Channel estimation for time division duplex communication systems

Channel estimation for time division duplex communication systems

These Patent application claims priority from US Provisional Patent Application No. 60/175167, filed on 7 January 2000.

Background of the Invention

the Invention relates generally to wireless communication systems. the in particular invention relates to a channel estimation in a wireless communication system.

<figref idrefs="S18">1</figref> shows an illustration of a wireless communication system <figref>10</figref>, The communication system <figref>10</figref> includes base stations <figref>12₁</figref> to <figref>12₅</figref> on, with user devices (user equipment, UE) <figref>14₁</figref> to <figref>143</figref> communicate. Each base station <figref>12₁</figref> is an operating range assigned, by comparing them to the user equipment <figref>14₁</figref> to <figref>14₃</figref> communicates.

In some communication systems, such as in code division multiple access (CDMA) and time division duplex systems in which code division multiple access techniques (TDD / CDMA) are used, a plurality of communications over the transmitted same frequency spectrum. These communications are typically based on their chip code sequences distinguished. To more efficiently utilize the frequency spectrum, use TDD / CDMA communication systems for communication repeating Frames that are divided into time slots. An in such a System transferred Communication is one or more associated chip codes and have time slots which it based on the Kommunikatioinsbandbreite be assigned to.

Because multiple communications in the same frequency spectrum and transmit simultaneously can be, has to be receiver in such a system between the plurality of communications differ. A method for detecting such signals is a single-user detection methods. In a single-user detection detected a receiver only the communications from a desired recipient under Use of the desired Stations assigned code and treats signals of other transmitters as Interference or disturbance. Another method is referred to as joint detection. At a joint detection, multiple communications are simultaneously detected.

at the use of these detection techniques, it is desirable the radio channel via transmitted to each communication will appreciate. In a typical TDD system, the channel estimation using Midamble sequences executed in communication bursts.

On typical communication burst <figref>16</figref> has a midamble <figref>20</figref>. a guard period <figref>18</figref> and two data bursts <figref>22</figref>. <figref>24</figref> on, as in <figref idrefs="S18">2</figref> is shown. The midamble<figref>20</figref> separated the two data bursts <figref>22</figref>. <figref>24</figref>And the protection period <figref>18</figref> separated the communication bursts <figref>16</figref>To a time difference in the arrival times of the transmitted from different transmitters bursts <figref>16</figref> permit. The two data bursts<figref>22</figref>. <figref>24</figref> contain the data of the communication bursts. The midamble<figref>20</figref> contains a for the channel estimation provided training sequence.

After this a receiver a communication burst <figref>16</figref> having received, he appreciates the Channel using the received midamble sequence. When a receiver several bursts <figref>16</figref> receives in a time slot, he estimates typically the channel for each burst <figref>16</figref>, A method for such channel estimation for over several Channels transmitted Kom munikationsbursts is a channel estimation method according to Steiner. the channel estimation by Steiner is typically used for uplink communications from several user units <figref>14₁</figref> to <figref>14₃</figref> used, wherein the channel estimator more channels estimate have to be.

In an article by Steiner and Jung titled "Optimum and sub-optimum Channel Estimation for the uplink of CDMA Mobile Radio Systems with Joint and Detection "is a channel estimation method described. According to a The method is used, a single cyclic correlator. Among Using the known transmitted Midamble sequences constructed a matrix M. The received Midamble vector e is multiplied by the first column of the matrix M. The multiplication is the cyclic correlator over P values and executed by (2P-1) times the displacement values. P is a period of the midamble codes.

&lt;?page 3?&gt;

In some cases be several bursts <figref>16</figref> transmitted over the same radio channel. According to a system is used a service in which a high data rate used is, for example 2 Mbit / s. In such a system, a transmitter transmit multiple bursts in a single time slot. In this case, the Channel estimation method by Steiner by averaging the channel responses of all bursts <figref>16</figref> be applied. This method is, however, highly complex. Therefore, would it is desirable to if alternative channel estimation techniques to disposal stood.

Short description the invention

On single transmitter transmits K communication bursts in a shared spectrum in a time slot of a time division duplex communication system. The system associated with N midamble sequences. Each burst is a midamble sequence assigned. A receiver receives one of the transferred Midamble sequences of the K communication bursts corresponding vector. A matrix having N identical right circulant matrix blocks is based in part on the known N midamble sequences constructed. The radio channel between the transmitter and receiver is based partly on one of the N blocks and the received vector estimated.

Brief Description of Drawings

<figref idrefs="S18">1</figref> shows a wireless communication system;

<figref idrefs="S18">2</figref> shows an illustration of a communication burst;

<figref idrefs="S19">3</figref> shows a simplified Multi Burst transmitters and receivers; and

<figref idrefs="S20">4</figref> shows a flow chart of multiburst channel estimation method.

Detailed Description of Preferred embodiments

<figref idrefs="S19">3</figref> shows a simplified multicode transmitter <figref>26</figref> and a multi-code receiver <figref>28</figref> in a TDD / CDMA communication system. In a preferred application, such as in a 2 Mbit / s downlink service, the recipient <figref>28</figref> a user device <figref>14₁</figref>, and the transmitter <figref>26</figref> is a base station <figref>12₁</figref>, although recipient <figref>28</figref> and the station for other applications are usable.

Of the transmitter <figref>26</figref> transmits data over a radio channel <figref>30</figref>, The data are transmitted in K communication bursts. data generators <figref>32₁</figref> to <figref>32_K</figref> in the transmitter <figref>26</figref> produce at the recipient <figref>28</figref> to be transmitted Data. Modulation / spreading and Trainigssequenzeinfügungsvorrichtungen<figref>34₁</figref> to <figref>34_K</figref> spread the data and multiplexing the spread reference data time with a midamble training sequence in the appropriate assigned time slot and codes for spreading the data to the K communication bursts to create. Typical values ​​for K for Downlink bursts transmitted base station <figref>12₁</figref> range 1 to 16 The communication bursts are by a combiner <figref>48</figref> combined and by a modulator <figref>36</figref> a radio frequency (Radio Frequency, RF) modulated. An antenna<figref>38</figref> radiates RF signal over a radio channel <figref>30</figref> from an antenna <figref>40</figref> Recipient <figref>28</figref>, The for the modulation type communication used, any be in the art known modulation type, such as a binary phase shift (BPSK) or quadrature phase shift keying (QPSK).

the antenna <figref>40</figref> Recipient <figref>28</figref> receives various Radio frequency signals. The received signals are through a demodulator <figref>42</figref> demodulated to produce a baseband signal. The baseband signal is, for example, by a channel estimator <figref>44</figref> and a data acquisition device <figref>46</figref> in the time slot and with appropriate codes processed which the transmitted communication bursts assigned. The data acquisition device<figref>46</figref> can be a multiuser detector or a single user detector. The channel estimator<figref>44</figref> used the midamble training sequence component in the baseband signal to channel information provide, for example, channel impulse responses. The channel information is represented by the data acquisition device <figref>46</figref> to estimate the Transmission data of the received communication bursts as hard symbols used.

To the Explain an implementation of a multi-burst channel estimation is used the following midamble type, although the Multi Burst channel estimation to other midamble types is applicable. The K midamble codes<img img-content="tx" img-format="tif" he="5" wi="7" file="00050001.tif" />. where k = 1 ... K, are referred to as time-shifted versions of a periodic single Midamble codes <img img-content="tx" img-format="tif" he="3" wi="5" file="00060001.tif" /> of Periods (P) chips derived. The length of each midamble code is L_m = P + W - &lt;?page 4?&gt;1, where W is the length of the is user channel impulse response. Typical values ​​for L_m are 256 and 512 chips. W denotes the length of the user channel impulse response. Although the following discussion is based on that each Burst another midamble code which can some midambles have the same code. Therefore, based the Analysis on N midamble codes N &lt;K. also the system may have a maximum number of permissible N midamble codes. Recipient <figref>28</figref> in such a system, the channel with respect to the maximum number N codes estimate even if less than N codes transmitted will.

the elements of <img img-content="tx" img-format="tif" he="3" wi="5" file="00060001.tif" /> to take Values ​​from the set {1, -1} integers on. The sequence<img img-content="tx" img-format="tif" he="3" wi="5" file="00060001.tif" /> is first in a complex sequence <img img-content="mf" img-format="tif" he="5" wi="26" file="00060002.tif" /> converted, where i = 1 ... P is. The values<img img-content="tx" img-format="tif" he="5" wi="7" file="00050001.tif" /> will obtained by K sub-sequences of length L_m of a Sequence of length 2P be taken that by linking of two periods <img img-content="tx" img-format="tif" he="4" wi="5" file="00060003.tif" /> educated becomes. The ith element of<img img-content="tx" img-format="tif" he="5" wi="7" file="00050001.tif" /> stands With <img img-content="tx" img-format="tif" he="4" wi="5" file="00060003.tif" /> according to equation 1 in relationship.

<img img-content="mf" img-format="tif" he="14" wi="115" file="00060004.tif" /> equation 1

Therefore will shift the starting point of <img img-content="tx" img-format="tif" he="5" wi="7" file="00050001.tif" />. k = 1 ... K to W chips to the right when k from 1 to K increases.

the combined received midamble sequences are a superposition of K folds. The k th convolution represents the convolution of<img img-content="tx" img-format="tif" he="5" wi="7" file="00050001.tif" /> With <img img-content="tx" img-format="tif" he="4" wi="5" file="00060005.tif" /> represent. <img img-content="tx" img-format="tif" he="4" wi="5" file="00060005.tif" /> is the channel response of the k th user. The preceding data field falsified burst the first (W-1) chips of the received midamble. Therefore, for channel estimation only the last P of L_m Chips.

below is a multi-burst channel estimation in conjunction with the flow diagram of <figref idrefs="S20">4</figref> explained. Around the individual channel responses <img img-content="tx" img-format="tif" he="4" wi="5" file="00060005.tif" /> to obtain, Equation 2 is used. <img img-content="mf" img-format="tif" he="31" wi="156" file="00070001.tif" /> equation 2 where r_W... r_LM the received combined chips of the midamble sequences designate. The m values ​​are the elements of <img img-content="tx" img-format="tif" he="3" wi="5" file="00060001.tif" />,

equation 2 can also in shorter Form can be rewritten as an equation. 3

<img img-content="mf" img-format="tif" he="10" wi="131" file="00070002.tif" />

Each expression M^(K) denotes a KW × W matrix. r - designated the received midamble chip responses. If all bursts over the transmit the same channel can be <img img-content="tx" img-format="tif" he="4" wi="5" file="00070003.tif" /> ... <img img-content="tx" img-format="tif" he="4" wi="5" file="00060005.tif" /> replaced - by h be, as shown in Equation 4 (50).

<img img-content="mf" img-format="tif" he="11" wi="126" file="00070004.tif" />

&lt;?page 5?&gt;

G is according to equation 5 defined. <st32:df xmlns:st32="http://lighthouseip.com/">G = [M<st32:sup>(1)</st32:sup>, ..., M<st32:sup>(K)</st32:sup>, ..., M<st32:sup>(K)</st32:sup>] equation 5</st32:df>

Thereby G is a KW × KW matrix. Since G is a right circulant matrix, Equation 4 can be implemented using be rewritten by K identical right circulant matrix blocks B, as shown by equation 6 is (52).

<img img-content="mf" img-format="tif" he="24" wi="127" file="00080001.tif" />

B is a right cyclic W × W matrix. The number of B-blocks is K. Using Equation 6, Equation 4 rewritten in Equation 7 will. <st32:df xmlns:st32="http://lighthouseip.com/">That is - = r - Equation 7</st32:df>

equation 7 describes an over-determined System with dimensions KW × W. A method for dissolution DC Tung 7 is a least squares method (54). the solution according to the method the least squares of Equation 7 is given by Equation eighth <img img-content="mf" img-format="tif" he="6" wi="126" file="00080002.tif" />D^H is a Hermitian of D.

By Applying Equation 6 to Equation 8 is obtained equation. 9

<img img-content="mf" img-format="tif" he="8" wi="126" file="00080003.tif" />

Of the received vector r - of dimension KW can be decomposed according to equation 10th

<img img-content="mf" img-format="tif" he="24" wi="129" file="00080004.tif" />

the Dimension of r -_k is W. Substituting Equations 9 and 10 into Equation 8, the solution according to the method of least squares for the channel coefficients according to equation 11 receive. <img img-content="mf" img-format="tif" he="11" wi="129" file="00090001.tif" />r -_k represents the average of the segments of r - represents Since B is a square matrix, Equation 11 in Equation 12 rewritten.

<img img-content="mf" img-format="tif" he="6" wi="129" file="00090002.tif" />

Because B is a right circulant matrix and the inverse of a right cyclic matrix is ​​also right cyclically, the channel estimator by a single cyclic correlator of dimension <figref>57</figref> or be implemented by a discrete Fourier transform (DFT) solution.

On W point DFT method is implemented as follows. Because B right is cyclic, equation may be used thirteenth <st32:df xmlns:st32="http://lighthouseip.com/">B = D<st32:frac>-1<st32:over>W</st32:over></st32:frac> Λ<st32:sub>C</st32:sub> · D<st32:sub>W</st32:sub> equation 13</st32:df>

&lt;?page 6?&gt;

D_W designated the W point DFT matrix according to equation 14th

<img img-content="mf" img-format="tif" he="38" wi="136" file="00090003.tif" />

Λ_C designated is a diagonal matrix whose main diagonal of the discrete Fourier transform (DFT) of the first column of B, as shown in Equation 15 is. <st32:df xmlns:st32="http://lighthouseip.com/">Λ<st32:sub>C</st32:sub> = diag (D<st32:sub>W</st32:sub>(B (:, 1))) Equation 15</st32:df><img img-content="mf" img-format="tif" he="6" wi="18" file="00100001.tif" /> Therefore D_W of the DFT operator so that D_W<u>x</u> the W-point DFT of vector <u>x</u> represents. By Substituting equation 13 in equation 12, and Using <img img-content="mf" img-format="tif" he="9" wi="17" file="00100002.tif" /> Equation 16 is obtained. <img img-content="mf" img-format="tif" he="8" wi="129" file="00100003.tif" />D * / W denotes the element-wise complex conjugate of D_W,

alternative , an equivalent Shape can be obtained, the h - instead regarding Λ_C respect AR expresses. Λ_R is is a diagonal matrix whose main diagonal of the DFT of the first row of B, as shown in equation 17th <st32:df xmlns:st32="http://lighthouseip.com/">Λ<st32:sub>R</st32:sub> = diag (D<st32:sub>W</st32:sub>(B (:, 1))) Equation 17</st32:df>

Because the transpose of B, B^TAlso right is cyclical, and its first column is the first row of B, can B^T be represented by Equation 18th <st32:df xmlns:st32="http://lighthouseip.com/">B<st32:sup>T</st32:sup> = D<st32:frac>-1<st32:over>W</st32:over></st32:frac> · Λ<st32:sub>R</st32:sub> · D<st32:sub>W</st32:sub> equation 18</st32:df>

Among Using Equation 18 and under the condition that DT / W = D_W, Λ T / R = Λ_R is, and that invertible for each Matrix A, (A^T)^-1= (A^-1)^T applies, B can be represented by Equation 19 can be, <st32:df xmlns:st32="http://lighthouseip.com/">B = D<st32:sub>W</st32:sub> · Λ<st32:sub>R</st32:sub> · D<st32:frac>-1<st32:over>W</st32:over></st32:frac> equation 19</st32:df>

By Substituting equation 19 in equation 12, and under the condition, that <img img-content="mf" img-format="tif" he="9" wi="16" file="00100004.tif" /> is, Equation 20 is obtained.

<img img-content="mf" img-format="tif" he="8" wi="129" file="00100005.tif" />

the Equations 16 or 20 can be used to determine the equation for h - to solve. Since all DFTs the same length W have, the complexity is for the solution the equations substantially reduced.

&lt;?page 7?&gt;

On A method of using a single cyclic correlator following. Because B^-1 the inverse form of a right is cyclic matrix, they can be represented by Equation 21 will.

<img img-content="mf" img-format="tif" he="30" wi="129" file="00110001.tif" />

the first row of the matrix T is the inverse DFT of the main diagonal equal of Λ -1 / R. Therefore, completely determines the matrix T by Λ -1 / R.

the Elements of the channel response h - be an inner or scalar successive rows of T with the average of segments the length W of the received vector r - received. The successive lines T are cyclically right shifted versions of the previous Row. Using registers to generate the inner product holds the first register the averaged segments of r -, and the second register is a shift register that holds the first row of the matrix T. The second register is cycled at a predetermined clock rate. At each clock cycle is the dot product of the two in Registers stored vectors, a new element of h - determined. It is advantageous, rather, the first row of the matrix T as the received Midambles to push. This is for the midamble no additional Memory required. The midamble still remain in the receive buffer, that holds the entire burst. Because the Korrelatorlänge only amounts to W, is obtained in the channel estimation is a significant reduction in complexity.