Procedure for the recognition of active code sequences

Abstract

A procedure for determining active code sequences of a plurality of overlaid code sequences (ca(v)·ga), wherein the active code sequences have a greater gain factor (ga) than the inactive code sequences, includes: (1) formation of a cost function (L) dependent upon unknown estimated values ({tilde over (g)}a,ĝa) of the gain factors (ga) of composite code sequences (ca(v)·ga), (2) partial differentiation of the cost function (L) in accord with the unknown estimated values ({tilde over (g)}a ĝa) of the gain factors (ga), (3) formation of an equation system from the presupposition that all partial differentials of the cost function (L) are zero and a minimum of the cost function (L) is present, (4) determination of the estimated values ({tilde over (g)}, ĝa) of the gain factors (ga) by solving the equation system, and (5) determining that an active code sequence exists, if the estimated values ({tilde over (g)}a, ĝa) of the corresponding gain factors (ga) are greater than a specified threshold value.

Description

BACKGROUND OF THE INVENTION

The invention concerns a procedure for the determination of active code sequences, in particular, of identification sequences (midambles) in mobile radio systems, especially for the TDD-Mode of the Standards 3GPP.

The employment of Time Division Duplex (TDD) for the uplink (connection of the mobile station to the base station) and the downlink (connection of the base station to the mobile station) for various mobile radio standards has been made known, for example, by T. Ojanperä, R. Prasad “Wideband CDMA for Third Generation Mobile Communications”, Artech House, 1998, ISBN 0-89006-735-X, Pages 261 to 277. Therein, a TDD Modus is presented, in which each downlink and uplink slot of the TDD framework is split up into a plurality of code channels with an orthogonal spreading code. Each code channel comprises a first data zone, a second data zone and an identification sequence (midamble) placed between the said data zones. Although the data chip sequences, because of multiplication with orthogonal spreading codes, are orthogonal to one another, the identification sequences (midambles) are not orthogonal to each other.

In certain operational situations checks must be made, to see which identification sequences (midambles) are active. Fundamentally, this could be done by correlation of the received data sequence in the midamble area, inclusive of all allowable midambles (identification sequences). By the squaring of the correlation-coefficients, a capacity centered evaluation can be achieved. If the square of a defined correlation coefficient, in relation to the entire capacity of the sum of the midambles oversteps in a logarithmic scale a certain threshold, then the conclusion could rest on an active midamble, and hence on an active code channel. This procedure adapts itself, however, only to the detection of the active midambles, providing that the midamble-code-sequences exhibit a satisfactory cross-correlation characteristic. In the case of short midamble-code-sequences and a high degree of disturbance, then, because of the poor cross correlation characteristics, erroneous or failed detection comes to the fore, because no clear-cut threshold can be found, which separates a valid hypothesis from a faulty hypothesis.

All references cited herein are incorporated herein by reference in their entireties.

BRIEF SUMMARY OF THE INVENTION

Thus the invention has the purpose of creating a procedure for the recognition of code sequences along with a corresponding computer program, which operates with a high degree of reliability, even in the case of small signal/noise relationships and poor cross correlation characteristics of the to-be-recognized code signals.

The basis of the invention is the recognition, that by means of the establishment of a cost-function and the partial differentiation after the gain factors of the individual code sequences a particularly secure procedure for the detection of the active code sequences can be created.

It is particularly of advantage to compute and to store, on a one-time basis, the matrix of the equation systems, which arises from the partial differentiation of the cost-function. Meanwhile, the actual detection procedure, in that case, can then be referred back onto these previously computed and saved coefficients, so that the invented procedure, with a relatively small investment in implementation, can be accomplished with a relatively small time spent in computation.

BRIEF DESCRIPTION OF SEVERAL VIEWS OF THE DRAWINGS

In the following, an embodiment of the invention will be more closely described with the aid of the drawings, wherein:

FIG. 1 is a TDD framework of the ESTI wideband-CDMA TDD mode, in which the invented procedure can be applied,

FIG. 2 is a model of the sender based on the invented, procedure, and

FIG. 3 is a model of the transmission channel, said model being based on the invented procedure.

DETAILED DESCRIPTION OF THE INVENTION

FIG. 1 shows a TDD-framework of the ETSI (European Telecommunications Standards Institute) wideband—CDMA (Code Division Multiple Access) TDD (Time Division Duplex) mode, whereby, with the aid of FIG. 1, an application example of the invented procedure can be explained. The application of the invented procedure is, however, not only for the TDD-mode by a mobile radio system, but entirely generally applicable for the recognition of code signals with little signal/noise ratio.

In the case of the TDD-Mode, different time slots of the TDD-frame are used for the down-link (connection between base station BS and the mobile station MS) and the up-link (connection between the mobile station MS and the base station BS) in a time-multiplex process. In the exemplary schematic presented in FIG. 1, some slots are continually in the down-link, other slots are placed continually in up-link service, while a plurality of slots are variable and can be assigned to the up-link (MS TX) and down-link (BS TX). Each slot arrays itself in different code channels K_o to K_N1−1. Each code channel comprises a first data area Dat.1 and a second data area Dat.2 and an identification sequence placed between the said data areas, which, hereafter will be referred to as “midamble”. Each code channel K_a can be assigned to a different midamble C_a(1)·g_a, whereby Ca(1) denotes the code sequence of the midamble and g_a stands for the gain factor. Then, according to whether the corresponding code channel K_a is active, then also its dedicated midamble is active, that is, for active code channels where g_a>0 is valid. Conversely, for non-active code channels the valid expression is g_a=0.

Since, in the data areas Dat.1 and Dat.2, the data symbols with orthogonal spreading codes are multiplied and the chip sequences, on this account, are also orthogonal, the code sequences of the midambles of the various code channels, namely, K_o to K_N1−1 are not orthogonal.

In certain operational situations, the determination must be made, as to which midambles are active and which midambles are not active. To this purpose, the present invention can be utilized, which will be described and explained below.

In the description hereafter, the following formula symbols will be used:

• • v time index on the chip surface
  • c_a(v) normalized capacity, descrambled, non-deformed chip signal of the a-ten midamble code
  • g_a gain factor of the a-ten midamble code
  • {tilde over (g)}_a {tilde over (g)}_a, ĝ_a and the estimated gain factor of the a-ten midamble code
  • j square root of −1
  • M relative capacity threshold
  • n(v) additive disturbance
  • N length of the midamble code
  • N₁ number of the midamble codes
  • r(v) measurement signal
  • REAL{ . . . } real (built-in) function
  • s(v) reference signal

In FIG. 2 is schematically shown a block circuit diagram of a representation, in keeping with the invented procedure, of the sender 1. The undistorted midamble codes c_a(v) are multiplied by the gain factors g_a and summarized in an adding device 3 and transmitted in parallel. The codes of the midambles c_a(v) are, in general not orthogonal.

The representation of the transmission channel 4, which is in agreement with the invented procedure, as it is presented schematically in FIG. 3, takes into consideration an additive disturbance n(v), which, in adding device 5 overlays the referring signal and biases the measurement signal r(v).

The invented procedure for the detection of the midambles uses a common maximum favorable probability estimation, which employs the following cost function:

$\begin{array}{cc}L\left({\tilde{g}}_a\right)=\sum _{v=0}^{N-1}{r\left(v\right)-\sum _a{\tilde{g}}_a·c_a\left(v\right)}^2,& \left(1\right)\end{array}$

in order to estimate the gain factors of the midambles. In this equation, r(v) is the measurement signal, c_a(v) is the complex, capacity normed, undistorted signal of the a-ten midamble and g_a is the gain factor of the a-ten midamble. Also, {tilde over (g)}_a denotes the trial value of the gain factor g_a.

For the calculation of the partial differentiation of the cost function in accord with the unknown parameters, the following formality is employed: an unknown parameter x is a real number, the constants c and d are complex numbers and a general cost functionL=|c·x+d|²=(c·x+d)·(c·x+d)*=|c|²·x²+c*d·x+c·d*·x+|d|²  (2)employs the square of the amount. Then, the partial derivative can be computed in this manner:

$\begin{array}{cc}\frac{\partial L}{\partial x}=2·{c}^2·x+2·\mathrm{REAL}\left\{c·d^{*}\right\}.& \left(3\right)\end{array}$ Having equation (3), a partial differentiation will yield, in accord with the estimated value ĝ_a of the gain factors of the midambles:

$\begin{array}{cc}\frac{\partial L}{\partial {\hat{g}}_a}=2\sum _{v=0}^{N-1}{c_a\left(v\right)}^2·{\hat{g}}_a+2\sum _{v=0}^{N-1}\mathrm{REAL}\left\{-c_a\left(v\right)·a_3^{*}\left(v\right)\right\}=0& \left(4\right)\end{array}$

with

$\begin{array}{cc}{a}_3\left(v\right)=r\left(v\right)-\sum _{\mu \ne a}{\tilde{g}}_{\mu }·c_{\mu }\left(v\right).& \left(5\right)\end{array}$

The equations (4, 5) can be condensed into a matrix-vector mode, giving:└A_a,μ┘·└ĝ_μ┘=[b_a]  (6)

whereby the coefficients of the lines reduce themselves to:

$\begin{array}{cc}{b}_a=\sum _v\mathrm{REAL}\left\{c_a\left(v\right)·r^{*}\left(v\right)\right\}& \left(7\right)\\ A_{a,\mu }=\sum _v\mathrm{REAL}\left\{c_a\left(v\right)·c_{\mu }^{*}\left(v\right)\right\}& \left(8\right)\end{array}$

If the linear equation is solved, then the sought for optimal estimated value ĝ_a of the gain factors g_a becomes known.

Using then, as a starting point, the estimated value {tilde over (g)}_a of the midambles, the entire capacity of the midambles can be approximated by

$\begin{array}{cc}{\hat{P}}_{\mathrm{Midamble}}=\sum _a{\left({\hat{g}}_a\right)}^2& \left(9\right)\end{array}$

If the capacity of a midamble code oversteps

$\begin{array}{cc}10{\log }_{10}\frac{{\hat{g}}_a^2}{{\hat{P}}_{\mathrm{Midamble}}}>M& \left(10\right)\end{array}$ the above relative, logarithmic capacity threshold M, then the midamble code is classified as an active midamble code.

The probability of detection of error in the procedure presented here is clearly much less, in comparison to a simple correlation procedure.

The greatest expenditure of time and effort in the procedure lies in the computation of the correlation coefficients b_a between the undeformed midamble codes and the measurement signal according to the equation (8). For the solving of the equation system, the coefficients of the inverse matrix A⁻¹ can be computed ahead of time and stored in memory. For this operation, one requires, for the solving of the equation system one needs only the second power of the number N₁ of the midamble codes operations. The numerical complexity of the presented procedure is also, in comparison to the simple correlation procedure, only a slight bit greater, in case the number N₁ of the midamble codes in comparison to the length N of the midamble codes is too small.

While the invention has been described in detail and with reference to specific examples thereof, it will be apparent to one skilled in the art that various changes and modifications can be made therein without departing from the spirit and scope thereof.

Claims

1. A process for recognizing active code sequences in mobile radio systems amongst a plurality of combined code sequences (ca(v)·ga) to minimize erroneous or failed detection from inactive code sequences, wherein the active code sequences exhibit greater gain factors (ga) than do the inactive code sequences, the process comprising: receiving the combined code sequences;forming a cost function (L) dependent upon unknown but estimated values ({tilde over (g)}a, ĝa) of the gain factors (ga) of the combined code sequences (ca(v)·ga);partially differentiating the cost function (L) in accord with the unknown estimated values ({tilde over (g)}a, ĝa) of the gain factors (ga);forming a correlation system from an assumption that all partial differentiations of the cost function (L) are zero and thus a minimum of the cost function (L) exists;determining optimal estimated values ({tilde over (g)}a, ĝa) of the gain factors (ga) by solving equation systems;storing the optimal estimated values ({tilde over (g)}a, ĝa);comparing the saved optimal estimated values ({tilde over (g)}a, ĝa) of the corresponding gain factors (ga) to a specified threshold;classifying any of the received combined code sequences as active code sequences when the saved optimal estimated values ({tilde over (g)}a, ĝa) of the corresponding gain factors (ga) are greater than the specified thresholds; andconveying the classified active code sequences from the combined code sequences for communication based on data areas corresponding to only the classified active code sequences and thus minimizing erroneous or failed detection from inactive code sequences.

2. The process of claim 1, wherein the cost function (L) is linearized by series development before the partial differentiation.

3. The process of claim 2, wherein the cost function L is

4. The process of claim 3, wherein by means of partial differentiation of the linearized cost function L in accord with the unknown estimated values ĝa the gain factor ga the equation system to be solved └Aa,μ┘·[ĝa]=[ba]

5. The process of claim 4, wherein the coefficients

6. The process of claim 1, wherein the estimated values of ĝa of the gain factors ga obtained by the solution of the equation system are squared and the sum squares

7. The process of claim 1, wherein the code sequences (Ca(v)·ga) are identification sequences of a code channel of a CDMA-mobile radio system.

8. A computer program with a programmed code stored on a computer readable medium, wherein the program executes the process of claim 1 when the program is run in a computer.