WO2010142449A1

WO2010142449A1 - Process for detecting transmitted symbols based on a ternary alphabet and corresponding receiver

Info

Publication number: WO2010142449A1
Application number: PCT/EP2010/003500
Authority: WO
Inventors: Rodolfo Verzegnassi; Jianliang Shi
Original assignee: ST Ericsson SA; ST Ericsson France SAS
Current assignee: ST Ericsson SA; STMicroelectronics Grand Ouest SAS
Priority date: 2009-06-12
Filing date: 2010-06-11
Publication date: 2010-12-16
Anticipated expiration: 2011-12-12
Also published as: JP5615912B2; JP2012529815A; US8315580B2; IN2012DN00232A; EP2262194B1; EP2262194A1; US20120157104A1

Abstract

1. Detection process of a communication channel based on a ternary alphabet, said channel transmitting a signal transporting the same transmitted symbol for N sequential slots within one Time Transmission Interval), involving the steps of: - subtracting (21) the received signal (k) to signal (k+1) so as to eliminate signal contribution and isolate the noise contribution; - estimating (22) the noise variance v; - summing (23) all consecutive received signals of the same TTI so as to isolate the transmitted signal contribution, plus some residual noise; - squaring (24) the variable so as to obtain a chi squared distribution, central if the transmitted signal was 0, non-central if it was +1 or -1. - determining (25) a threshold proportional to said variance; - using (26) said threshold so as to perform detection of the communication channel.

Description

Process for detecting transmitted symbols based on a ternary alphabet and corresponding receiver

Technical field

The invention relates to the field of wireless communication and more particularly to a detection process of a channel transmitting symbols based on a ternary alphabet and a receiver for doing the same.

Background Art

High Speed Uplink Packet Access (HSUPA) is a 3G mobile telephony protocol providing high speed uplink transmission.

In the 3GPP Release 6 (HSUPA) standard there are two downlink channels, namely E-RGCH and E-HICH, used for controlling the transmission power of the UE and which are based on the use of a ternary alphabet. Those two channels transmit one among three different signals to the User Equipment (UE) : UP, DOWN or HOLD for E-RGCH₁ ACK, NACK or DTX for the E-HICH. These signals are merely transmitted as +1 , -1 or 0, and are received as +S, -S, or 0, possibly plus noise.

As known by the skilled man, when the UE has to detect a received signal that could be +S, -S or 0, plus some noise N, and a decision has to be taken on what was transmitted, it is necessary to set two thresholds: -a first threshold between 0 and +S in order to detect a +1 or a 0

- l - -a second threshold between 0 and -S so as to detect a 0 or a -1

Usually the two thresholds are symmetrical so the problem is really reduced to setting one threshold.

Now, as shown in Figure 1 , the use of a ternary alphabet results in the existence of three gaussian distributions, referred to as 11 , 12 and 13, requiring the need to determine two thresholds for minimizing the misdetection as well as false alarms.

Now, the problem is that in the case of the two channels E-RGCH and E- HICH in HSUPA, but not exclusively, one has unfortunately no information on the value of S, nor on the noise N, deriving in a distribution of the signal around the three values -1 , 0 and +1.

Consequently, receiving a signal Y alone does not allow the discrimination between the three possibilities : just noise N, or +S + N or, at least -S + N.

Therefore, there is a need for a specific mechanism allowing the determination of an appropriate threshold to be used in the detection process of a channel transmitting symbols based on a ternary alphabet.

Summary of the invention

It is an object of the present invention to provide a new process achieving determination of the appropriate threshold to be used in a detection process of a channel transmitting symbols based on a ternary alphabet.

It is a further object of the present invention to provide a process for determining the appropriate threshold to be used for the detection of the E-RGCH and E-HICH channels when one has no information of the amplitude/power of the received signal. It is still another object of the present invention to provide a detection process of the E-RGCH and E-HICH channels which can further provide the control of the percentage of misdetection and false alarms.

These and other objects of the invention are achieved by means of a detection process of a signal which is transmitted over a communication channel, said signal consisting in a sequence of a same symbol which is repeated during N sequential slots within one Time Transmission Interval (TTI), said symbol being extracted from a ternary alphabet, the process involving the following stepts to be performed separately on every received signal, ie sequence of N symbols:

- subtracting the symbol received at a certain time instant k to the following symbol received at the following time instant k+1 so as to eliminate symbol contribution and isolate the noise contribution;

- estimating the noise variance v by performing the preceding operation N-1 times , that is to say over all N symbols in the received signal and thus isolating N-1 noise contributions;

- summing all consecutive received signals of the same TTI so as to isolate the transmitted signal contribution, plus some residual noise in order to generate a sum which will later be compared to a given threshold for the purpose of determine which particular element of the ternary alphabet was originally transmitted;

- squaring said sum so as to obtain a chi squared distribution, central if the transmitted signal was 0, non-central if it was +1 or -1. - determining a threshold proportional to said variance;

- using said threshold so as to perform detection of the communication channel

In one particular embodiment a filtering is applied on the estimated variance, such as a low-pass filter , before using it for the determination of the threshold. The process is particularly suitable for detecting the E-HICH and E-RGCH communication channels which are decoded after dispreading and signature correlation.

In one particular embodiment, a look-up table is used for the purpose of providing one particular coefficient to be multiplied to the estimation of the variance so as to provide an accurate control of the detection process. The look-up table therefore provides different thresholds, corresponding to different scenarios, i.e. to different required ratios of missed detection (a +S or a -S detected as a 0) and false detection (a 0 detected as a +S or a -S).

A first scenario corresponds to the situation of a transmission of E-HICH channel in single link, and the look-up table provides an appropriate coefficient matching that scenario.

Alternatively, the look-up table provides one particular coefficient corresponding to the situation of a transmission of E-HICH channel in inter-cell handover, where the RLS (Radio Link Set) does not contain the serving E-DCH cell.

Further, the look-up table provides one particular coefficient corresponding to the situation of a transmission of E-HICH channel in inter-cell handover, where the RLS (Radio Link Set) contains the serving E-DCH cell.

In one embodiment, the look-up table provides one particular coefficient corresponding to the situation of a transmission of E-RGCH channel in single link.

In one further embodiment, the look-up table provides one particular coefficient corresponding to the situation of a transmission of E-RGCH channel in inter-cell handover where the E-RGCH belongs to the non-serving RL (Radio Link)

The invention further achieves a receiver for receiving a signal which is transmitted over a communication channel, said signal consisting in a sequence of the same symbol which is repeated during N sequential slots within one Time Transmission Internval, said symbol being taken from a ternary alphabet.

The receiver comprises

- means for subtracting the symbol received at a certain time instant k to the following symbol received at the following time instant k+1 so as to eliminate signal contribution and isolate the noise contribution;

- means for summing (23, 11 , 12) all consecutive received signals of the same TTI so as to isolate the transmitted signal contribution, plus some residual noise, said sum being later compared with a given threshold so as to determine which element of the ternary alaphabet was originally transmitted ;

- means for squaring said sum so as to obtain a chi squared distribution, central if the transmitted signal was 0, non-central if it was +1 or -1 ;

- means for determining a threshold proportional to said variance;

- means for using said threshold so as to perform detection of the communication channel

The invention is particularly useful for achieving a telephone mobile fitted with a receiver achieving detection of the E-HICH and E-RGCH channels in the case of 3GPP Release 6 (HSUPA)

Description of the drawings

Other features of one or more embodiments of the invention will best be understood by reference to the following detailed description when read in conjunction with the accompanying drawings. Figure 1 illustrates the distribution (of the Gaussian type) of the received signal around the three values forming the ternary alphabet.

Figure 2 illustrates one embodiment of a process for detecting the E-HICH and E-RGCH channels.

Figure 3 illustrates one particular embodiment of a block diagram of the process for detecting the E-HICH and E-RGCH channels.

Description of the preferred embodiment

There will now be described one preferred embodiment of the invention which achieves the detection of the E-HICH and E-RGCH channels in the case of 3GPP Release 6 (HSUPA).

Briefly, the principle takes advantage of the fact that the signal (that may be either E-RGCH or E-HICH) transports the same transmitted symbol for N sequential slots, where N depends on the length of the TTI (Time Transmission Interval), whereas the noise contribution on each received signal is different. Therefore, by subtracting received signal (k) to signal (k+1) one can eliminate the signal contribution (if this is done within the same TTI, assuming the channel conditions do not vary much within the same TTI) and one isolates the noise contribution. From the noise contributions one can estimate the noise variance v. On the other hand, if one sums all consequent received signals of the same TTI, one can isolate the transmitted signal contribution, plus some residual noise. This variable follows a normal distribution where the mean is the unknown transmitted signal and the variance is what was estimated. By squaring the variable, one obtains a chi squared distribution, central if the transmitted signal was 0, non-central if it was +1 or -1. The cumulative distribution function is proportional to the variance, so we can set a threshold proportional to the variance in order to fix the % of values that pass the threshold. In such way we can control the % of missed detections and false detections.

With respect to Figure 2, there is now described with more details the basic steps of one process for achieving detection of a communication channel based on a ternary alphabet, such as the E-HICH and E-RGCH channels.

The process starts with a step 21 which consists in the extraction of the communication channels to be detected and the subtracting of two consen itiv^p received signals so as to isolate the noise contribution. As known by the skilled man, in the case of the E-HICH and E-RGCH communication channel for the 3GPP TS 25.211 ( which is one particular embodiment), the indicator is required to be transmitted repeatedly for a number N of times (3, 12 or 15), each time modulated with a defined signature sequence. Different signature sequences are orthogonal between them so that transmitted signal on the E-HICH and the E-RGCH channel, which share the same OVSF code, do not interfere with each other.. After correlating the received signal with a local signature sequence which is the same as the one used in transmission, the correlation output at transmit repetition time ^l can be expressed as

9?, = h, x S₁ + N,

where hi is the channel coefficient, Si is the transmitted signal correlated with the local signature, the first term hi x Si is the usefui component for detection of indicator and the second term Ni is due to the background noise.

Both components depend on the propagation characteristics of the signal transmission since coherent demodulation is applied in Slave Rake. However the noise component is assumed to be Gaussian. The signal component depends on the channel estimation which is proportional to the channel coefficient.

Assuming the channel fading to be slow enough, if one correlates in the next repetition time (time' + l ) the received symbol level signal with the inverse of signature sequence of interest, one has :

*,_+I = ^_+l * S,.₊₁ + N_i+1 + I,._+l

where the term ^I/+l corresponds to the interference between the signature sequence of interest and those not of interest. This term is negligible and can be ignored.

Once the appropriate communication channel has been extracted, the process proceeds with a step 22 which consists in the estimation of the variance of the noise during a period coinciding with the TTI.

This is achieved by performing, in one embodiment, first a ZF (zero-forcing)

algorithm, ie divide ^l and ^;+1 by respectively ' and ^l+] and sums

^' j^M I h ⁿ' i⁽+⁺l^l . One notices that the noise component is essentially independent of the signature, and that rnuitipiying by the inverse of the signature

and summing ^/+1 ' ^/+1 to ' ' is the same as subtracting ⁱ⁺¹ ' ^;+1 to

I 9

By repeating the procedure over the whole TTI ( »+i /

and so on) so as to compute the

variance of the resulting noise symbols ^/+1 , ^/+2 etc, one has :

= Var(N_i) + Var(N_M)

= ^iV o It should be noticed that, due to scaling factors applied during the signal processing in the receiver, a scaling is also applied to the RSCP value i.e. our h ² estimation of ' .

Once we have estimated the signal variance we can consider that the signal has a Gaussian distribution with the mean value proportional to the useful signal (+1 , 0 or -1) and the variance which was estimated.

The process then proceeds to a step 23, which is a summation of the received signal over the N times it is transmitted, together with the application of a scaling for dynamic purposes (basically a division by K with K being 3, 5, 12...) in order to produce a value (corr_sig) representative of the received signal.

Then, in a step 24, the process proceeds with the squaring of the means received signal in order to generate a chi distribution of that signal.

Indeed, one has noticed that the difference between the cumulative distribution function of a normal distribution centered in 0 and of one centered in S is simply a shift of S. On the other hand, the difference between the cumulative distribution function of a chi-square distribution and the one of a non-central chi- square distribution is not a simple shift. The non-central chi-square distribution has a flatter profile, which means that higher values are more probable than in a chi- square distribution, even compensating the S shift.

Therefore, the embodiment illustrated in figure 2 takes advantage of such property and the squaring step 24 results in that the signals corresponding to a +/-1 will statistically have more higher values than the ones corresponding to a 0, and this spread will be higher than in a normal distribution case.

Consequently, the squaring step 24 applied on the received signal results in reducing the probabilities of false alarm or missed detection. This is already a first significant advantage resulting from the process illustrated in figure 2.

In one embodiment, the detection process is based on a threshold which is determined, in a step 25. as being proportional to the estimated variance.

It has been observed, however, that for a constant value of the CDF (e.g.

95%), the value of the threshold grows linearly with the variance of the normal distribution that generated the chi-square. To compute the linearity coefficient by using the formula for chi-square and non-central chi-square distributions would be clearly too power-consuming.

In order to solve that problem and also for the purpose of providing a more accurate control of the detection process in the case of different situations, the process of figure 2 further involves a LUT (look-up table) which is read for the purpose of providing coefficients to multiplied to the noise variance estimation in order to fix the proper threshold for each different transmission scenario and 3GPP requirement.

In one particular optional embodiment, a filtering process is applied to the noise variance before the multiplication with the coefficient read from the LUT, so as to smooth the variability of the variance estimation.

Once determined, the threshold can be used by the process, in a step 26, so as to achieve proper detection of the communication channel.

A block diagram implementing the process described above is particularly illustrated in figure 3, which can be used for detecting one ternary alphabet communication channel, be it E-RGCH or E-HICH .

One assumes that the entries of the block diagram, respectively referred to as reference 1 and 2, consist in the estimation of the channel power and the received signal (E-RGCH or E-HICH) already separated from other channels via orthogonal sequences and spreading codes.

- i i - More particularly the channel power is represented by a parameter RSCPJnt (Received Signal Code Power) , and the received signal is represented by a parameter accu_sum_int ).

Figure 3 shows that, as mentioned above, a scaling is applied on the RSCP value by means of a divide-by-64 block 3 having an input receiving the value RSCPJnt .

A dividing block 4 has two inputs, a first input receiving the signal (accu_sum_int) and a second input receiving the value generated by block 3.

It can be seen that blocks 3 and 4 embody the ZF zero forcing process of step 22.

One then sees that the output of block 4 is subject to two distinctive processing paths, a first path arranged for the estimation of the variance as described above, and a second path arranged for the computation of the average of the received signal over the whole TTI.

The first path includes the succession of blocks 6, 7, 8 and 10, together with the two delays 5 and 9, respectively having an input connected to the output of block 4 and the output of block 8.

Block 6 is a subtractor having a first input receiving the output of block 4 and a second input receiving the output of delay 5 providing the difference of two consecutive received signals, needed for the estimation of the noise contribution.

Block 7 performs a squaring of the output of substractor 6.

Block 8 is an adder having a first input respectively receiving the output of squaring block 7 and the output of delay 9, so as to provide the whole summation of all noise contributions.

Block 10 is a divider by a scalar - (length_tti_int) so as to normalize the value of the variance estimation. The second path is based on adder 12, having a first input receiving the output of block 4 and a second input receiving the delayed version of adder 12, so as to achieve the summation of all the received signals - namely a value corr_sig - over the whole period TTI.

With respect to figure 3 again, the variance estimation is then inputted into an optional filtering block 13 - basically a low pass filter - before being forwarded to a first input of a multiplier 14, a second input of which receives the output of the Look- up-table 15.

On the other hand, the summation of all the received signals, namely signal corr_sig is entered into a squaring block 16 so as to achieve the chi - conversion of the distribution.

The two outputs of multipliers 14 and squaring block 16 are entering into a decision block 17 which, also receives the value of the sign of the corr_sign before it is squared by squaring block 16.

Decision block performs a comparison of the squared received signal, together with the consideration of the sign, with the threshold computed in accordance with the contents stored in the LUT table, so as to achieve the determination and the detection of the signal received from the considered communication channel, E-HICH or E-RGCH.

It will now be discussed the arrangement of the look-up-table.

Clearly, different LUT may be considered for the purpose of providing appropriate coefficients to be used for the determination of the threshold, and such coefficients will closely depend on the particular application being considered, and particularly on the communication channels ( E-HICH and E-RGCH) to be detected.

In one particular embodiment, the LUT (Look-up table) consists in 9 values, each corresponding to one specific predetermined scenario. The 9 scenarios have been chosen so as to fully match with the 3GPP specification TS 34.121-1 V7.5.0 (2007-06), chapter 10 "Performance requirement (E-DCH)". each scenario corresponding to one particular set of requirements.

Each of the 9 values of the LUT, multiplied by the estimate of the variance, will give a decision threshold that will satisfy the good detection requirements proper to that scenario.

The 1^st and 2^nd scenarios being coded in the LUT, correspond to a transmission of E-HICH channel in single link, known to the skilled man. The reader is invited to refer more particularly to Chapter 10.2.1 :

• In case of integration length = 12 slots ( note: "Integration length" is the number of times that a same E-HICH or E-RGCH symbol is repeated, namely 3, 12 or 15.) o table 10.2.1.1.5.2 for MACK (Missed ACK => a '1' is transmitted and is interpreted as a '0' or '-1') o table 10.2.1.1.5.3 for FACK (False ACK => a '0' is transmitted and is interpreted as a '1 ') o to satisfy these requirements the decision threshold shall be set by multiplying the estimated variance with a coefficient in the LUT denominated Li

• In case of integration length = 3 slots o table 10.2.1.2.5.2 for MACK (Missed ACK => a '1 ' is transmitted and is interpreted as a '0' or '-1') o table 10.2.1.2.5.3 for FACK (False ACK => a '0' is transmitted and is interpreted as a '1 ') o to satisfy these requirements the decision threshold shall be set by multiplying the estimated variance with a coefficient in the LUT denominated L? The 3^rd and 4^th scenarios being coded in the LUT correspond to a transmission of E- HICH channel in inter-cell handover, where the RLS (Radio Link Set) does not contain the serving E-DCH cell.

The requirements for these scenarios are more particularly listed in chapter 10.2.2.1 :

• In case of integration length = 12 slots o table 10.2.2.1.1.5.2 for MACK (Missed ACK => a '1' is transmitted and is interpreted as a '0' or '-1 ') o table 10.2.2.1.1.5.3 for FACK (False ACK => a '0' is transmitted and is interpreted as a '1 ') o to satisfy these requirements the decision threshold shall be set by multiplying the estimated variance with a coefficient in the LUT denominated L₃

• In case of integration length = 3 slots o table 10.2.2.1.2.5.2 for MACK (Missed ACK => a '1' is transmitted and is interpreted as a '0' or '-1 ') o table 10.2.2.1.2.5.3 for FACK (False ACK => a '0' is transmitted and is interpreted as a '1 ') o to satisfy these requirements the decision threshold shall be set by multiplying the estimated variance with a coefficient in the LUT denominated L₄

The 5^th and 6^th scenarios correspond to a transmission of E-HICH channel in inter- cell handover, where the RLS (Radio Link Set) contains the serving E-DCH cell. The requirements for these scenarios are listed in chapter 10.2.2.2:

• In case of integration length = 12 slots o table 10.2.2.2.1.5.2 for MACK (Missed ACK => a '1' is transmitted and is interpreted as a '0' or '-T) o table 10.2.2.2.1.5.3 for FACK (False ACK => a '0' is transmitted and is interpreted as a 'T) o to satisfy these requirements the decision threshold shall be set by multiplying the estimated variance with a coefficient in the LUT denominated L₅ • In case of integration length = 3 slots o table 10.2.2.2.2.5.2 for MACK (Missed ACK => a '1' is transmitted and is interpreted as a '0' or '-1') o table 10.2.2.2.2.5.3 for FACK (False ACK => a '0' is transmitted and is interpreted as a '1 ') o to satisfy these requirements the decision threshold shall be set by multiplying the estimated variance with a coefficient in the LUT denominated L₆

The 7^th and 8^th scenarios correspond to a transmission of E-RGCH channel in single link.

The requirements for these scenarios are described in chapter 10.3.1 :

• In case of integration length = 12 slots o table 10.3.1.1.5.2 for MUD (Missed UP or DOWN => either a T is transmitted and is interpreted as a '0' or a '-1', or a '-1' is transmitted and is interpreted as a '0' or a '1') o table 10.3.1.1.5.3 for MHOLD (Missed HOLD => a O' is transmitted and is interpreted as a '1 ' or a '-1 ') o to satisfy these requirements the decision threshold shall be set by multiplying the estimated variance with a coefficient in the LUT denominated L₇

• In case of integration length = 3 slots o table 10.3.1.2.5.2 for MUD (Missed UP or DOWN => either a T is transmitted and is interpreted as a '0' or a '-1 ', or a '-1' is transmitted and is interpreted as a '0' or a '1 ') o table 10.3.1.2.5.3 for MHOLD (Missed HOLD => a '0' is transmitted and is interpreted as a '1 ' or a '-1 ') o to satisfy these requirements the decision threshold shall be set by multiplying tne estimated variance witn a coefficient in ihe LUT denominated L₈ The 9^th scenario corresponds to a transmission of E-RGCH channel in inter-cell handover where the E-RGCH belongs to the non-serving RL (Radio Link) The requirements for this scenario are described in chapter 10.3.2:

• There is only one case of integration length = 15 slots o table 10.3.2.5.2 for MHOLD (Missed HOLD => a '0' is transmitted and is interpreted as a '1 ' or a '-1 ') o table 10.3.2.5.3 for MDOWN (Missed DOWN => a '-1' is transmitted and is interpreted as a '0' or a '1 ') o to satisfy these requirements the decision threshold shall be set by multiplying the estimated variance with a coefficient in the LUT denominated Lg

As described above with details, the invention solves the problem of determination of a threshold to be used in the decision making process based on an estimation of the noise variance and the assumption that the noise is originally AWGN (Additive White Gaussian Noise). To achieve this, one takes advantage of the statistical characteristics of Gaussian distributions and of its closely-related non- central chi squared distribution. This will permit us to set a threshold to detect whether a +S, a -S or a 0 was received.

The invention provides optimized detection process achieving proper minimizing of false alarms ( a "0" which is being detected as a "1" or a "-1" ) and also misdetection ( either a "+1" or a "-1" being detected as a "0"). This is achieved even though one has no information on the amplitude S of the signal, nor on the value of the noise. The threshold determination is based, as described below, on noise variance estimation, zero-forcing, and on the properties of non-central chi-square distribution.

This algorithm has been tested against the 3GPP E-HICH/E-RGCH detection requirements and it satisfies all requirements with a minimum margin of 1.2 dB.

Claims

1. Detection process of a signal that is transmitted over a communication channel, said signal consisting in a sequence of a same symbol which is repeated during N sequential slots within one Time Transmission Interval (TTI), said symbol being extracted from a ternary alphabet, said process involving the steps of:

- subtracting (21) the symbol received at a certain time instant k to the following symbol received at the following time instant k+1 so as to eliminate symbol contribution and isolate the noise contribution;

- estimating (22) the noise variance v by performing said operation (21) N-1 times corresponding to a period of N symbols in the received signal and thus isolating N-1 contributions;

- summing (23) all consecutive received signals of the same TTI so as to isolate the transmitted signal contribution, plus some residual noise so as to generate a sum which will be used in a comparison step with a predetermined threshold so as to determine which element of the ternary alphabet was originally transmitted; - squaring (24) said sum (23) so as to obtain a chi squared distribution;

- determining (25) a threshold proportional to said variance;

- using (26) said threshold for said comparison so as to perform detection of the communication channel

2. Detection process according to claim 1 characterized in that the estimated variance is subject to a filtering before being used for determining said threshold.

3. Detection process according to claim 2 characterized in that said filtering is a low- pass filtering for averaging the variation of the variance over the whole TTI.

4. Detection process according to anyone of claims 1 to 3 characterized in that said communication channel is the E-HICH channel decoded after dispreading and signature correlation, e.g. such as defined in 3GPP specification (TS 25.211 V7.2.0. (2007-05).

5. Detection process according to anyone of claims 1 to 3 characterized in that said communication channel is the E-RGCH channel decoded after dispreading and signature correlation, e.g. such as defined in 3GPP specification (TS 25.211 V7.2.0. (2007-05).

6. Detection process according to claims 4 or 5 characterized in that it said threshold is computed from the multiplication of said varience with a coefficient which is provided by a look-up-table.

7. Detection process according to claims 4 or 5 characterized in that said look-up table provides one particular coefficient corresponding to the situation of a transmission of E-HICH channel in single link.

8. Detection process according to claims 4 or 5 characterized in that said look-up table provides one particular coefficient corresponding to the situation of a transmission of E-HICH channel in inter-cell handover, where the Radio Link Set R. L. S. does not contain the serving E-DCH cell.

9. Detection process according to claim 4 or 5 characterized in that said look-up table provides one particular coefficient corresponding to the situation of a transmission of E-HICH channel in inter-cell handover, where the Radio Link Set R.L.S contains the serving E-DCH cell.

10. Detection process according to claims 4 or 5 characterized in that said look-up table provides one particular coefficient corresponding to the situation of a transmission of E-RGCH channel in single link.

11. Detection process according to claim 4 or 5 characterized in that said look-up table provides one particular coefficient corresponding to the situation of a transmission of E-RGCH channel in inter-cell handover where the E-RGCH belongs to the non-serving Radio Link RL .

12. Receiver for receiving a signal which is transmitted over a communication channel, said signal consisting in a sequence of a same symbol repeated during N sequential slots within one Time Transmission Interval, said symbel being extracted from a ternary alphabet, said receiver including:

- means for subtracting (21 , 5, 6) the symbol received at a certain time instant k to the following symbol received at the following time instant k+1 so as to eliminate symbol contribution and isolate the noise contribution;

- means for summing (23, 11 , 12) all consecutive received signals of the same TTI so as to isolate the transmitted signal contribution, plus some residual noise so as to generate a sum which is to be compared with a predetermined threshold so as to determine which element of the ternary alphabet was originally transmitted;

- means for squaring (24, 16) said sum so as to obtain a chi squared distribution,

- means for determining (25, 14, 15) a threshold proportional to said variance;

- means for using (26, 17) said threshold so as to perform detection of the communication channel

13. Receiver according to claim 12 characterized in that said communication channel is the E-HICH channel or a E-RGCH channel decoded after dispreading and signature correlation.

14. Receiver according to claim 13 characterized in that it includes a look-up-table (LUT) for the purpose of generating one coefficient in each of the following situations :

- transmission of E-HICH channel in single link ; - transmission of E-HICH channel in inter-cell handover, where the RLS (Radio Link Set) does not contain the serving E-DCH cell ;

- a transmission of E-HICH channel in inter-cell handover, where the RLS (Radio Link Set) contains the serving E-DCH cell ;

- a transmission of E-RGCH channel in single link ; - a transmission of E-RGCH channel in inter-cell handover where the E-RGCH belongs to the non-serving RL (Radio Link).

15. Mobile telephone comprising a receiver as defined in anyone of claims 12-14.