1281795 九、發明說明: 本發明係有關信號處理領 離(B S S )技術將_源 ^ 寸別㈣㈣盲信號分 ,、你秸唬混合分離。 先前技術 目^號分離係涉及從組合 — 信號係包含源信號之混合:二:::匕號’其中該虹合 資訊來執行。 、。说上之政應相關之有限 一例係宴會上之個人可將單聲 分離時之熟悉,,雞尾酒 有晕曰組合 機及個人益線通俨壯/並f二源勿硪特別可應用至手 多頻册孫、由四夕〇衣 八令通常共存於相同頻譜中之許 ^係被〇射頻發射器混亂。共頻道 被預期只會隨著年復一年低功率,如藍 = 網路之未授權無線技術發展而更糟。 -他個人邊域 三,普遍使用盲信號分離技術係為主分量分析(pcA),獨 立分量分析(ICA)及單值分解(SVD)。主分量分析係涉及源 信號第-及第二矩統計值’且被用於源信號之信號雜訊比 很尚時。另外,獨立分量分析係被使用涉及主分量分析處 理之後之源信號第三及第四矩統計值。可替代是,單值分 解可被用來以其特徵值為基礎將源信號與源信號混合物二 離0 無論被應用之盲信號分離技術為何,複數感測器係被用 來接收來自各信號源之不同源信號混合物。各感測器係輸 7 1281795 出源k 5虎唯加總之源信號混合物。通常,接收器並不知 頻道係數及原始源信號。信號唯一加總係被用來集結混合 矩陣。適當盲信號分離技術接著被施加至混合矩陣以便將 預期源彳a號與源信號混合物分離。 、 如一例,美國專利號第6,799,170係揭示使用獨立分量分 析將獨立源信號與源信號混合物分離。複數感測器係接收 源信號混合物,而處理器隨時採取源信號混合物之樣本並 儲存各樣本當作資料向量來創造資料組。各感測器係輸出 源信號唯一加總之源信號混合物。獨立分量分析模組係執 行該貧料向量之獨立分量分析將源信號混合物中之獨立源 信號與其他信號分離。 感測器係被空間性彼此分離,而該處理器僅為各感測器 產生一資料向量來創造資料組。\^17〇專利案亦揭示感測器 數N等於或大於源數M,也就是N>M可集結資料組。該^ 施之一問題係當源數M增加時,感測器數N亦增加。小^ 可攜式通信裝置係對大量感測器數N具有少許可用體積, 而衣。又δ亥感測益於通信裝置外侧對使用者是個問題。 η吴國專利號第6,931,362係揭示使用盲信號分離分離信 唬之另一方法。被揭示盲信號分離技術係形成具有可最小 化因干擾發射ϋ及高斯雜訊所造成之均方差之混合矩陣筆 ?應陣列權重之混合矩陣。該混合權重係最大化信號干擾 雜訊比。如同’17G專利案’感測器亦被空間性彼此分離, 而感測器數N等於或大於源數M以集結混合矩陣。再者, 各感測ϋ係提供單以至產生A量面積給可攜式通信裝置 8 1281795 之混合矩陣。 發明内容 、=\上述背景’本發明目的係提供一通信震置,包含可 以盲仏號分離技術接收源信號混合物使預期源信號可 離之一小型天線陣列。 f據本發明之此及其他目的,特性及優點係藉由通信裝 置提供分離Μ信號源所提供之源信號,該通信裝置 可接收該Μ源信號不同加總之一天線陣列。—接收器或接 收器組件係被連接至該天線陣列,而一盲信號分離處理器 係被連接至該接收器以形成混合矩陣。該混合矩陣係包含 被該=線陣列接收之該Μ源信號不同加總。該盲信號分離 處理裔接著將預期源信號與該混合矩陣分離。 除了使用空間分離感測器為該混合矩陣提供Μ源信號不 同加總之外,小型天線陣列可被替代使用。針對可攜式通 信裝置,因為天線陣列提供一個以上輸入至該混合矩陣而— 仍維持緊密,所以盲信號分離技術可被使用。 特別是,路徑選擇可被用來提供信號不同加總來做盲信 號分離處理。路徑選擇係被執行使被用來集結混合矩陣之 所有源號加總均被產生關聯(第一及第二矩)及/或統計 (第二及第四矩)獨立。也就是說,天線波束係被選擇性形成〜 使入射信號得以被選擇性挑選來提供混合矩陣申之源信號 新加總來取代不被產生關聯及/或統計獨立之加總。 天線陣列可包含N天線元件來接收]V[源信號至少N不 9 12817951281795 IX. INSTRUCTIONS: The present invention relates to the signal processing separation (B S S ) technique, which divides the _ source ^ inch (four) (four) blind signal, and mixes and separates your straw. The prior art is related to the separation - the signal system contains a mixture of source signals: two::: apostrophe' where the rainbow information is executed. ,. A limited number of cases related to politics should be familiar with the individual at the banquet, cocktails with syncope and personal benefits. The frequency book grandchildren, who are coexisting in the same spectrum by the four eves, are plagued by RF transmitters. Common channel is expected to only worsen with low power year after year, such as blue = network unlicensed wireless technology. - His personal domain 3. The commonly used blind signal separation techniques are principal component analysis (PCA), independent component analysis (ICA) and single value decomposition (SVD). Principal component analysis involves the source signal first- and second moment statistics' and the signal-to-noise ratio used for the source signal is very good. In addition, the independent component analysis is performed using the third and fourth moment statistics of the source signal after the principal component analysis process. Alternatively, the single value decomposition can be used to separate the source signal from the source signal mixture based on its eigenvalues. Regardless of the blind signal separation technique applied, complex sensors are used to receive signals from each source. Different source signal mixtures. Each sensor is fed with a source signal mixture of 7 1281795. Usually, the receiver does not know the channel coefficients and the original source signal. The unique summation of the signals is used to assemble the mixing matrix. A suitable blind signal separation technique is then applied to the mixing matrix to separate the expected source 彳a number from the source signal mixture. As an example, U.S. Patent No. 6,799,170 discloses the use of independent component analysis to separate an independent source signal from a source signal mixture. The complex sensor receives the source signal mixture, and the processor takes a sample of the source signal mixture at any time and stores each sample as a data vector to create a data set. Each sensor is a source signal mixture whose output source signal is uniquely summed. The independent component analysis module performs an independent component analysis of the poor vector to separate the independent source signals in the source signal mixture from the other signals. The sensors are spatially separated from one another, and the processor creates a data set for each sensor to create a data set. The \^17 patent also discloses that the sensor number N is equal to or greater than the source number M, that is, the N>M can aggregate the data set. One problem with this is that as the number of sources M increases, the number of sensors N also increases. The small ^ portable communication device has a small available volume for a large number of sensors N. It is also a problem for the user to measure the outside of the communication device. η Wu Guo Patent No. 6,931,362 discloses another method of separating and separating signals using a blind signal. The disclosed blind signal separation technique forms a mixing matrix with a matrix of weights that can minimize the mean square error caused by interfering emission chirp and Gaussian noise. This blending weight maximizes the signal interference noise ratio. As with the '17G Patent' sensors, they are also spatially separated from each other, and the sensor number N is equal to or greater than the source number M to assemble the mixing matrix. Furthermore, each sensing system provides a mixing matrix that is single to produce an A-area area to the portable communication device 8 1281795. SUMMARY OF THE INVENTION The present invention is directed to providing a communication stimuli comprising a small antenna array that can receive a mixture of source signals by a blind nick separation technique to isolate a desired source signal. According to this and other objects, features and advantages of the present invention, a source signal provided by a separate source of signal is provided by a communication device that can receive an antenna array of different sums of the source signals. - A receiver or receiver component is coupled to the antenna array and a blind signal separation processor is coupled to the receiver to form a mixing matrix. The hybrid matrix includes different summations of the source signals received by the = line array. The blind signal separation process then separates the expected source signal from the mixing matrix. In addition to the use of spatial separation sensors to provide different summation of the source signals for the hybrid matrix, small antenna arrays can be used instead. For portable communication devices, blind signal separation techniques can be used because the antenna array provides more than one input to the mixing matrix - still maintaining tightness. In particular, path selection can be used to provide different signal summations for blind signal separation processing. The path selection is performed such that all source numbers used to assemble the mixing matrix are summed (first and second moments) and/or statistics (second and fourth moments) independently. That is, the antenna beam is selectively formed - the incident signal is selectively selected to provide a mixed matrix source signal to add up to replace the sum that is not associated and/or statistically independent. The antenna array may comprise an N antenna element to receive]V [source signal at least N not 9 1281795
冋加總,N及Μ大於2 〇 —批也丨吳叮、士 A 谨裡从, 控制益可被連接至天線陣列來 選擇性形成至少N天線波束。 4木 一接收器組件,被連接至今綠圭 少N笊她 亡丄至°亥天線陣列接收該Μ源信號至 "目“號分離處理器,被連接至該接收哭 、、且?形成包含該Μ源信號至少2 N不同加總之混合矩陣: 目信號分離處理H亦可決u源錢不同 :關聯或統計獨立,若否,則與控制器共同操作來形= 接收m5#b新不同加總之不同波束來取代混合矩陣中不 f相關聯聽計獨源信號不同加總。預期源信號係 藉由盲信號分離處理器被與混合矩陣分離。 另-實施例中,接收器組件可包含N耙式接收器,各孝巴 式接收器#包含可針對被連接至此之各天線元件所接收之 Μ源信號不同加總選擇让不同多路分量之让耙指。盲信號 分離處理器可被連接至N耙式接收器來形成混合矩陣,其U 可包含達Μ源信號至少N不同加總之至少kN不同多路分 里。該此合矩陣係具有等於達kN之排序。 N天線70件可被相關聯。該N相關聯天線元件可包含N 主動天線元件使該天線陣列得以形成相位陣列。可替代 是,該N關聯天線元件可包含至少一主動天線元件,及至 多N-1被動天線元件使該天線陣列得以形成切換波束天線。 Μ源信號各加總係為線性。盲信號分離處理器可以主分 量分析,獨立分量分析及單值分解至少其中之一為基礎將 預期源信號與混合矩陣分離。 對路徑選擇實施例之一增強係涉及可接收附加信號加總In addition, N and Μ are greater than 2 〇—Batch 丨 叮 叮 叮, 士 A 里 从 , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , 4 wood-receiver component, is connected to the current green Gui Shao N笊 her death to ° Hai antenna array to receive the source signal to the "mesh separation processor, is connected to the receiving cry, and? The source signal is at least 2 N differently mixed matrix: The target signal separation processing H can also be different depending on the source money: association or statistical independence, if not, then operate with the controller to form = receive m5#b new different sum Different beams are used to replace the non-f-related listener unique sources in the mixing matrix. The expected source signals are separated from the mixing matrix by the blind signal separation processor. In another embodiment, the receiver component may comprise N耙The receivers, each of the sinister receivers #include a different total of the multiplexed signals that can be received for the respective antenna elements connected thereto, and the different signal components can be selected. The blind signal separation processor can be connected to The N-type receiver is used to form a mixing matrix, and the U may include at least N different sums of at least N different demultiplexed signals. The combined matrix has an order equal to kN. 70 pieces of N antennas can be correlated Union. The N The associated antenna element can include an N active antenna element such that the antenna array can form a phase array. Alternatively, the N associated antenna element can include at least one active antenna element, and at most N-1 passive antenna elements enable the antenna array to form a switch The beam antenna is linear. The blind signal separation processor can separate the expected source signal from the mixing matrix based on at least one of principal component analysis, independent component analysis and single value decomposition. One enhancement relates to the ability to receive additional signals