WO2007115434A1

WO2007115434A1 - Strong interference canceling method for neighboring cells using same frequency in cdma system

Info

Publication number: WO2007115434A1
Application number: PCT/CN2006/000652
Authority: WO
Inventors: Peng Geng; Hai Jiang; Ping Li; Zhihong Li
Original assignee: Zte Corporation
Priority date: 2006-04-12
Filing date: 2006-04-12
Publication date: 2007-10-18
Also published as: CN101395812B; CN101395812A

Abstract

A strong interference canceling method for neighboring cells using the same frequency in a CDMA system comprises the following steps: (a) the channel estimation on all users in neighboring cells using the same frequency is processed, and one or more strong interference user(s) is selected, whose power is strongest among all channel impulse response in neighboring cells; (b) the interference of the strong interference user in neighboring cells using the same frequency, which is obtained in step (a), is subtracted from training sequence part in received signals, then the channel estimations for all users in local cell are processed; (c) the joint detection for multi users is obtained for the data part in received signals, using the channel impulse response estimations on strong users in neighboring cells using the same frequency which is obtained from step (a) and the channel impulse response estimations on all users in local cell which is obtained from step (b), then the symbol information of all users in local cell is demodulated. Present invention could cancel the interference which local user signal detection is affected from the strong interference user in neighboring cells using the same frequency, and enhance the performance of signal detection for users in local cell.

Description

Method for eliminating strong interference in CDMA system co-frequency neighboring cell

Technical field

The present invention relates to a method for interference cancellation in a CDMA mobile communication system, and more particularly to a method for strong interference cancellation of a co-frequency neighboring cell in a third-generation TD-SCDMA mobile communication system based on time division duplexing.

Background technique

As we all know, the CDMA system is a self-interference system. In addition to the traditional frequency division (if equipped with multiple frequency points) and time division (such as time slots in TD-SCDMA systems), the method of multiple access is introduced into the concept of code division, although all CDMA systems use The channelization codes are all orthogonal codes (such as OVSF codes in TD-SCDMA systems), but the multipath introduced in the wireless weakened channel still causes interference to a large extent; at the same time, interference from neighboring co-frequency cells cannot pass positive The characteristics of the cross code are eliminated. Many existing interference cancellation technologies, such as multi-user detection (MUD), serial or parallel interference cancellation (SIC/PIC), are aimed at eliminating multiple-site interference in a cell, but cannot solve the same-frequency neighboring cell band. The interference.

Summary of the invention

The technical problem to be solved by the present invention is to provide a method for strong interference cancellation in a neighboring cell, so as to eliminate the interference of the strong neighboring cell in the same-frequency code division multiple access system to the user signal detection of the local cell, and enhance the signal detection performance of the user in the cell.

To solve the above problem, the present invention provides a strong interference cancellation method for a co-frequency neighboring cell in a code division multiple access system, which includes the following steps: (a) performing channel estimation on all users of the same-frequency neighboring cell, and in all neighboring cell users One or more strong interfering users with the strongest power are selected in the channel impulse response;

(b) subtracting the interference of the interfering neighbor cell strongly interfered with the user in the training sequence portion of the received signal, and then performing channel estimation on all users of the cell; (c) using the channel impulse response of the interfering neighbor cell strongly interfered by the user in step (a) and the channel impulse response of all users in the cell obtained in step (b) to perform multi-user joint detection on the data part of the received signal, Call up the symbol information of all users in the community.

Further, the method for performing channel estimation on the intra-frequency neighboring cell in the step (a) includes the following steps: (al) using the training sequence part of the received signal to perform initial channel estimation for all users in the cell; (a2) according to the step (al) Obtaining a user channel impulse response, retaining one or more taps of the strongest signal power of the cell; (a3) subtracting the interference of the strong signal of the cell obtained in step (a2) in the training sequence portion of the received signal, and then according to the Channel estimation is performed for all users of the same frequency neighboring cell.

Further, step (a2) retains a sequence of initial channel estimation for all users of the cell. The K-tap with the largest medium power, where K is the number of users in the current time slot of the cell, or exceeds the set tap power threshold ϊ3⁄4 ₎ , and the remaining taps are cleared, and the processed channel impulse response sequence is obtained. . .

Further, in step (a3), according to = e, „- Μ. · subtract the interference of the K taps of the strongest signal power of the local cell obtained in step (a2) in the training sequence portion of the received signal, and according to fi, = M^S, „Cell estimation for all users of the same-frequency neighbor cell, where the training sequence midamble part receives data in the middle of the time slot, M. The right loop matrix formed by the basic midamble code of the cell, i = \, 2...K _nc , ,. The number of the largest neighboring cell, Μ, is the right circular matrix formed by the basic midamble of the i-th neighboring cell.

Further, the method for one or more strong interfering users with the strongest power output in all neighbor cell user channel impulse responses in step (a) further includes: (a4) a channel impulse response sequence for each of the neighboring cells ^ Only retain the N taps whose power is the largest, and the rest are cleared, where N ranges from 1 to the number of activated users in the neighbor cell; (a5) selects and retains the strongest K _u among all ^ non-zero taps One, _c taps, the above K _u→c taps correspond to the total physical code channel resources occupied by the user should not exceed the physical code channel resources of the current time slot of the current cell; (a6) for this. The non-zero taps are processed after the neighboring cell threshold decision, and the taps whose tap power exceeds the threshold I3⁄4r. are retained, and finally hi ... K _{m is obtained} .

Further, the threshold in step (a6) is calculated as follows: r/^ = .fif. , the value can be between 1/4 and 3/4. Further, step (b) subtracts the strong signal of the same-frequency neighboring cell obtained by the step (a6) according to -e, „ - ]M, h, in the training sequence portion of the received signal, and according to £.=Μ^ ·έ,,,. An enhanced local cell channel estimation is performed.

Further, the step (b) further comprises: performing a threshold decision post-processing on the channel response of the cell, retaining the tap that activates the power in the user window exceeding the threshold, and clearing the remaining taps to obtain a channel impulse response after the cell processing. .

Further, step (c) comprises the following steps:

(cl) according to the equivalent virtual code channel, all activated neighboring cells strongly interfere with the user's channel impulse response convolution corresponding cell's scrambling code spreading code to generate respective b vectors, and after processing with the local cell in order The b vectors generated by the scrambling code spreading codes corresponding to the channel impulse response convolutions together form a composite V matrix as shown in the following formula;

v = b(» b ₀₂ .. b 0, K _vm ncl b where b. denotes the b-vector corresponding to the i-th equivalent virtual code channel of the cell, which is the total number of equivalent virtual code channels of the cell, b„ _c>i denotes the b-vector corresponding to the i-th equivalent virtual code channel in all κ _νηι ―, _c neighbor cells;

(c2) constituting the composite V matrix of the composite system matrix A of all users participating in the joint detection as shown in the following formula in the form of the original block Toepliz;

V

(c3) Demodulation of user data in accordance with a multi-user joint detection algorithm.

Further, the multi-user joint detection of the step (c3) is based on the MMSE-BLE algorithm, and the post-tuning symbol is as follows:

d = (A^R^A + R- ¹ ) — ¹ A^R^e^

Where e _d is a slot data part receiving sequence, which is a noise correlation matrix, R _rf is a signal correlation matrix; and demodulation symbol 3 contains a symbol of a neighboring cell strong interfering user, its position and the neighboring cell user b in the system matrix A The vector positions correspond one-to-one. Compared with the prior art, the neighboring cell strong interference cancellation method of the present invention greatly eliminates the interference of the neighboring cell strong interference user to the user of the cell by allowing the neighboring cell to strongly interfere with the user to participate in the joint signal measurement of the local cell, thereby greatly enhancing the local cell. User signal detection performance, which in turn improves the performance of the entire CDMA receiver system.

BRIEF abstract

Figure 1 is a schematic diagram of the conventional slot structure of the physical layer of the TD-SCDMA system.

2 is a flow chart of a method for eliminating strong interference in a co-frequency neighboring cell according to an embodiment of the present invention. BEST MODE FOR CARRYING OUT THE INVENTION

The present invention is applicable to a base station receiver and a terminal receiver of a TD-SCDMA mobile communication system, and is also applicable to any CDMA mobile communication system.

Hereinafter, a specific embodiment of the present invention will be described with reference to the accompanying drawings, taking a TD-SCDMA base station receiver as an example. This scheme can be extended to base station receivers and terminal receivers for any other CDMA mobile communication system.

1 is a schematic diagram of a conventional time slot structure of a physical layer of a TD-SCDMA system. A time slot consists of 864 chips, consisting of 2 data parts and an intermediate training sequence of 144 chips.

The specific operation steps of the baseband processing of the TD-SCDMA base station of the present invention are as follows:

Step 10: The receiver uses the training sequence part of the received signal to perform initial channel estimation for all users in the cell according to formula (1);

h ₀ = Mo ¹ -e _m ( 1 )

Wherein e is the digital baseband sequence of the midamble portion of the training sequence in the middle of FIG. 1 (ie, the digital baseband sequence of the midamble received signal obtained by down-converting and digitally sampling the antenna interface radio frequency data, all of the received signal sequences described below are digital baseband sequences ), M. The right loop matrix formed by the basic midamble code of the cell. Step 20: Retain the sequence. In the specific implementation, the K taps with the highest power can be simply taken , = , where the number of users in the current time slot of the cell is activated, and the remaining taps are cleared to obtain the processed channel impulse response sequence ϋ. And calculate the post-processing threshold of the neighboring cell channel estimation

Thr _nc = ~-h ^H , . , ii denotes fi. The conjugate transpose, the superscript H represents Hermitian; for the present invention, the reason why the Ku taps are retained is that the cell should have at least one main path energy; in fact, we can also set a power threshold. Keep £. The threshold is exceeded: 3⁄4 of the taps, and the remaining taps are cleared.

Because the accuracy of the channel estimation of the neighboring cell is relatively poor, only the channel impulse response of the strong interference multipath can be accurately estimated. The neighboring cell channel estimation post-processing threshold set here is used to select "strong interference". The value can be between 1/4 and 3/4, and the simulation shows that the effect is better when it is equal to 1/2.

Step 30: Perform a midamble partial interference cancellation process according to formula (2), that is, subtract the interference of the strong signal of the local cell in the midamble original receiving sequence, and then perform channel estimation on all the same-frequency neighboring cells according to formula (3).

e ~ _m = e _m -M ₀ - (2)

(3)

Where ί = 1,2.. „., K _nc is the maximum number of neighbor cells, M _; is the right circular matrix formed by the basic midamble of the i-th neighboring cell;

The simulation shows that if the interference cancellation is performed on the midmable part of the received sequence used in the channel estimation of the neighboring cell, it is beneficial to improve the channel estimation effect of the neighboring cell.

Step 40: The channel impulse response sequence of each of the neighboring cells only retains the N taps whose power is the largest, and the remaining clears, where N ranges from 1 to the number of activated users in the neighboring cell, which can be simple in practical applications. Take N=l; then at all. Among the non-zero taps, the most powerful _„. taps are selected, and the total physical code channel resources (the number of RUs) occupied by the above-mentioned taps corresponding to the user should not exceed the physical code channel resources of the current time slot of the current cell ( Taking a 12.2k voice user with a spreading factor of 8 as an example, then = 8 - , ); Finally, the non-zero tap is subjected to the neighbor cell threshold decision post-processing, and the reserved tap power exceeds the threshold J3⁄4r. The taps, other taps are cleared, and finally get h, h, ...; The strong interference user is discovered through detection. In fact, from a statistical point of view, the probability that a neighboring cell forms a strong interference number for the local cell at a certain moment is 1 is the largest, so the channel estimate for each neighbor cell is Only a maximum of one of the multiple taps after the threshold is processed.

Step 50: The above sequence is subtracted from the original received sequence according to formula (4), and all neighboring cell midamble receiving sequences used for channel estimation interference cancellation in the local cell are reconstructed, and are performed according to formula (5): Cell channel estimation, performing traditional threshold post-processing, retaining the tap of the active user window power exceeding the threshold 73⁄4 (the channel estimation post-processing decision threshold, the size is proportional to the noise tap power, and the engineering can take 1.5 to 3 times the noise tap power) The remaining taps are cleared to obtain the channel impulse response & processed by the cell. ; ^o = e _m -∑M,. .h _; (4)

S ₀ = M^ .S„,, ₀ (5)

Step 60: According to the equivalent virtual code channel (equivalent SF=16), all activated neighboring cells strongly interfere with the user (the number of equivalent virtual code channels is the channel impulse response convolution corresponding cell scrambling code expansion) The frequency code words generate respective b vectors, and sequentially form the b vectors generated by the scrambling coded codewords corresponding to the processed channel impulse response convolutions obtained by the cell in step 50 to form a composite V as shown in equation (6). a matrix, and then the composite V matrix in the form of the original block Toepliz constitutes a composite system matrix A of all users participating in joint detection (including neighboring cells strongly interfering users) as shown in equation (7);

Wherein, the ratio of the b-vector corresponding to the i-th equivalent virtual code channel of the cell is the total number of equivalent virtual code channels of the cell, b„., indicating the i-th (in all) equivalent virtual code of the neighboring cell. The b vector corresponding to the track;

V

V!

A = (7)

i V

Step 70: Perform user data demodulation according to a multi-user joint detection algorithm, as shown in equation (8). Demodulated symbols based on the MMSE-BLE algorithm:

(A 'A + R ¹ ) — R;; (8)

Where e _rf is the data part receiving sequence shown in Figure 1, R, is the noise correlation matrix, ^ is the signal correlation matrix; and demodulation symbol 3 contains the symbol of the neighboring cell strong interfering user, its position and the neighbor of the system matrix A The b-vector positions of the cell users correspond one-to-one.

In the method of the present invention, it is not necessary to demodulate the neighboring cell strong interference user symbol, but the neighboring cell strong interference user must be included in the system equation of multi-user detection, that is, the neighboring cell user symbol must be utilized in the demodulation process of the user symbol. The cell strongly interferes with the user's interference cancellation process, which is implemented in equation (8).

Industrial applicability

The invention can be used to introduce the strong interfering user information of the neighboring cell into the channel estimation and the joint detection of the local cell, which greatly suppresses the interference of the strong interference of the neighboring cell to the demodulation of the user signal of the local cell, and improves the accuracy of the channel estimation of the cell and the joint. The effect of signal detection.

Claims

A method for eliminating interference in a co-frequency neighboring cell of a code division multiple access system, comprising the following steps -

(a) Perform channel estimation for all users of the same-frequency neighboring cell, and pick out one or more strong interfering users with the strongest power among all neighbor cell user channel impulse responses;

(b) subtracting the interference of the interfering neighbor cell strongly interfered with the user in the training sequence portion of the received signal in step (a), and then performing channel estimation on the user owned by the cell;

(c) using the channel impulse response of the interfering neighbor cell strongly interfered by the user in step (a) and the channel impulse response of all users in the cell obtained in step (b) to perform multi-user joint detection on the data part of the received signal, Call up the symbol information of all users in the community.

2. The method according to claim 1, wherein: in step (a), the same frequency neighboring cell is entered.

The method for line channel estimation includes the following steps: (al) using the training sequence portion of the received signal to perform initial channel estimation for all users of the cell; (a2) retaining the signal power of the cell according to the user channel impulse response obtained in step (al); Strong one or more taps; (a3) subtracting the interference of the strong signal of the local cell obtained in step (a2) in the training sequence portion of the received signal, and then performing channel estimation on all users of the same-frequency neighboring cell accordingly. .

3. The method of claim 2 wherein: step (a2) retains a sequence of initial channel estimates for all users of the cell. The K-tap with the highest power in the medium, where K is the number of users activated in the current time slot of the cell, or exceeds the set number of tap power thresholds, and the remaining taps are cleared to obtain the processed channel impulse response sequence &. .

4. The method according to claim 3, wherein: in step (a3), according to = e _m -M. . . . The interference of the K taps of the strongest signal power of the local cell obtained in the step (a2) is subtracted from the training sequence portion of the received signal, and according to ^^^! ^ Perform channel estimation for all users of the same-frequency neighboring cell, where e _m is the training sequence midamble part of the time slot receiving data, M. The right loop matrix formed by the basic midamble code of the cell, i = l, 2~K _nc , ,. The number of the largest neighboring cell, M _; is the right circular matrix formed by the basic midamble of the i-th neighboring cell.

5. The method according to claim 4, wherein: in step (a), only one strong interfering user with the strongest power is selected in the channel impulse response of all neighboring cells, and the method further comprises: (a4) The channel impulse response sequence fi of each neighboring cell, only retains the N taps whose power is the largest, I clear zero, where N ranges from 1 to the number of active users in the neighbor cell; (a5) selects and retains the strongest power among all non-zero taps. Taps, above. The total physical code channel resources occupied by the taps corresponding to the user shall not exceed the physical code channel resources of the current time slot of the current cell; (a6) The K _U —,, _C non-zero taps are subjected to the neighbor cell threshold decision processing, The tap whose power exceeds the threshold Th' _nc is retained, and finally, h ₂ ... _{∞ is obtained} .

6. The method of claim 5 wherein: the threshold in step (a6) is calculated such that: Thr „ _c = K ii ₀ , "the value is between 1/4 and 3/4.

7. The method according to claim 5 or 6, wherein: step (b) is obtained by subtracting step (a6) from the training sequence portion of the received signal according to o = e„, - M, . co-frequency adjacent cells strong signals' interference, according to £. .s _m, _Q for the present cell enhanced channel estimate.

8. The method according to claim 7, wherein: step (b) further comprises: performing a threshold decision post-processing on the channel response of the local cell, and retaining a tap that activates the user window to exceed the channel estimation and processes the decision threshold of 23⁄4^, The remaining taps are cleared to obtain the channel impulse response after processing in the cell.

9. The method of claim 8 wherein: step (c) comprises the steps of:

V = L ^b 01 ^b 02 ··· ^b . , K _v „, ^b »cl c2 ··· ^h nc,K _w „_„ _c _

Where b. _,; B represents an i-th vector of the present cell equivalent code channel corresponding to the virtual ", the virtual present total equivalent number of code channel cell, denotes the i th code channel corresponding to the virtual equivalent in all neighbor cells b vector;

(c2) forming the composite V matrix of the composite system matrix A of all users participating in the joint detection as shown in the following formula in the form of the original block Toepliz; "Ϋ1

A - :...

(c3) User data demodulation according to the multi-user joint detection algorithm.

10. The method according to claim 9, wherein: the multi-user joint detection of the step (c3) is based on the MMSE-BLE algorithm, and the demodulated symbol is as follows:

d = (A ^ff R ¹ A + R- ¹ )" ^, A^R ¹ e _ii

Where e _rf is the slot data part receiving sequence, R„ is the noise correlation matrix, R _rf is the signal correlation matrix; and demodulation symbol 3 contains the symbol of the neighboring cell strong interfering user, its position and the neighbor cell user in system matrix A The b vector positions correspond one-to-one.