KR100958917B1 - A fast index search algorithm for codebook-based beamforming scheme - Google Patents

Abstract

In a MIMO system, a codebook consisting of a vector index of 3 bits or 6 bits is arbitrarily used for a plurality of randomly quantized channel state information to provide a complete channel state information to a transmitter while using less bandwidth. The present invention calculates both the inner product of the weight vector of each vector index and the weight vector of channel state information, and selects the vector index having the maximum value. By calculating the inner product size by selecting only the vector indices of a specific range using the angle of the weighted vector calculated from the information, a new fast index search method with less computational differences and less performance difference is achieved than the conventional vector index search method. Suggest.

MIMO, Codebook, Index, Feedback, Precoding

Description

A fast index search algorithm for codebook-based beamforming scheme

The present invention relates to a beamforming system using a codebook. In particular, a fast index search method can be used in a feedback process of channel state information using a codebook in order to apply a beamforming technique in a MIMO (Multi Input Multi Output) system. It is about how to make one.

In general, systems using codebooks in MIMO systems include WiMAX, WiBro, and LTE techniques. WiMAX stands for World Interoperability for Microwave Access. It is a standard for mobile Internet, originally developed by Intel Corporation in the United States to develop and activate IEEE 802.16 technology standard. Recently, WiBro (Wireless Broadband Internet), which ensures the mobility of WiMAX, has been approved as the IEEE 802.16e international standard. Long Term Evolution (LTE), a fourth-generation mobile communication technology, is being recently researched.

In the MIMO system, frequency division duplexing (FDD) is used to classify communication paths using frequency as a duplex method, and time division duplexing (TDD) is used to classify communication paths using time. In case of the time division method, the feedback is not necessary because it only distinguishes transmission / reception in time and uses channels of the same frequency.However, in the frequency division method, the frequency band used for transmission / reception is different (ie, Because different channels are used, a feedback process for channel state information from a receiver to a transmitter is necessary.

In the MIMO system, a plurality of randomly quantized channel state information provides a codebook configured with a vector index of 3 bits or 6 bits to perform this feedback process while occupying less uplink bandwidth. Codebooks provide 8 or 64 vector indices, respectively. A method of searching for the most efficient vector index among multiple vector indices is to calculate the inner product of the weight vector of each vector index and the weight vector of channel state information. The vector index with the maximum value is selected.

However, as the MIMO system is actively researched, a new index search method is required to reduce the computation amount or improve the performance more than the conventional index search method.

It is an object of the present invention to provide a method for enabling a fast index search method in a feedback process of channel state information using a codebook to apply a beamforming technique in a MIMO system.
Another object of the present invention is to provide a dot product magnitude value using a weight vector calculated from channel state information having a number of streams 1 and a weight vector of vector indices of a codebook in a feedback process of a MIMO system using a frequency division scheme (FDD). In this paper, we provide a new fast index search technique that reduces the computational complexity compared to the conventional method with little performance difference.

It is still another object of the present invention to provide a fast index search technique in a beamforming system using a codebook that can search a vector index satisfying a specific range by using angles of a weight vector of channel state information and a weight vector of a codebook. Is in.
The objects of the present invention are not limited to the above-mentioned objects, and other objects which are not mentioned will be clearly understood by those skilled in the art from the following description.

In order to achieve the above object, a new fast index search scheme according to an embodiment of the present invention compares an angle and a weight vector of vector indices of a codebook using a specific angle of a weight vector calculated from channel state information. The size of the vector indices and the dot product belonging to is obtained, and the vector index having the maximum value is selected to perform the feedback process.

According to the new fast index search method of the present invention, the weight vector of the vector indices of the codebook is compared with the angle by using a specific angle of the weight vector calculated from the channel state information, and the vector indexes and dot products belonging to the specific range are compared. Considering the size of, the number of comparison vector indices is reduced by half, which shows little difference in performance.

Specific details of other embodiments are included in the detailed description and the drawings.

Advantages and features of the present invention and methods for achieving them will be apparent with reference to the embodiments described below in detail with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but may be embodied in various different forms, and the present embodiments merely make the disclosure of the present invention complete, and are common in the art to which the present invention pertains. It is provided to inform those skilled in the art to the fullest extent of the invention, the invention being defined only by the scope of the claims. Like reference numerals refer to like elements throughout.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a conceptual diagram illustrating a fast index search technique according to an embodiment of the present invention.

As shown in FIG. 1, a fast index search technique according to an embodiment of the present invention includes a precoding 110, a transmitter 120, a receiver 130, a feedback 140, and the like.

In this case, when the receiver 130 completely estimates the channel state information, the channel state information is transmitted to the precoding 110 through the feedback 140, in which a 3-bit or 6-bit vector index is used to use less uplink bandwidth. Use codebooks V (2,1,3), V (3,1,3), V (3,1,6), V (4,1,3), V (4,1,6), etc. Each codebook is optimized for generation efficiency and storage memory, and has both a transmitter and a receiver. The notation V (N _t , S, L) refers to a vector codebook, where N _t is the number of transmitter antennas, S is the number of streams, and L is the number of bits required for an index that can represent any vector. If L = 3, 3-bit feedback is performed, and if L = 6, 6-bit feedback is performed. 2 ^L is defined as the total number of vector indices for each transmitter antenna in each codebook. Table 1 shows the V (2,1,3) defined in the WiMAX technical paper.
Vector
index One 2 3 4 5 6 7 8 c ₁ One 0.7940 0.7940 0.7941 0.7941 0.3289 0.5112 0.3289 c ₂ 0 -0.5801 + j0.1818 0.0576 + j0.6051 -0.2978-j0.5298 0.6038 + j0.0689 0.6614 + j0.6740 0.4754-j0.7160 -0.8779-j0.3481

delete

delete

은 Transmitter 안테나 1에 대한 가중치 벡터이며,

는 Transmitter 안테나 2에 대한 가중치 벡터로서, 이때 각각의 벡터 인덱스들에 대한 실수 값 혹은 복소수 값은 임의로 양자화된 채널의 가중치 벡터를 나타낸다.

Is the weight vector for Transmitter antenna 1.

Is a weight vector for the transmitter antenna 2, where a real value or a complex value for each vector index represents a weight vector of a randomly quantized channel.

Before describing a new fast index search method according to the present invention, the existing index search method is described as follows.

는 [수학식 1]과 같고, 이때의 가중치 벡터

는 [수학식 2]와 같다.Random channel when there are two transmitter antennas and one receiver antenna

Is equal to [Equation 1], and the weight vector at this time

Is the same as [Equation 2].

delete

이 때,

은 Transmitter 안테나 1에 대한 채널이며,

는 Transmitter 안테나 2에 대한 채널이다.

,

는 가중치 벡터를 구성하는 요소로써

,

,

그리고

는 모두 임의의 복소수 값을 가진다.

At this time,

Is the channel for Transmitter Antenna 1,

Is the channel for Transmitter Antenna 2.

,

Is a component of the weight vector.

,

,

And

All have arbitrary complex values.

delete

delete

로 표현할 때, [수학식 4]에서처럼

와

의 내적의 크기를 벡터 인덱스 개수 만큼 모두 구한 후 이 중 최대값

를 가지는 벡터 인덱스

를 Precoding(110)으로 보내게 된다.A weight vector value corresponding to each vector index in the codebook is expressed as [Equation 3].

In terms of Equation 4,

Wow

Find the size of the dot product of as many as the number of vector indices and then take the maximum of them

Vector index with

Is sent to Precoding (110).

delete

delete

에 해당하는 가중치 벡터

를 Transmitter에 전송하며, Transmitter에서는 이 가중치 벡터를 데이터 신호

와 함께 Receiver로 전송하게 된다. Receiver는 [수학식 5]와 같이 수신 신호 y를 결정하고, [수학식 6]과 같이 수신단의 가중치

를 구해 최종 복조 신호

를 [수 학식 7]과 같이 계산한다.Precoding (110) gives the vector index

Weight vector corresponding to

Is transmitted to the transmitter, and the transmitter sends this weight vector to the data signal.

It is sent to Receiver together with. The receiver determines the reception signal y as shown in [Equation 5], and the weight of the receiver as shown in [Equation 6].

Obtain Final Demodulation Signal

Is calculated as shown in [Equation 7].

delete

delete

delete

를 찾기 위해 L = 3인 경우에는 8개의 벡터 인덱스, L = 6인 경우에는 64개의 벡터 인덱스를 모두 계산해야 한다. 하지만 아래에 서술한 새로운 빠른 인덱스 탐색기법은 기존의 인덱스 탐색 개수를 절반으로 줄여준다.Conventional index search method is the maximum value in [Equation 4]

To find, we need to calculate all 8 vector indices when L = 3 and 64 vector indices when L = 6. However, the new fast index finder method described below cuts the number of existing index searches in half.

For Table 1, the angles of the complex values corresponding to the respective vector indices for the codebook V (2,1,3) are shown in Equation 8.

delete

에서

는 [1 …

]에 해당하는 안테나들이며,

는 벡터 인덱스로 [1 …

]을 의미한다. 코드북 V(2,1,3)의

행은 실수 값이고,

행은 복소수 값이므로

행의 모든 벡터 인덱스들의 가중치 벡터의 각도는 0°이다.

행의 복소수 값에 대해 복소평면 상에서 이를 나타내면 도 2와 같다. notation

in

[1….

] Antennas,

Is the vector index and [1...

Means. Of codebook V (2,1,3)

Row is a real value,

Row is a complex value,

The angle of the weight vector of all the vector indices of the row is 0 °.

This is shown in the complex plane for the complex value of the row as shown in FIG.

에 대한 가중치 벡터는

,

인데

에 대한 복소평면 상에서의 각도

를 [수학식 9]를 이용하여 0°로 변환,

를 실수 값으로 만들어 준다.On any channel if N _t = 2

The weight vector for

,

Indeed

Angle on complex plane with respect to

To 0 ° using Equation 9,

Makes a real value.

delete

또한 같이 나타낸다.2 shows the new complex value of the second row in Equation 9

Also shown as:

에 대한 가중치 벡터와 N_t = 4 인 경우 임의의 채널에

에 대한 가중치 벡터에서의

는 [수학식 10], [수학식 11]과 같다.On any channel if N _t = 3

For any channel if N _t = 4

In the weight vector for

Is the same as [Equation 10], [Equation 11].

delete

delete

을 [수학식 12]와 같이 정의한다.Reference angle using [Equation 9], [Equation 10] and [Equation 11]

Is defined as shown in [Equation 12].

delete

을 구하는 것이 첫 번째 단계이다.The new fast index search method finds the angle on the complex plane from the weight vector of the channel state information, and then subtracts all the angles of the previous row from the angle of the last row.

Finding is the first step.

을 구하는 것이다.The second step is to find the angle on the complex plane from the weight vector of the codebook, and then subtract all the angles of the previous row from the angle of the last row.

To save.

= 3인 경우에는 [수학식 13], 코드북 V(4,1,3), V(4,1,6)과 같이

= 4인 경우에는 [수학식 14]와 같이 복소평면에서의 각도를 구할 수 있다. 각 수학식의 마지막 행에서

을 구할 수 있다.Like Codebook V (3,1,3), V (3,1,6)

= 3, Equation 13, codebooks V (4,1,3), V (4,1,6)

In the case of = 4, the angle in the complex plane can be obtained as shown in [Equation 14]. In the last row of each equation

Can be obtained.

delete

delete

을 [수학식 15]와 같이 정의하며, 이때의 표기법

는 [수학식 8]에서 정의한 것과 같다.Through [Equation 8], [Equation 13] and [Equation 14]

Is defined as [Equation 15], and the notation at this time

Is as defined in [Equation 8].

delete

을 기준으로 ± 90°라는 특정 범위에 속하는

를 선택한 후, 각각의

에 해당하는 벡터 인덱스만을 [수학식 4]에 고려하는 것인데, 이는

를 기준으로 ± 90°라는 특정 범위를 벗어나게 되면 도 2에서처럼 그 크기와 각도를 고려한 코드북과 채널 상태 정보의 가중치 벡터에 대한 내적의 크기가 180° 범위 안에 있는 다른 벡터 인덱스들보다 그 값이 작기 때문이다. 즉, 8개 각각의 벡터 인덱스에 해당하는 포인트에 근접해야 하는데 [수학식 10]처럼

을 기준으로 ± 90°를 벗어나게 되면 인접성이 범위 안의 다른 벡터 인덱스들보다 떨어지게 되어 [수학식 4]에 의한 연산 시 ± 90° 범위 안의 벡터 인덱스들만을 고려하면 계산량이 절반으로 줄어들게 된다.The third step of the new fast index explorer method is shown in Equation 16.

Within a certain range of ± 90 °

After selecting the

Only vector indices corresponding to Equation 4 are considered.

Since the dot product for the weight vector of the codebook and channel state information considering the size and angle is smaller than other vector indices within the 180 ° range as shown in FIG. to be. In other words, it should be close to the point corresponding to each of the eight vector indices.

If the deviation is ± 90 °, the adjacency becomes lower than other vector indices in the range, and the calculation amount is halved when considering only the vector indices in the ± 90 ° range during the calculation by [Equation 4].

을 기준으로 ± 90° 범위를 보여주며, 도 3은 L = 6인 코드북에 대한 특정

을 기준으로 ± 90° 범위를 보여준다.2 is specific for a codebook with L = 3.

± 90 ° range, with reference to FIG. 3, which is specific for a codebook with L = 6.

The range is ± 90 °.

delete

을 기준으로 ± 90°의 범위에 속하는 벡터 인덱스들만을 고려하면 [수학식 4]에서

에 해당하는 개수가 L = 3 인 경우에는 8 → 4, L = 6 인 경우에는 64 → 32 개로 감소한 다.As a result, using Equation 12, Equation 15, and Equation 16,

Considering only the vector indices in the range of ± 90 ° with reference to Equation 4

If L = 3, then 8 → 4, and if L = 6, then 64 → 32.

In FIG. 4, the performance of the codebook V (2,1,3) to which the new fast index search method is applied is shown. In FIG. 5, the performance of the codebook V (3,1,3) to the new fast index search method is shown. 6 shows the performance of the codebook V (3,1,6) to which the new fast index search scheme is applied.

In addition, FIG. 7 shows the performance of the codebook V (4,1,3) to which the new fast index search method is applied, and FIG. 8 shows the performance of the codebook V (4,1,6) to the new fast index search method. Indicated.

Although the embodiments of the present invention have been described above with reference to the accompanying drawings, those skilled in the art to which the present invention belongs may be embodied in other specific forms without changing the technical spirit or essential features of the present invention. You will understand that. Therefore, the embodiments described above are to be understood in all respects as illustrative and not restrictive.

1 is a conceptual diagram illustrating a fast index search technique according to an embodiment of the present invention.

을 기준으로 ± 90° 범위를 보여주는 개념도이다.2 is specific for a codebook with L = 3.

This is a conceptual diagram showing a range of ± 90 ° with reference to.

을 기준으로 ± 90° 범위를 보여주는 개념도이다.3 is specific for a codebook with L = 6

This is a conceptual diagram showing a range of ± 90 ° with reference to.

4 shows the performance of codebooks V (2,1,3) to which the new fast index search scheme is applied.

5 shows the performance of the codebook V (3,1,3) to which the new fast index search scheme is applied.

6 shows the performance of codebooks V (3,1,6) to which the new fast index search scheme is applied.

7 shows the performance of codebooks V (4,1,3) to which the new fast index search scheme is applied.

8 shows the performance of the codebook V (4,1,6) to which the new fast index search scheme is applied.

<Explanation of symbols for the main parts of the drawings>

110: Precoding 120: Transmitter

130: Receiver 140: Feedback

Claims (4)

In the fast index search method for feedback of channel state information using a codebook, Obtaining a first reference angle obtained from a weight vector of the channel state information by subtracting all angles of the previous row from the angle of the last row using the angle on the complex plane with respect to the complex value of the channel state information; Subtracting all angles of the previous row from the angle of the last row using the angle on the complex plane with respect to the corresponding complex value of the codebook vector indices to obtain a second reference angle obtained from the weight vector of the codebook; And A codebook comprising the step of performing a feedback process using a specific vector index having a maximum value by obtaining the size of the inner product for the vector index of the second reference angle belonging to a specific range based on the first reference angle Fast Index Search in Beamforming Systems The method of claim 1, wherein the first reference angle is, A fast index search technique in a beamforming system using a codebook, which is obtained through Equation 1 below. (Equation 1)

Where θ 'is the weight vector in the channel state, and θ is the angle on the complex plane in the channel state information of each transmitting antenna. The method of claim 1, wherein the vector index of the second reference angle, A fast index search technique in a beamforming system using a codebook, characterized in that the vector index corresponding to the second reference angle belonging to a specific range of ± 90 ° with respect to the first reference angle. The fast index search of a beamforming system using a codebook according to claim 1, wherein an optimal codebook index is selected by performing Equation 2 below on a vector index of a second reference angle belonging to the specific range. technique. (Equation 2)

Here, C _a is the weight vector of the codebook, W is the channel state information, b is the double maximum value after obtaining the size of the dot product of W and C _a as the number of vector indexes,

Means norm.

Images