshold between NACK and DTX given from a means for generating thresholds (30), respectively, a noise estimator (24) for performing estimation of despread signals from the despreader (20) to obtain a noise standard deviation after despreading, wherein the means for generating thresholds (30), is used for calculating a threshold between ACK and DTX that includes two values, the first being proportional to a noise standard deviation from the noise estimator (24), and the second depending on an estimated received signal for ACK, calculating a threshold between NACK and DTX that includes two values, the first being proportional to a noise standard from the noise estimator (24), and the second depending on an estimated received signal for NACK, Selecting the one with maximum absolute value from the two calculated values of the threshold between ACK and DTX ;

7. An ACK/NACK detector according to Claim 6, the means for generating thresholds (30) comprises, a first weight function generator (27) for generating a first weight function through using the channel estimates from the Rake combiner (26), the noise standard deviation from the noise estimator (24), a known power offset of the ACK, estimated channel estimation error variances and an estimated mean of the received pilot signal powers; a first threshold generator (28) for generating the first value through using norm of the channel estimates from the Rake combiner (26) and the noise standard deviation from the noise estimator (24), generating the second value through weighting the square root of a power offset of the ACK with the first weight function from the first weight function generator (27), and giving the one with maximum absolute value to the comparator (22); a second weight function generator (29) for generating a second weight function through using the channel estimates from the Rake combiner (26), the noise standard deviation from the noise estimator (24), a known power offset of the NACK, the estimated channel estimation error variances and the estimated mean of the received pilot signal powers; a second threshold generator (23) for generating the first value through using norm of the channel estimates from the Rake combiner (26) and the noise standard deviation from the noise estimator (24), generating the second value through weighting the square root of a power offset of the NACK with the second weight function from the second weight function generator (29), and giving the one with maximum absolute value to the comparator (22).

8. An ACK/NACK detector according to Claim 7, when the ACK signal is mapped to negative sign, the first threshold generator (28) generates the first value being proportional to a noise standard deviation after despreading as: - a_aσ_w h ; when the NACK signal is mapped to positive sign, the second threshold generator (23) generates the first value being proportional to a noise standard deviation after despreading is given as: αr_πσJh ; wherein the coefficients a_a and a_n meet a target performance requirement.

9. An ACK/NACK detector according to Claims 7 or 8, the first threshold generator (28) generates the second value given as the following: -«jAP_ag_a(h) , where AP _a

represents the known power offset for the ACK and g_a (h) represents the first weight function when the ACK signal is mapped to negative sign; the second threshold generator (23) generates the second value given as the following: «jAP_ng_n (h) , where AP_n represents the known power offset for the NACK and

g_π(h) represents the second weight function when the NACK signal is mapped to positive sign.

10. An ACK/NACK detector according to Claim 7, the first weight function

generator (27) generates the first weight function given as: g(h) = — h^ E(h | h) cx_n + σ,

where σ₀ ² = -σ* |h||² ,

+ ΔP_αVar(Re{h*h}|h) , the second weight function generator (29) generates the second weight function σ, given as: g(h) = ^• 0 h u H*E(h | h) σ_n + σ.

where σl 2 = — 1 σ h

+ ΔP_HVar(Re{h*h}|h) ,

wherein σ_w denotes the noise standard deviation after despreading, ll-l denotes the l|2 vector norm: h = h h , h are column vectors given by the channel estimates, and

£(h h) represents mean of the conditional probability distribution of h given h in case of Rayleigh fading; the E(hh) is calculated as: E(h | h) = Ah, where A is a diagonal matrix and its Mi diagonal element is given by

wherein h^ is the Mi element of h, |-| denotes absolute value, σ S²k is the variance of the channel estimate error for Rake finger k, E{\h^} is the mean pilot signal power received in Rake finger k.