JP2005039391A - Method and device for measuring reception pass phase of wireless communication - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a means of adding signals in phase in environment wherein only weak signals can be received, to enable position calculation using the weak signals, and to reduce a processing quantity which is a problem when a MUSIC method is applied. <P>SOLUTION: Provided is a method of putting a plurality of delay profiles together. According to the result of characteristic value analysis, selected is the MUSIC Method when separation into the plurality of delay profiles is possible or a pass detecting method based upon the shape of the delay profile when not. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

【００２０】
合成によって得られた遅延プロファイルＰｃ（τ）は元の遅延プロファイルに対して信号成分が同相で加算されるように位相回転補正項を含む複素重みで重み付け加算されていることから、信号対雑音電力比が改善される。よって、信号の受信タイミングを判定するパス検出の性能向上が可能となる。複素重みを得る方法についてはこの後述べる。
【００２１】
図１の手順を説明する。まず（ステップ１０１）で、端末は遅延プロファイルを作成する。遅延プロファイルの計算方法は、例えば特許文献１に記載されているが、ここでの特長は、複数のスナップショットを合成して合成された遅延プロファイルを得ることである。（ステップ１０１）の詳細な中身は図３に示されている。図３で、まず特定のＰＮオフセットの前後数チップに関する遅延プロファイルを作成する。これには、例えば、相関器の利用が考えられる。スライディング相関器であれば、特定の窓についての遅延プロファイル作成が容易に実施可能である。ここで作成する遅延プロファイルはＩ成分とＱ成分をもつ複素振幅の遅延プロファイルである。本発明では、複数のスナップショットを取得することを前提としており、それぞれのスナップショットに対して相関器を動作させて、複数の遅延プロファイルを作成する（ステップ２０１）。次に各スナップショット間の遅延プロファイルの相関を計算した相関行列を計算する（ステップ２０２）。計算方法は以下の式に示される。
【００２２】
【数２】

【００２３】
ここで、アスタリスク（＊）は複素共役演算を示している。ｅｉｇ（ Ｘ ）は（ステップ２０３）において実施する作業を示しており、行列Ｘの固有値（Ｖ）と固有ベクトル（Ｗ）を求める演算である。ここでは得られたスナップショット間の相関行列の固有値解析を行う。固有値と固有ベクトルは互いにペアーになっていて、固有ベクトルの複素共役が上記の合成重みとなり、固有値がその重みで抽出される成分の電力に相当する。一般には、固有値の大きいものを選択してその固有値に対応する固有ベクトルを選択して合成重みとし、数１に示す演算によって合成遅延プロファイルを計算すれば、特定の位置にピークがたったＳ／Ｎの良好な遅延プロファイルが得られるはずであるが、ＣＤＭＡの遠近問題が発生しているような場合、例えば該当する受信機が特定基地局付近にあり、その基地局からの信号が支配項となっているような場合には、逆拡散による拡散利得を得た後であっても、希望する基地局の受信信号電力が上記の特定基地局からの信号の受信電力以下となってしまい、干渉局である上記特定基地局の信号を同相で加算するような合成重みが選択され、希望する基地局の遅延プロファイルの主要なピークの検出することができない場合がある。
【００２４】
これに対する対策として、（ステップ１０２）を実行する。すなわち、全ての固有ベクトルを使って、それぞれに対応する複数の合成遅延プロファイルを作成し、そのピークと雑音レベルとの比が最も大きいものを選択する。ただし、この方法は多数の合成遅延プロファイルを作成することとなるので、例えば、固有値の大きな固有ベクトルから順にピークと雑音レベルとの比を求め、その比が閾値を超えるものが発見されたら、それを目的の合成遅延プロファイルとし、それ以降の固有ベクトルによる合成遅延プロファイルの作成をやめるといった処理の削減と組み合わせることが有効である。
【００２５】
最後に（ステップ１０３）を実行する。すなわち、得られた合成遅延プロファイルの最大ピークを中心に例えば±５チップ分のサンプルを切り取り、この部分について時間相関を作成すれば、上記の行列演算量を大幅に削減することができる。ＭＵＳＩＣ法の演算は以下の式で示される。まず、上記で切り出したＭ個のサンプルからなる部分集合｛τ｝に関する遅延プロファイルを数３に代入して時間相関行列を求め、これを固有値解析して固有値と固有ベクトルを計算する。
【００２６】
【数３】

【００２７】
固有値が閾値以上となる固有ベクトル（例えばＬ個あったとする）を除いて、残った固有ベクトルを数４に代入してΦを求める。このΦが一定以上の値となるときに、その位置にパスがあるとみなすことができる。
【００２８】
【数４】

【００２９】
以上の操作によって、受信信号が微弱であって、パス検出ができないような信号であっても、複数のスナップショットを利用したＭＵＳＩＣ法を次元数の低い行列演算によって実施可能となり、課題は解決される。
【００３０】
上記の例では、ＣＤＭＡのセルラ基地局からの信号を利用することを前提に本特許を説明したが、他のシステム、例えば無線ＬＡＮやデジタルセルラ方式においても、本特許を適用することができる。
【００３１】
また、図４に示すように雑音についても、相関行列演算を行い（ステップ２０６）、ここで得られた相関行列を上記（ステップ２０２）で得られた相関行列から差し引き、差し引いた相関行列を固有値解析して重みを求める（ステップ２０７）。その得られた重みを使って上記と同じ、ピークを検出する方法も本特許の範疇である。ここでの操作は以下の数５で示される。この操作によれば、（ステップ２０２）で得られた相関行列から雑音あるいは干渉による影響を低減し、固有値解析を行うことができる。よって課題は解決される。
【００３２】
【数５】

【００３３】
また、図５に示すように、同じく雑音による相関行列を用いて、その影響を取り除いた後に、重みを白色化することで、積極的に干渉波を低減することもできる。この操作も本発明の範疇である。演算式は数６に示される。この白色化を行った重みＷ’を用いるだけで、（ステップ２０８）の白色化が実施される。よって遠近問題による干渉波の影響を受けることなく、ピークの検出が可能となり、課題は解決される。
【００３４】
【数６】

【００３５】
スナップショットの取得方法については、例えば、図６に示す手順に基づいて、異なる時間に取得されたスナップショットのうち、時間的に近接するスナップショット間を回転補正して単純加算し、それを上記の合成前の遅延プロファイルとすることもできる。２つのスナップショット間で、相関を取り、相関値が高い場合には、この位相補正後の加算を行うことができる。位相補正は上記の相関演算で得られた複素相関の共役をかけることで実施できる。この操作により、スナップショットに関する相関行列演算の次元数を削減することができることから演算量を削減することができる。上記の実施例にこうした処理を加えることも本発明の範疇である。
【００３６】
また、上記実施例では、（ステップ１０２）において、全ての固有ベクトルを使って、それぞれに対応する複数の合成遅延プロファイルを作成し、そのピークと雑音レベルとの比が最も大きいものを選択することを示したが、このとき利用する式は数７で与えられる。
【００３７】
【数７】

【００３８】
但し、干渉電力が大きい場合に、雑音レベル（Ｎｏｉｓｅ）を求める際は注意が必要である。この雑音レベルには干渉レベルが含まれ、干渉レベルはスナップショット選択性があるため、上記重みについて有色となる。したがって、特定の重みの場合にその電力が変化してしまう。よって、数８によって有色時の雑音レベルを重み付けする必要がある。これによって、干渉レベルが大きい場合の問題も解決される。
【００３９】
【数８】

【００４０】
（２）環境への対応
図２を用いて、環境への対応について説明する。図２は、本発明からなる第２の実施例の流れを示すフローチャートである。（ステップ１０１）および（ステップ１０２）については、図１で示された第１の実施例と同じであるため、ここでは説明を省略する。続く（ステップ１０４）では、（ステップ１０２）で抽出された閾値以上のピークを持つ遅延プロファイルの数を数え、それが複数ある場合にのみ、使用するサンプル数を制限したＭＵＳＩＣ法を適用する（ステップ１０３）。ここで使用するＭＵＳＩＣ法は上記の例で紹介した方法と同じであるため、ここでは説明を省略する。ＭＵＳＩＣ法の出力Φは、閾値以上の値がないと、パス検出ができない。これは、特に信号レベルが弱い場合に発生すると考えられるが、こうした場合にはパスが失敗したとして、別のパス検出方法を実施する。ＭＵＳＩＣ法によるパス検出が成功したならば、それを最終出力としてパス検出を終了する（ステップ１０５）。（ステップ１０５）で失敗したと判断された場合、あるいは（ステップ１０４）で閾値以上のピークを持つ遅延プロファイルが１つしかない場合には、別のパス検出方法を利用する。この方法に先立ち、上記の複素の合成遅延プロファイルを電力プロファイルに変換する必要がある。電力化はＩおよびＱ成分の信号をＸＩ、ＸＱとすると、Ｐ＝ＸＩ＊ＸＩ＋ＸＱ＊ＸＱで得られる。次に上記で得られた電力の合成遅延プロファイルの形状を利用したパス検出を行う。例えば、単純に遅延プロファイルの前方から順にサーチしていき、閾値以上の値を持つサンプルとする方法や、あるいは前方からサーチしていき、閾値以上で、最初のピークを検出する方法、あるいは、特許文献２で開示されている遅延プロファイルの形状に注目してパス検出を行う方法が利用できる。
【００４１】
本発明の特長は、（ステップ１０４）において、遅延プロファイルの解析結果を分析し、準定常環境とフェージング環境とを判断して、それぞれにあったパス検出方法を採用することである。すなわち、スナップショット間の相関が高く、ピークを持つ遅延プロファイルが複数作成できない場合には、準定常状態であると判断して、スナップショット間の相関からピークを強調する重みで遅延プロファイルを合成し、合成された遅延プロファイルの形状に注目してパス検出を行う方法を選択する。他方、スナップショット間の相関が低く、ピークを持つ遅延プロファイルが複数作成できた場合には、フェージングが発生していると判断し、ピーク付近のサンプルを取り出してＭＵＳＩＣ法を実施するパス検出方法を選択する。これにより、どちらの環境においても、パス検出が精度よく実施することが可能となる。よって課題は解決される。
【００４２】
（３）端末
図７を用いて、本発明からなる第３の実施例を説明する。図７は、本発明を実施する端末の構成を示す図である。
【００４３】
図で、電波を受信したアンテナ１からの信号はＲＦ部２において信号変換されベースバンド信号に変換される。その信号は一旦メモリー７に蓄積される。メモリー７に蓄積されたスナップショットは、適宜、相関器６に転送され、ここで相関演算が行われて遅延プロファイルが作成される。作成された遅延プロファイルはメモリー７に蓄積される。蓄積された遅延プロファイルに対してＣＰＵ８は信号処理演算を行い、上記で述べてきた信号処理を実施して遅延プロファイルの合成を行う。また、パス検出を行う。位置計算に関しては、本ＣＰＵ８で行っても良いし、ここには記述されていない無線回線上に配置されたサーバー装置にパス検出結果を送り、そこで実施してもよい。本発明のポイントは遅延プロファイルの合成方法とパス検出の方法であるので、いずれの方法であっても遅延プロファイルに上記の方法の遅延プロファイルの合成方法が使用される場合は、本発明の範疇である。ここに本発明のポイントとなる手順および式が記されたプログラム４００はメモリー４１０内に格納されている。本発明の実施形態には、様々なものが考えられる。たとえば図で示す装置をディスクリートの部品で構成する場合はもちろんのこと、上記プログラムを格納するメモリー４１０を具備する端末装置も本発明の実施形態となる。また、本発明を含むメモリー４１０自身も本発明の実施する形態の一つである。また、例えば図内で破線５００や６００で示したようなブロックを含む集積回路をもつ装置や集積回路そのものも本発明の実施形態の１つである。
【００４４】
【発明の効果】
本発明によれは、微弱な信号しか受信できない環境においても、信号を同相で加算する手段を与え、上記の微弱信号を用いた位置計算が可能となる。また、ＭＵＳＩＣ法を適用する際に課題となる処理量を削減する。さらにはフェージングや準定常環境に応じて適用なパス検出方法を選択してパス検出を行うので、精度のよいパス検出を行うことができる。
【図面の簡単な説明】
【図１】本発明の第１の実施例のフローチャート。
【図２】本発明の第２の実施例のフローチャート。
【図３】本発明の遅延プロファイル合成のフローチャート。
【図４】本発明の遅延プロファイル合成のフローチャート。
【図５】本発明の遅延プロファイル合成のフローチャート。
【図６】本発明からなる遅延プロファイル合成のフローチャート。
【図７】端末装置の構成。
【符号の説明】
１．．．アンテナ、２．．．ＲＦ部、３．．．ベースバンド装置、４．．．制御チャネル用の逆拡散器、５．．．受信機、６．．．パイロット信号用の相関器、７．．．メモリー装置、８．．．ＣＰＵ、１０１、１０２、１０３、１０４、１０５、１０６、２０１、２０２、２０３、２０４、２０５、２０６、２０７、２０８、３０１、３０２．．．信号処理のステップ、４００．．．信号処理プログラム、４１０．．．メモリー装置、５００．．．メモリーを含むベースバンド部、６００．．．ＲＦ部を含む無線器ユニット。[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a path detection method for a wireless position measuring device that receives a signal transmitted from a wireless radio wave source and measures the position of a terminal. In particular, it describes a location calculation method for a system that identifies the location of a terminal using a signal from a cellular base station.
[0002]
[Prior art]
Conventionally, the principle of trilateral surveying in which the position of a terminal is detected using a signal from a radio wave source is disclosed in Patent Document 1. The configuration of the terminal will be described using IS-95 as an example. The terminal includes an antenna, an RF device, a baseband device, a memory device, and a CPU. The signal received by the antenna is converted into a baseband signal by the RF unit. Furthermore, a received signal for a certain period is accumulated in the memory device. The accumulated received data is called a snapshot.
[0003]
A procedure for performing position measurement will be described. An IS-95 base station transmits a pilot signal which is a fixed pattern. The transmission timing of each signal is transmitted later than the system time based on the PN offset that the base station has individually. In the CDMA system, all base stations transmit signals in the same frequency band. Therefore, the terminal can receive a plurality of base station signals at the same time by receiving signals in this frequency band, and store the received signals for a predetermined time as snapshots. In other words, it shows that the snapshots stored in the above-mentioned memory device are added and stored with the signals of all receivable base stations. Since signals transmitted from multiple base stations interfere with each other, when extracting a specific base station signal, it is necessary to despread the received signal with a spread signal tuned to the PN offset of the pilot signal of the specific base station There is.
[0004]
When powered on, the terminal needs to synchronize with the PN offset of the nearest base station. For this reason, the terminal first operates the pilot channel correlator, and sequentially searches for all PN offsets while sequentially changing the PN offsets of the pilot signals to be correlated, and finds the PN offset with the largest correlation peak. look for. The detected maximum peak PN offset indicates the synchronization timing to the base station considered to be closest. In the baseband apparatus, a despreader for the control channel is provided, and a despreading operation is performed with the PN offset of the nearest base station detected above, and a control channel signal is taken out. Further, the extracted control channel signal is detected by a receiver and demodulated into significant information. The CPU extracts the ID of the received base station from the extracted detection output information, searches the information table of the surrounding base stations stored in advance in the memory, and identifies the surrounding base stations to be observed. Take out the PN offset. A complex delay profile is created using a pilot signal correlator for a certain width (window) before and after the PN offset of the extracted nearest base station and surrounding base stations. Further, complex information is converted into a power delay profile by a product-sum operation. The created power delay profile is stored in the memory device. The CPU analyzes the delay profile stored in the memory device and takes out the timing when the path is detected. The CPU calculates the position of the terminal using a solution such as a least square method.
[0005]
[Patent Document 1] Japanese Patent Laid-Open No. 7-181242
[Patent Document 2] Japanese Unexamined Patent Publication No. 2000-197863
[Problems to be solved by the invention]
In a wireless position measurement device, for example, when it is desired to receive a signal in an environment where the radio wave is weak, such as indoors, it is effective to extract the weak signal using long-term received data. At this time, the handling method differs depending on the environment. The environment can be a fading environment or a quasi-stationary environment. The fading environment is generated when the surrounding object moves or changes, or when the wireless position measuring device itself moves, and the propagation path changes within the data collection time. On the other hand, the quasi-steady state indicates a case where the fluctuation of the propagation path is slow and almost negligible. The acquisition of the data is intermittently performed in burst units called snapshots. In the fading state, the propagation path state fluctuates between the acquired snapshots. The MUSIC method is known as a method for performing path detection in such an environment. In the MUSIC method, the received snapshot is despread, a correlation matrix in the time direction is created from the obtained delay profile, and eigenvalue analysis is performed to perform path detection. The MUSIC method is a powerful tool when the propagation environment changes between the above snapshots, but there are some problems.
[0006]
The first problem is the amount of processing. For example, when a CDMA signal is received, if the base station signal is weak with respect to the total received power, it cannot be determined where the signal is generated. It is necessary to evaluate the whole by the MUSIC method.
[0007]
The second problem is performance in a quasi-steady state. In the MUSIC method, when acquired snapshots are highly correlated with each other, the rank of the obtained correlation matrix is low, and the ability to separate multipaths is significantly reduced. In such a case, it is rather considered that the method of performing path detection paying attention to the shape of the delay profile disclosed in Patent Document 2, for example, is more effective.
[0008]
As shown above, there has been path detection like the conventional MUSIC method or path detection as shown in Patent Document 2, but each has different characteristics, so these are fused according to the environment. The problem is that there was no usage.
[0009]
[Means for Solving the Problems]
The above problem is to perform a correlation operation with a specific signal for at least two snapshots, create a delay profile corresponding to each snapshot, and obtain a cross correlation matrix of the obtained delay profiles. Step 1 of calculating a plurality of sets of weights from the eigenvalue analysis and generating a plurality of combined delay profiles by weighted combining the plurality of delay profiles with the weights, and noise for the plurality of combined delay profiles Step 2 for extracting a delay profile having a peak higher than the threshold value from the level, and extracting only a part of samples including the peak higher than the threshold value of the delay profile extracted in Step 2 above, applying to the MUSIC algorithm, This is solved by the path detection method comprising step 3 of detecting the.
[0010]
Alternatively, the above problem is to perform a correlation operation with a specific signal for at least two snapshots, create a delay profile corresponding to each snapshot, and obtain a cross-correlation matrix of the obtained plurality of delay profiles. And calculating a plurality of sets of weights from the eigenvalue analysis, creating a plurality of combined delay profiles by weighted combining the plurality of delay profiles with the weights, and for the plurality of combined delay profiles Step 2 for extracting a delay profile having a peak equal to or higher than a threshold value from the noise level, Step 4 for determining the number of delay profiles extracted in Step 2 above, and extraction in Step 2 when there are a plurality of determinations in Step 4 Extract only some samples that contain peaks above the above threshold of the delayed profile Step 3 for detecting the path phase by applying the MUSIC algorithm, Step 5 for determining whether the path detection has failed in Step 3, and in the case of failure in Step 5 or if the determination in Step 4 is singular. This is solved by the second path detection method that implements the path detection method that detects the rising edge.
[0011]
Or the said subject is the said 2nd path | pass detection method, Comprising: When the said step 1 performs the correlation calculation with the said specific signal, in addition to the said step 1, the signal by which a correlation does not generate | occur | produce Generating a noise profile using the sequence, further obtaining a cross-correlation matrix of the noise profile, obtaining a difference of the cross-correlation matrix of the noise profile from the cross-correlation matrix of the delay profile, and further calculating the correlation This is solved by a third path detection method including step 8 of performing eigenvalue analysis on the matrix difference and calculating the weight.
[0012]
Alternatively, the problem is the third path detection method, wherein the step 8 performs eigenvalue analysis on a correlation matrix having the noise profile as a denominator and a numerator being a difference of the correlation matrix, This is solved by a fourth path detection method characterized by including step 9 for calculating the weight.
[0013]
Alternatively, the problem is the first path detection method, in which the step 1 selects a plurality of delay profiles having close observation time intervals from the obtained plurality of delay profiles, and performs phase comparison on the selected delay profiles. The present invention is solved by a fifth path detection method characterized by including a step 10 of performing in-phase addition processing for rotational adjustment and addition.
[0014]
Alternatively, the problem is the first path detection method, wherein the step 2 performs weighted addition of a delay profile using an eigenvector for one or more eigenvalues and eigenvectors generated by eigenvalue analysis, The seventh path detection method is characterized in that it includes a step 11 of selecting an eigenvector having a maximum ratio between the maximum peak and the average noise level.
[0015]
DETAILED DESCRIPTION OF THE INVENTION
A first embodiment of the present invention will be described with reference to FIG. FIG. 1 is a flowchart showing the flow of a first embodiment according to the present invention. As already described in the prior art, in the case of the CDMA system, the received signal includes signals from a plurality of base stations, but here, in order to simplify the explanation, attention is paid to one base station. To explain. The signal processing method is the same for other base stations, and only the spreading code used for despreading is different.
[0016]
The present invention has the following two features. The first feature is a reduction in the processing amount of the MUSIC method, and the second feature is a response to the environment.
[0017]
(1) Reduction of treatment amount of MUSIC method
When a weak signal is received, even if it is despread, in the delay profile obtained from a single snapshot, the value where the peak appears is buried in the background noise generated by noise and interference, and path detection is not possible. Is possible. It is also difficult to know which part of the created delay profile has a signal. Therefore, when the MUSIC method is applied to such data, it is necessary to create a correlation matrix by obtaining time correlation for all created profiles. However, for example, when a chip rate is a signal of 1 MHz and a hexagonal cell is arranged with a base station interval of about 3 km, an observation interval of about ± 32 chips may be necessary if a plurality of base stations are to be observed. When the sample rate of the observation signal is set to twice the chip rate, the observation window size of the delay profile reaches 128 samples. When the time correlation of this signal is obtained, the dimension of the matrix is considered to be about 100. In general, in order to perform eigenvalue analysis of a matrix, an operation proportional to the cube of the dimension of the matrix is required. In the above case, a huge amount of calculation is required. Therefore, as an effective method, it is conceivable to extract the vicinity of a sample having a high signal correlation as a main term and apply the MUSIC method to the sample. However, as described above, in the delay profile obtained from one snapshot, it is difficult to determine where the path is. Therefore, in the present invention, a plurality of received data (snapshots) are acquired, a delay profile corresponding to each snapshot is created, and a composite delay profile obtained by synthesizing them is created, thereby reducing the S / N of the signal. Improvement is performed, a peak search is performed on the combined delay profile, a main part having a high correlation is detected, the part is cut out, and a path search by the MUSIC method is performed on the cut out delay profile. Thereby, the processing amount can be reduced. A specific method will be described below.
[0018]
First, snapshots of received signals acquired at different times are acquired. This is despread to obtain a delay profile. The resulting delay profile can be written as Pn (τ). Here, τ indicates a delay time, and n is an index of a snapshot corresponding to the delay profile. The delay time is a discrete sample, and is usually set to a value that is twice or more the chip rate. The delay profile is weighted and synthesized with complex weights. If written in an expression, it is expressed by the following formula 1.
[0019]
[Expression 1]

[0020]
The delay profile Pc (τ) obtained by the synthesis is weighted and added with a complex weight including a phase rotation correction term so that signal components are added in phase with the original delay profile. The ratio is improved. Therefore, it is possible to improve the path detection performance for determining the signal reception timing. A method for obtaining the complex weight will be described later.
[0021]
The procedure of FIG. 1 will be described. First (in step 101), the terminal creates a delay profile. The delay profile calculation method is described in, for example, Patent Document 1, and the feature here is that a composite delay profile is obtained by combining a plurality of snapshots. The detailed contents of (Step 101) are shown in FIG. In FIG. 3, first, a delay profile for several chips before and after a specific PN offset is created. For example, the use of a correlator can be considered. With a sliding correlator, it is possible to easily create a delay profile for a specific window. The delay profile created here is a complex amplitude delay profile having an I component and a Q component. In the present invention, it is assumed that a plurality of snapshots are acquired, and a correlator is operated for each snapshot to create a plurality of delay profiles (step 201). Next, a correlation matrix that calculates the correlation of the delay profile between the snapshots is calculated (step 202). The calculation method is shown in the following formula.
[0022]
[Expression 2]

[0023]
Here, an asterisk (*) indicates a complex conjugate operation. eig (X) indicates the work performed in (step 203), and is an operation for obtaining the eigenvalue (V) and eigenvector (W) of the matrix X. Here, eigenvalue analysis of the correlation matrix between the obtained snapshots is performed. The eigenvalue and the eigenvector are paired with each other, and the complex conjugate of the eigenvector becomes the above-described composite weight, and the eigenvalue corresponds to the power of the component extracted with the weight. In general, if an eigenvector corresponding to the eigenvalue is selected by selecting the one having a large eigenvalue and is set as a synthesis weight, and a synthesis delay profile is calculated by the calculation shown in Equation 1, the S / N having a peak at a specific position is calculated. A good delay profile should be obtained, but when a CDMA perspective problem occurs, for example, the corresponding receiver is near a specific base station, and the signal from that base station becomes the dominant term. In such a case, even after obtaining the spreading gain due to despreading, the received signal power of the desired base station is less than or equal to the received power of the signal from the specific base station. In some cases, a combination weight is selected such that signals from a specific base station are added in phase, and a main peak of a desired base station delay profile cannot be detected.
[0024]
As a countermeasure against this, (step 102) is executed. That is, a plurality of composite delay profiles corresponding to each eigenvector are created using all eigenvectors, and the one having the largest ratio between the peak and the noise level is selected. However, since this method creates a large number of composite delay profiles, for example, the ratio between the peak and the noise level is calculated in order from the eigenvector having the largest eigenvalue, and if the ratio exceeds the threshold value, it is found. It is effective to combine it with the reduction of processing such as making the desired synthesis delay profile and stopping the creation of the synthesis delay profile using the eigenvectors thereafter.
[0025]
Finally (step 103) is executed. That is, if, for example, samples of ± 5 chips are cut around the maximum peak of the obtained composite delay profile and a time correlation is created for this portion, the above matrix calculation amount can be greatly reduced. The operation of the MUSIC method is expressed by the following equation. First, the time correlation matrix is obtained by substituting the delay profile for the subset {τ} consisting of the M samples cut out above into Equation 3, and this is subjected to eigenvalue analysis to calculate eigenvalues and eigenvectors.
[0026]
[Equation 3]

[0027]
Φ is obtained by substituting the remaining eigenvectors into Equation 4 except for eigenvectors having eigenvalues equal to or greater than the threshold (for example, L). When this Φ is a certain value or more, it can be considered that there is a path at that position.
[0028]
[Expression 4]

[0029]
By the above operation, even if the received signal is weak and the signal cannot be detected, the MUSIC method using a plurality of snapshots can be performed by a matrix operation with a low number of dimensions, and the problem is solved. The
[0030]
In the above example, the present patent has been described on the assumption that a signal from a CDMA cellular base station is used. However, the present patent can also be applied to other systems such as a wireless LAN and a digital cellular system.
[0031]
Further, as shown in FIG. 4, the correlation matrix calculation is also performed for noise (step 206), the correlation matrix obtained here is subtracted from the correlation matrix obtained in the above (step 202), and the subtracted correlation matrix is the eigenvalue. The weight is obtained by analysis (step 207). The method of detecting the same peak as described above using the obtained weight is also within the scope of this patent. The operation here is expressed by the following equation (5). According to this operation, the influence of noise or interference can be reduced from the correlation matrix obtained in (step 202), and eigenvalue analysis can be performed. Thus, the problem is solved.
[0032]
[Equation 5]

[0033]
Further, as shown in FIG. 5, the interference wave can also be actively reduced by whitening the weights after removing the influence using the correlation matrix due to noise. This operation is also within the scope of the present invention. The arithmetic expression is shown in Equation 6. The whitening of (Step 208) is performed only by using the weight W ′ subjected to the whitening. Therefore, the peak can be detected without being affected by the interference wave due to the perspective problem, and the problem is solved.
[0034]
[Formula 6]

[0035]
Regarding the snapshot acquisition method, for example, based on the procedure shown in FIG. 6, among snapshots acquired at different times, rotation correction is performed between snapshots that are close in time, and this is simply added. It is also possible to use a delay profile before synthesis. When the correlation is taken between the two snapshots and the correlation value is high, the addition after the phase correction can be performed. Phase correction can be performed by applying the conjugate of the complex correlation obtained by the above correlation calculation. This operation can reduce the number of dimensions of the correlation matrix calculation related to the snapshot, thereby reducing the calculation amount. It is also within the scope of the present invention to add such processing to the above embodiment.
[0036]
In the above embodiment, in (step 102), using all the eigenvectors, a plurality of combined delay profiles corresponding to each are created, and the one having the largest ratio between the peak and the noise level is selected. As shown, the formula used at this time is given by Equation 7.
[0037]
[Expression 7]

[0038]
However, when the interference power is large, care must be taken when obtaining the noise level (Noise). This noise level includes an interference level, and since the interference level has snapshot selectivity, the weight is colored. Therefore, the power changes in the case of a specific weight. Therefore, it is necessary to weight the noise level at the time of coloring by the equation (8). This also solves the problem when the interference level is high.
[0039]
[Equation 8]

[0040]
(2) Response to the environment
The response to the environment will be described with reference to FIG. FIG. 2 is a flowchart showing the flow of the second embodiment according to the present invention. Since (Step 101) and (Step 102) are the same as those of the first embodiment shown in FIG. 1, the description thereof is omitted here. In the subsequent (step 104), the number of delay profiles having peaks equal to or greater than the threshold extracted in (step 102) is counted, and the MUSIC method in which the number of samples to be used is limited is applied only when there are a plurality of delay profiles (step 104). 103). Since the MUSIC method used here is the same as the method introduced in the above example, the description is omitted here. If the output Φ of the MUSIC method does not have a value equal to or greater than the threshold, path detection cannot be performed. This is considered to occur particularly when the signal level is weak. In such a case, it is assumed that the path has failed, and another path detection method is performed. If the path detection by the MUSIC method is successful, the path detection is terminated using this as the final output (step 105). If it is determined that the process has failed in (Step 105), or if there is only one delay profile having a peak equal to or greater than the threshold value in (Step 104), another path detection method is used. Prior to this method, it is necessary to convert the complex composite delay profile to a power profile. Electric power can be obtained by P = XI * XI + XQ * XQ where I and Q component signals are XI and XQ. Next, path detection using the shape of the combined delay profile of power obtained above is performed. For example, simply search from the front of the delay profile in order and use a sample with a value above the threshold, or search from the front and detect the first peak above the threshold, or patent A method of performing path detection by paying attention to the shape of the delay profile disclosed in Document 2 can be used.
[0041]
The feature of the present invention is that, in (Step 104), the analysis result of the delay profile is analyzed, the quasi-stationary environment and the fading environment are judged, and the path detection method suitable for each is adopted. In other words, if there is a high correlation between snapshots and multiple delay profiles with peaks cannot be created, it is determined to be in a quasi-steady state, and a delay profile is synthesized with a weight that emphasizes the peaks from the correlation between snapshots. Then, a method for performing path detection is selected by paying attention to the shape of the synthesized delay profile. On the other hand, when the correlation between snapshots is low and a plurality of delay profiles having peaks can be created, it is determined that fading has occurred, and a path detection method is performed in which samples near the peaks are extracted and the MUSIC method is performed. select. This makes it possible to accurately perform path detection in either environment. Thus, the problem is solved.
[0042]
(3) Terminal
A third embodiment according to the present invention will be described with reference to FIG. FIG. 7 is a diagram showing a configuration of a terminal that implements the present invention.
[0043]
In the figure, the signal from the antenna 1 that has received the radio wave is converted into a baseband signal by the RF unit 2. The signal is temporarily stored in the memory 7. The snapshot stored in the memory 7 is appropriately transferred to the correlator 6, where a correlation calculation is performed to create a delay profile. The created delay profile is stored in the memory 7. The CPU 8 performs a signal processing operation on the accumulated delay profile, and performs the signal processing described above to synthesize the delay profile. Also, path detection is performed. The position calculation may be performed by the CPU 8, or the path detection result may be sent to a server device arranged on a wireless line not described here and executed there. Since the point of the present invention is the delay profile synthesis method and the path detection method, the delay profile synthesis method of the above method is used for the delay profile in any of the methods. is there. A program 400 in which the procedure and formulas that are the points of the present invention are written is stored in the memory 410. Various embodiments of the present invention are conceivable. For example, in the case where the device shown in the figure is configured with discrete components, a terminal device including the memory 410 for storing the program is also an embodiment of the present invention. Further, the memory 410 itself including the present invention is also one embodiment of the present invention. Further, for example, an apparatus having an integrated circuit including a block as indicated by broken lines 500 and 600 in the figure and the integrated circuit itself are one embodiment of the present invention.
[0044]
【The invention's effect】
According to the present invention, even in an environment where only weak signals can be received, a means for adding the signals in phase is provided, and position calculation using the above weak signals becomes possible. In addition, the amount of processing that becomes a problem when the MUSIC method is applied is reduced. Furthermore, since path detection is performed by selecting an appropriate path detection method according to fading or a quasi-stationary environment, path detection with high accuracy can be performed.
[Brief description of the drawings]
FIG. 1 is a flowchart of a first embodiment of the present invention.
FIG. 2 is a flowchart of a second embodiment of the present invention.
FIG. 3 is a flowchart of delay profile synthesis according to the present invention.
FIG. 4 is a flowchart of delay profile synthesis according to the present invention.
FIG. 5 is a flowchart of delay profile synthesis according to the present invention.
FIG. 6 is a flowchart of delay profile synthesis according to the present invention.
FIG. 7 shows a configuration of a terminal device.
[Explanation of symbols]
1. . . Antenna, 2. . . RF section, 3. . . Baseband device, 4. . . 4. despreader for control channel; . . 5. receiver, . . 6. Correlator for pilot signal, . . Memory device, 8. . . CPU, 101, 102, 103, 104, 105, 106, 201, 202, 203, 204, 205, 206, 207, 208, 301, 302. . . Signal processing step, 400. . . Signal processing program, 410. . . Memory device, 500. . . Baseband section including memory, 600. . . Radio unit including RF section.

Claims (9)

A path detection method for storing a received signal from a radio base station as a snapshot and detecting a path of the received signal,
At least two or more times of the above snapshots are correlated with a specific signal, a delay profile corresponding to each snapshot is created, a cross correlation matrix of the obtained delay profiles is obtained, and its eigenvalue Step 1 of calculating a plurality of sets of weights from the analysis and creating a plurality of combined delay profiles by weighted combining the plurality of delay profiles with the weights, and a threshold value based on a noise level for the plurality of combined delay profiles Step 2 for extracting the delay profile having the above peaks, and extracting only a part of the samples including the peaks equal to or higher than the threshold of the delay profile extracted in Step 2 above, and applying the MUSIC algorithm to detect the path phase. A path detection method comprising step 3. A path detection method for storing a received signal from a radio base station as a snapshot and detecting a path of the received signal,
At least two or more times of the above snapshots are correlated with a specific signal, a delay profile corresponding to each snapshot is created, a cross correlation matrix of the obtained delay profiles is obtained, and its eigenvalue Step 1 of calculating a plurality of sets of weights from the analysis and creating a plurality of combined delay profiles by weighted combining the plurality of delay profiles with the weights, and a threshold value based on a noise level for the plurality of combined delay profiles Step 2 for extracting a delay profile having the above peaks, Step 4 for determining the number of delay profiles extracted in Step 2 above, and the determination of the delay profile extracted in Step 2 above when there are a plurality of determinations in Step 4 Extract only some samples that contain peaks above the threshold of MUSIC algorithm The path rises when the path phase is detected in Step 3, Step 5 for determining whether the path detection has failed in Step 3, and in the case of failure in Step 5, or when the determination in Step 4 is singular. A path detection method for implementing a path detection method for detecting a path. 3. The path detection method according to claim 1, wherein the step 1 uses a signal sequence that does not generate a correlation with the transmission signal in addition to the step 1 when performing a correlation calculation with the specific signal. Generating a noise profile, and obtaining a cross-correlation matrix of the noise profile, obtaining a difference of the cross-correlation matrix of the noise profile from the cross-correlation matrix of the delay profile, and further obtaining a difference of the correlation matrix A path detection method comprising the step 8 of performing eigenvalue analysis on and calculating the weight. 4. The path detection method according to claim 3, wherein the step 8 performs eigenvalue analysis on a correlation matrix having the noise profile as a denominator and a numerator being a difference of the correlation matrix to calculate the weight. A path detection method including step 9. 3. The path detection method according to claim 1 or 2, wherein the step 1 selects a plurality of delay profiles having close observation time intervals from the obtained plurality of delay profiles and performs phase rotation adjustment on the selected delay profiles. A path detection method comprising: a step 10 for performing in-phase addition processing for adding together. 3. The path detection method according to claim 1, wherein the step 2 performs weighted addition of a delay profile using an eigenvector for one or more eigenvalues and eigenvectors generated by eigenvalue analysis, and the maximum peak thereof. And a step 11 for selecting an eigenvector that maximizes the ratio of the average noise level. A memory device storing a program for causing a CPU to execute all or part of the signal processing according to claim 1. 3. An integrated circuit having a built-in memory device in which a program for causing a CPU to execute all or part of the signal processing according to claim 1 or 2 is stored. A wireless position measuring device including a memory device in which a program for causing a CPU to execute all or part of the signal processing according to claim 1 is stored.

Images