TW200840256A - Method of precoding in a multi-antenna system - Google Patents

Abstract

The present invention provides a method of precoding in a multi-antenna system. The method generates precoding codebooks with extension Fourier series, based on the number of transmit antennas in the system and the number of groups of user data streams, so as to obtain a larger range for selection of precoding codebooks, which is helpful to find out which is the most suitable codebook.

Description

200840256 九、發明說明： 【發明所屬之技術領域】 本發明係關於一種無線電通信技術，尤其係關於一種用 於多天線系統中之預編碼方法。 【先前技術】 與單輸入單輸出（Single Input Single 0utput，SIS0)天線 無線通信系統相比，多輸入多輸出（Muhiple Input output，ΜΙΜΟ)無線通信系統能夠提供更高之通道容量。 對於多使用者ΜΙΜΟ系統而言，預編碼為一項能夠獲得較 高之系統效能及較低之接收處理複雜度之發射器優化技 術。藉由預編碼，發射器能夠對發射信號進行預矯正，以 減輕無線廣播通道之一些影響，包括符號間干擾、衰落及 通道相關性等。 對於預編碼而言，理想情況為將完整之ΜΙΜ〇通道狀態 資訊發送回發射器，但此會給反饋傳輸帶來較大負擔，在 實際系統中幾乎不可能實現。因此，在有限之反饋資訊傳 輸中會考慮採用碼薄（Codebook)來表示反饋資訊，即在預 編碼時產生一定數目之碼薄，而接收機選擇適當之碼薄發 回發射器。如此，如何在預編碼階段對碼薄進行優化設計 成為其中一個關鍵問題。 在現有之預編碼方案中，每使用者單一速率控制（p^_ User Unitary Rate Contro卜 PU2RC)為一種主流方案。假設 一個ΜΙΜΟ系統之節點B(N〇de…具有％個發射天線，相應 之PU2RC系統可參照圖！所示。其中一組預編碼矩陣 119650.doc 200840256 Ε = {ε(0) 妨） ^g) "Λ/-1 為第g個預編碼矩陣， e m 為此組中之第m個預編碼矢量。每一使用者設備將針對 此組預編碼矩陣E中之每一矩陣裏之每一矢量計算一個通 道品質指示符（Channel Quality Indicator，CQI)。一 個預編 碼矩陣中之預編碼矢量可被設計成多使用者或者同一使用 者之多個資料流以支持空間劃分多工（Spatial Divisi〇n200840256 IX. Description of the Invention: [Technical Field] The present invention relates to a radio communication technology, and more particularly to a precoding method for use in a multi-antenna system. [Prior Art] Compared with a single input single output (SIS0) antenna wireless communication system, a multi-input and multi-output (Muhiple Input Output) wireless communication system can provide higher channel capacity. For multi-user systems, precoding is a transmitter optimization technique that achieves higher system performance and lower reception processing complexity. By precoding, the transmitter can pre-correct the transmitted signal to mitigate some of the effects of the wireless broadcast channel, including intersymbol interference, fading, and channel correlation. For precoding, the ideal situation is to send complete channel state information back to the transmitter, but this puts a heavy burden on the feedback transmission, which is almost impossible to implement in practical systems. Therefore, in a limited feedback information transmission, a codebook is considered to represent the feedback information, that is, a certain number of codebooks are generated during precoding, and the receiver selects an appropriate codebook to send back to the transmitter. Thus, how to optimize the codebook in the pre-coding stage becomes one of the key issues. In the existing precoding scheme, per-user single rate control (p^_User Unitary Rate Contro Bu PU2RC) is a mainstream scheme. Suppose a node B of a system (N〇de... has % transmit antennas, and the corresponding PU2RC system can be referred to Figure! One set of precoding matrix 119650.doc 200840256 Ε = {ε(0) )) ^g ) "Λ/-1 is the gth precoding matrix, em is the mth precoding vector in this group. Each user equipment will calculate a Channel Quality Indicator (CQI) for each of the vectors in the set of precoding matrices E. The precoding vector in a precoding matrix can be designed to be multi-user or multiple streams of the same user to support spatial division multiplexing (Spatial Divisi〇n

Multiplexing，SDM)或空間劃分多向存取（Spatia][ Divisi()nMultiplexing, SDM) or spatial partitioning multi-directional access (Spatia][ Divisi()n

ΟΟ

Multiple Access，SDMA)技術。 藉由選擇適當之G值以及判斷使用者設備需要反饋給節 點β之資訊量，節點B可選擇反饋之資訊量以獲得不同之 適應性。例如： 為獲得最高之適應性，使用者設備可反饋Ε中每一矩陣 之每一 CQI之值，如此總共每一使用者設備要反饋個 CQI之值； 而為獲得最低之管理成本，使用者設備只需反饋最佳之 CQI值以及對應之矢量索引，此時每一使用者設備除實際 之CQ1值以外’需要額外傳輸l〇g2(GM)位元之資料。由於 每使用者只能被安排在某一個特定之波束上，此種反饋 八支持有限之SDMA。而當一個使用者根據時頻單元被安 排時，此種反饋僅支持發射波束形成（Beam_F〇rming); 每一使用者設備亦可反饋對應於首選或最佳組之Μ個 CQ1之值，以便獲得一個較佳之SDM及SDMA之混合。 節點Β可選擇一個組g並伴隨一個ε⑻，同時基於預定之 叹。十將最多Μ個獨立資料流安排給不同使用者設備或者同 H9650.doc 200840256 一個使用者設備。對於碼薄之要求為它必須係唯一的，以 保證預編碼不會改變通道容量。 在PU2RC系統中碼薄之設計係根據下面之等式（1)之傅立 葉變換來進行： 2m ^(g) 4mMultiple Access, SDMA) technology. By selecting the appropriate G value and determining the amount of information that the user equipment needs to feed back to node β, Node B can select the amount of information to be fed back to achieve different fitness. For example: In order to obtain the highest adaptability, the user equipment can feed back the value of each CQI of each matrix in the ,, so that each user equipment should feed back the value of CQI; and in order to obtain the lowest management cost, the user The device only needs to feed back the best CQI value and the corresponding vector index. At this time, each user equipment needs to transmit additional data of l〇g2 (GM) bits in addition to the actual CQ1 value. Since each user can only be placed on a particular beam, this feedback eight supports limited SDMA. When a user is arranged according to the time-frequency unit, such feedback only supports transmit beamforming (Beam_F〇rming); each user equipment can also feed back the value of one CQ1 corresponding to the preferred or optimal group, so that A better mix of SDM and SDMA is obtained. The node Β can select a group g with an ε(8), based on the predetermined sigh. Ten will arrange up to one independent data stream to different user devices or a user device with H9650.doc 200840256. The requirement for the codebook is that it must be unique to ensure that the precoding does not change the channel capacity. In the PU2RC system, the design of the codebook is performed according to the Fourier transform of the following equation (1): 2m ^(g) 4m

M G為使用者分組數，M為發射天線數 〇 PU2RC之碼薄中給出兩個實例，即： 當 M=2，G=2 時， '1 1 " E]=' "1 1 " l -1_ ，' J 一 j_ 當 M=4，G=2 時， m +MG is the user group number, M is the number of transmitting antennas 〇 PU2RC codebook gives two examples, namely: When M=2, G=2, '1 1 " E]=' "1 1 &quot ; l -1_ , ' J a j_ when M=4, G=2, m +

G (1) 0<n<M-1 (2) 1 1 1 1 1 1 ejn'2 ejn ej^'2 77 1 ejn ej2n ββπ 1 ej3n!2 ej3n ej9n'2 ϋ 1111 1 y· -1 -） (3) 1-11-1 1 一7 —1 j 1 1 1 1 一 1 ejnlA ejMA ej5nf4 ，/4 λ/4 ejnl1 ej述 β；1〇π/4 严2 /TIM β·/9π/4 γ.15π/4 eJ2\nlA 1 1 1 119650.doc (4) 200840256 對應之PU2RC處理過程包括： 1·控制器UK)收集反饋資訊，其中包含一個首選之預編 碼矩陣之索引以及該矩陣中所有預編碼矢量對應之CQI之 值； " 2·使用者分組單元200將要求同一個首選之預編碼矩陣 之使用者編為一組； 3. 多工單元则在不同之組中選擇具有最高優先級之彼 組，並在敎之組中選擇具有最高優先級之資料流（該資 料流可能來自不同使用者或者同一使用者）； 4. 自適應調變編碼（Adaptive M〇dulati〇n a" AMC)單以⑻根據相關之吻，針對所敎之資料流執行 適當之AMC處理； 5. 預編碼單元·對選定之組進行預編碼處理。 其中預編碼矩陣亦為該之預編碼碼薄。 Ο 數ST二處理^及碼薄之缺點在於，所能提供之碼薄 =二在實際系統中，只能藉由計算所有可能之 生=來得到優化之系統輸出。當無線傳播通道發 二:化：’碼薄選擇之可修改範圍亦报有限，此為由於； ^天i數目、使用者數目及碼薄分組數目被辞定時 所示。 夺碼璋便固定為等式（2) [發明内容】 本發明之目的之一為提出一 ^ 夕大^線系統之預结》级古 法，以基於擴展傅立葉級數對碼薄進行優化。 馬方 119650.doc 200840256 根據本發明之一種用於多天線 深糸統之混合自動重傳請求 發射器方法，包括步驟： 一種用於多天線系統之預編码方法，包括步驟： a•選擇適當之使用者資料流分組數； b ·根據系統發射天線數及該使用者 從用者貝枓流分組數，利用 擴展傅立葉級數生成預編碼碼薄； c·獲得來自使用者之包含首選 ^之預、、扁碼碼簿以及相關G (1) 0<n<M-1 (2) 1 1 1 1 1 1 ejn'2 ejn ej^'2 77 1 ejn ej2n ββπ 1 ej3n!2 ej3n ej9n'2 ϋ 1111 1 y· -1 -) (3) 1-11-1 1 - 7 - 1 j 1 1 1 1 - 1 ejnlA ejMA ej5nf4 , /4 λ/4 ejnl1 ej describes β; 1〇π/4 严 2 /TIM β·/9π/4 γ .15π/4 eJ2\nlA 1 1 1 119650.doc (4) 200840256 The corresponding PU2RC processing process includes: 1. The controller UK) collects feedback information, including an index of a preferred precoding matrix and all pre-preparations in the matrix. The value of the CQI corresponding to the coding vector; " 2. The user grouping unit 200 groups users who require the same preferred precoding matrix; 3. The multiplex unit selects the highest priority among the different groups. The other group, and select the data stream with the highest priority in the group (the data stream may come from different users or the same user); 4. Adaptive modulation coding (Adaptive M〇dulati〇n a" AMC (8) Perform appropriate AMC processing for the selected data stream according to the relevant kiss; 5. Precoding unit • Perform precoding processing on the selected group. The precoding matrix is also the precoding codebook. The disadvantage of the number of ST two processing and codebook is that the codebook can be provided = two In the actual system, the optimized system output can only be obtained by calculating all possible lifetimes. When the wireless transmission channel is sent two: The following: The configurable range of the codebook selection is also limited, which is due to; ^ The number of days i, the number of users and the number of codebook groups are indicated by the resignation. The code grab is fixed to the equation (2) [Summary of the Invention] One of the objects of the present invention is to propose a pre-knot-level ancient method for optimizing the codebook based on the extended Fourier series. Ma Fang 119650.doc 200840256 A hybrid automatic repeat request transmitter method for a multi-antenna deep system according to the present invention, comprising the steps of: a precoding method for a multi-antenna system, comprising the steps of: a The number of user data stream packets; b. According to the number of transmitting antennas of the system and the number of packets from the user's stream stream, the pre-coded codebook is generated by using the extended Fourier series; c· obtaining the preferred method from the user Pre-, flat code book and related

U CQI之值之反饋資訊； d·根據該反饋資訊對使用者進行分組；以及 e·根據預定規則選擇一個使用者組並對選定之使用者組 進行預編碼處理。 與現有技術相比’本發明之多天線系統之預編碼方法能 夠利用擴展傅立葉級數獲得更大之碼薄選 利於找到最合適之碼薄。 攸而有 晴專利範圍中之内 ，本發明之其他目 藉由參考以下結合附圖之說明以及申 容，並且隨著對本發明之更全面之理解 的及效果將變得更加清楚及易於理解。 【實施方式】 本發明預編瑪方法採 該擴展傅Α葉級數之 相對於現有技術中之PU2rc方案 用擴展傅立葉級數對碼薄進行優化 原理如下面之等式（5)所示：Feedback information of the value of U CQI; d. grouping users according to the feedback information; and e. selecting a user group according to a predetermined rule and performing precoding processing on the selected user group. Compared with the prior art, the precoding method of the multi-antenna system of the present invention can utilize the extended Fourier series to obtain a larger codebook to find the most suitable codebook. The scope of the present invention is to be understood by the following description and the accompanying drawings, [Embodiment] The pre-matrix method of the present invention adopts the extended Fourier series. Compared with the prior art PU2rc scheme, the extended Fourier series is used to optimize the codebook. The principle is as shown in the following equation (5):

2πη Μ 料吾+△2πη Μ material I +△

0 < Δ < 丄 G (5) 119650.doc 200840256 其中，△定義為一個起始點，並可舻M箱好 I 1根據預編碼竭薄分组 之要求在0到石之間變化，等式中· '' /、他又里之疋義與等式 (1)相同，如此，便獲得一個擴展傅立葉級數。 下面結合具體實施例詳細說明本發明如何基於擴展傅立 葉級數對碼薄進行優化。 假設系統中之發射天線數為Μ，經由此等_發射天線 發射之第0個使用者之符號為，其中 灸=1，…，尺，尺表示使用者數，上標τ表示對矩陣進行轉置 (T/anspose)操作，不同使用者之通道響應矩陣為，則第 灸個使用者之接收信號可被建模成： %==ΗλΣ«(/)^+η^ k=l (6) e/)之定義可參照等式（1)所示，為後面更佳地闡釋，假 設此處M=尺，此亦屬於通常情況，而實際上當時之處0 < Δ < 丄 G (5) 119650.doc 200840256 where △ is defined as a starting point, and 舻M box is good I 1 varies from 0 to stone according to the requirements of precoding exhaust grouping, etc. In the formula, '' /, he is the same as equation (1), so that an extended Fourier series is obtained. The present invention will be described in detail below with reference to specific embodiments to optimize the codebook based on the extended Fourier series. Assuming that the number of transmitting antennas in the system is Μ, the symbol of the 0th user transmitted via these _ transmitting antennas is, where moxibustion = 1, ..., ruler, ruler indicates the number of users, and superscript τ indicates the rotation of the matrix Set (T/anspose) operation, the channel response matrix of different users is, then the receiving signal of the moxibustion user can be modeled as: %==ΗλΣ«(/)^+η^ k=l (6) The definition of e/) can be referred to the equation (1), which is better explained later. Suppose M=foot is here, which is also the normal situation, but actually it is at that time.

理亦類似。對於迫零（Zer〇 Forcing，ZF)檢測如下面之等 式（7)所示： ⑺ 其中上&H及-1表示對矩陣進行共輛轉置（c〇njUgate Transpose)及反（Inverse)操作。 最合適之碼薄為自檢測出之符號中選擇一個能使信號干 擾加雜訊比（Signal to Inference and Noise Ratio，SINR)總 數最大，即： 119650.doc (8)200840256 ^ κ g = argmax [ ^ SINRis^)] gE(0,...G) k—1 為找到最佳之碼薄，等式（8)中之值將根據不同組之大 小之不同碼簿被計算多次。以M=2之情況為例，當M=2, G=4時，如果根據現有技術之等式（1)，碼薄為：E. — . _ _ _exP(j^/4) exp(j；r5/4)The same is true. For Zer〇Forcing (ZF) detection, as shown in the following equation (7): (7) where upper &H and -1 indicate that the matrix is transposed (c〇njUgate Transpose) and inverse (Inverse) operating. The most suitable codebook is to select one of the detected symbols to maximize the total number of Signal to Inference and Noise Ratio (SINR), namely: 119650.doc (8)200840256 ^ κ g = argmax [ ^ SINRis^)] gE(0,...G) k-1 To find the best codebook, the values in equation (8) will be calculated multiple times according to the different codebooks of different groups. Taking M=2 as an example, when M=2, G=4, if according to the prior art equation (1), the codebook is: E. — . _ _ _exP(j^/4) exp(j ;r5/4)

：VT :万 '1 1 一 1 一 _1 -1 "1 1 一 1 J 一 j一 ，ε3=7?_ (9) ί. 由此可看出，儘管分組大小自2變為4,但當g=4時，盆 個碼薄與G=2時之碼薄重複。此種重複對於有效地選擇 合適之碼薄係不利的，而等式⑺中之擴展傅立葉級數可 =每-組中之碼薄數目，即等式⑴中之傅立葉基數加上 起始點△，其優點在於：能夠為碼薄 葉級數，從而有利於尋找最合適 ：之傅立 之车算所有可能之碼薄之可行性才能獲得理超 糸、，先貝枓流量；當無線傳播， 之可修改範圍變得非當…夂化時碼溥選擇 傳輸之實際情況 更加適用於ΜΙΜ0預編石馬 此種優點可驻1 例： 错由下面兩個例子來具體描述，以Μ=2為 (1)假如 G=4，△ 碼薄為： 119650.doc -12- (10)200840256 T2 exp( j7i / 8) exp(】π9 / 8):VT : million '1 1 -1 _1 -1 "1 1 -1 J -j one, ε3=7?_ (9) ί. It can be seen that although the packet size changes from 2 to 4, However, when g=4, the codebook thinning is repeated with the codebook of G=2. Such repetition is disadvantageous for the effective selection of a suitable codebook, and the extended Fourier series in equation (7) can be = the number of codebooks per group, ie the Fourier base in equation (1) plus the starting point Δ The advantage is that it can be a thin leaf series, which is beneficial to find the most suitable: the feasibility of all possible codebooks can be obtained by the Fourier car, in order to obtain the super-small, first Bellow flow; when wirelessly propagated, it can be modified The scope becomes improper... The actual situation of the code selection and transmission is more suitable for the ΜΙΜ0 pre-programmed stone horse. This advantage can be stationed in 1 case: The error is described by the following two examples, Μ=2 is (1) If G=4, the △ codebook is: 119650.doc -12- (10)200840256 T2 exp( j7i / 8) exp(]π9 / 8)

E 1 E. E, exp(】π3 / 8) exp( j7il 1 / 8) exp(』π5 / 8) exp( j7il 3 / 8) exp(』π7 / 8) exp( j7il 5 / 8) 假如8 16 碼薄為：E 1 E. E, exp(]π3 / 8) exp( j7il 1 / 8) exp(』π5 / 8) exp( j7il 3 / 8) exp(』π7 / 8) exp( j7il 5 / 8) If 8 16 codebook is:

Er E, 1 1 1 1 exp(』π /16) exp(』π9 /16) 1 1 exp(』π3 /16) exp( j7il 1 /16) 1 1 exp(』π5 /16) exp( j7il 3 /16) 1 1 " exp(】π7 /16) exp( j7il 5 /16) 1 1 一 exp(】π9 /16) exp(』π25 /16) 一 1 1 exp( j7il 1 /16) exp(』π27 /16) 1 1 exp(』π13 /16) exp(』π29 /16) 1 1 exp(』π15 /16) exp(』π31 /16) (11) 由等式（2)、（10)及（11)可看出，在G=2，G=4及G=8時， 碼薄完全不相同。此便證明利用擴展傅立葉級數可獲得更 大碼薄選擇範圍，從而有利於找到最合適之碼薄。 碼薄為 (2)假如 G=4，△ 119650.doc -13 - 32 (12) 200840256 E, E, 1 E. '4i 1 "Ti exp(jTi / 32) exp( jm33 / 32) 1 1 exp( ju9 / 32) exp( ]π41/32) 1 1 _exp( j7il 7 / 32) exp(ju49/32) 1 1 _exp〇25 / 32) exp(』π57 / 32) 亦可獲得完全 自等式（10)及（12)可看出，甚至在G=4時， 不同之碼薄組。 f Ο 因此，參照圖2,本發明多天線系統之預編碼方法首先 遥擇適當之使用者資料流分組數目，以禮 ^心久讀之資訊量 (步驟10); 然後根據系統中發射天線之數目及該使用者資料流分組 數，利用前述擴展傅立葉級數生成預編碼碼薄1); 節點B接收來自使用者終端之反饋資訊，其中包含首選 之預編碼碼薄之索引以及該碼薄中所有預編碼矢量對應之 CQI之值（步驟12); " 根據該反饋資訊中首選之預編碼碼薄之選擇情況對使用 者進行分組（步驟13); 選擇-個具有最高優先級之使用者分組，以及根據預定 之預編碼方案對選定之使用者組進行預編碼處驟 14) ° 本發明多天㈣統之預編碼方法在硬體上基本可採用圖 1中之PU RC發射器結構來執行，但由於本發明多天線系 扁I方法具有更大之預編碼碼薄範圍，因而在效能 119650.doc -14- 200840256 上優於現有技術之pu2rc方案。 應注Μ上述貫施例用於說明、而非限制本發明，並 且’在不脫離所附申請專利範圍之保護範圍之前提下，熟 習此項技術者應當理解，對上述本發明所公開之多天線^ 統之預編碼方法，亦可在不脫離本發明内容之基礎上做出 各種改良。因此，本發明之保護範圍應當由所附申請專利 範圍之内容確定。此外，不應當將申請專利範圍中之任何 參考標記闡釋為限制申請專利範圍之保護範圍。 【圖式簡單說明】 現在結合附圖對本發明進行更詳細之描述，其中： 圖1展示一個pu2rc發射器之結構示意圖；以及 圖2展示本發明預編碼方法之流程圖。 【主要元件符號說明】 100 控制器 200 使用者分組單元 300 多工單元 400 AMC單元 500 預編碼單元 119650.doc _ 15.Er E, 1 1 1 1 exp(』π /16) exp(』π9 /16) 1 1 exp(』π3 /16) exp( j7il 1 /16) 1 1 exp(』π5 /16) exp( j7il 3 /16) 1 1 " exp(]π7 /16) exp( j7il 5 /16) 1 1 an exp(]π9 /16) exp(』π25 /16) a 1 1 exp( j7il 1 /16) exp( Ππ27 /16) 1 1 exp(』π13 /16) exp(』π29 /16) 1 1 exp(』π15 /16) exp(』π31 /16) (11) From equations (2), (10) And (11), it can be seen that when G=2, G=4 and G=8, the codebooks are completely different. This proves that the extended Fourier series can be used to obtain a larger code selection range, which is helpful for finding the most suitable codebook. The codebook is (2) if G=4, △ 119650.doc -13 - 32 (12) 200840256 E, E, 1 E. '4i 1 "Ti exp(jTi / 32) exp( jm33 / 32) 1 1 Exp( ju9 / 32) exp( ]π41/32) 1 1 _exp( j7il 7 / 32) exp(ju49/32) 1 1 _exp〇25 / 32) exp(』π57 / 32) can also obtain complete automorphism (10) and (12) can be seen, even when G=4, different codebook groups. f Ο Therefore, referring to FIG. 2, the precoding method of the multi-antenna system of the present invention first selects the appropriate number of user data stream packets to read the information amount of the long-term reading (step 10); The number and the number of the user data stream packets, using the foregoing extended Fourier series to generate a precoding codebook 1); the Node B receives feedback information from the user terminal, including an index of the preferred precoding codebook and the codebook The value of the CQI corresponding to all the precoding vectors (step 12); " grouping the users according to the selection of the preferred precoding codebook in the feedback information (step 13); selecting the user with the highest priority Grouping, and precoding the selected user group according to a predetermined precoding scheme. 14) The multi-day (four) precoding method of the present invention can basically adopt the PU RC transmitter structure of FIG. 1 on the hardware. Execution, but because the multi-antenna flat I method of the present invention has a larger range of pre-coded codebooks, it is superior to the prior art pu2rc scheme in performance 119650.doc -14-200840256. The above-described embodiments are intended to be illustrative, and not to limit the invention, and are to be considered as <Desc/Clms Page number> The precoding method of the antenna system can also be variously modified without departing from the scope of the present invention. Accordingly, the scope of the invention should be determined by the scope of the appended claims. In addition, any reference signs in the scope of the claims should not be construed as limiting the scope of the claims. BRIEF DESCRIPTION OF THE DRAWINGS The invention will now be described in more detail with reference to the accompanying drawings in which: FIG. 1 shows a schematic diagram of a pu2rc transmitter; and FIG. 2 shows a flow chart of the precoding method of the present invention. [Explanation of main component symbols] 100 Controller 200 User grouping unit 300 Multiplex unit 400 AMC unit 500 Precoding unit 119650.doc _ 15.

Claims (1)

200840256 十、申請專利範園： 1 · 一種用於多天線系統之預編碼方 /ir ’匕括步驟： a·選擇適當之使用者資料流分組數； 匕根據系統發射天線數及該使用者資料漭八 用擴展傅立葉級數生成預編碼碼薄； 1刀、、且，1 c·獲得來自使用者之包含首選 COI^ ^ ^ ^ ^ ^ 預、扁碼碼薄以及相關 CQI之值之反饋資訊； d·根據該反饋資訊對使用者進行分組·以 e·根據預定規則選擇一個使 电w… 使用者組並對選定之使用者 組進饤預編碼處理。 2·如請求項1之方法，其中步驟b 料办4二、 j用以下擴展傅立葉級 數來生成預編碼碼薄： m + — + A G ’⑷一一冷 0<Δ< 去 / 如其中G為使用者分組數，μ為 〇 發射天線數，。 3·如請求項2之方法，其中，△定義 攄箱绝M、 ，馬一個起始點，並可根 據預編碼碼溥分組之要求在〇到丄 ^ G之間變化。 4·如請求項1之方法，其中步驟d ^ r, 勺竹要未同一個首選之預 編碼矩陣之使用者編為一組。 5·如明求項丨之方法，其中步 Γηγ τ亦包括根據相關之 CQ1 ’採用自適應調變編碼，斜 之眘粗A Α ^ 針對所選定之使用者組中 之貝枓流進行處理。 119650.doc200840256 X. Application for Patent Park: 1 · A precoding method for multi-antenna systems/ir's steps: a. Select the appropriate number of user data stream packets; 匕 According to the number of system transmit antennas and the user data漭8 Use the extended Fourier series to generate the precoding codebook; 1 knife, and, 1 c· obtain feedback information from the user including the preferred COI^ ^ ^ ^ ^ ^ pre-, flat codebook and related CQI values d. Grouping users according to the feedback information. e. Selecting a user group according to a predetermined rule and performing precoding processing on the selected user group. 2. The method of claim 1, wherein step b is configured to generate a precoding codebook using the following extended Fourier series: m + — + AG '(4) one by one cold 0<Δ< For the number of users grouped, μ is the number of transmit antennas. 3. The method of claim 2, wherein Δ defines a box M, a starting point of the horse, and may vary from 〇 to 丄 ^ G according to the requirements of the precoding code grouping. 4. The method of claim 1, wherein the step d ^ r, the spoon bamboo is not grouped together with the user of the preferred precoding matrix. 5. The method of claim ,, wherein the step Γ γ τ also includes adaptive modulation coding according to the associated CQ1 ', and the oblique A A Α ^ is processed for the stream of the selected user group. 119650.doc