KR100860579B1 - Frequency recognition method for wireless cognitive terminal - Google Patents

Abstract

A frequency recognition method for a wireless recognition terminal is disclosed.

에 대해 추정된 CIR 벡터

를 구하여, 결합된 CIR 벡터를 생성하는 단계와; B) p번째 경로가 가장 강한 경로라 할 때, P개의 가장 강한 경로들에 대한 추정값을 결정하고, 추정값들을 이용하여 추정된 채널 진폭 벡터를 형성하는 단계와; C) 상기 B) 단계의 결과를 이용하여 1부터 P까지의 p에 대해 지연값

를 결정하는 단계; 를 포함하여 이루어짐에 기술적 특징이 있다.The present invention provides a tap amplitude and delay estimation method for frequency recognition in a wireless recognition terminal requiring a frequency or network recognition capability for recognizing an available wireless service and system. A) It is assumed that the number of samples per chip period is N _S. For all n _S from 1 to N _S ,

Estimated CIR Vectors for

Obtaining a combined CIR vector; B) when the p-th path is the strongest path, determining an estimate for the P strongest paths and using the estimates to form an estimated channel amplitude vector; C) a delay value for p from 1 to P using the result of step B) above

Determining; There is a technical feature made to include.

Description

Frequency Sensing Method for Cognitive Radio Terminal

1 is a conceptual diagram of a downlink system model for explaining the present invention.

<Description of Symbols for Main Parts of Drawings>

10: transmit RRC filter 20: mobile wireless channel

30: Adder 40: Receive RRC Filter

50: AD Converter 60: Tap Amplitude and Delay Estimator

The present invention relates to a wireless recognition technology, and to an efficient frequency recognition method that can be applied to a wireless recognition terminal requiring a frequency or network recognition capability for the recognition of available wireless services and systems.

In the ubiquitous age using not only mobile communication and digital multimedia broadcasting (DMB) but also ubiquitous sensor network (USN), telematics, home network, etc. Changes in use are required.

Therefore, research has been conducted on a new concept of a system called flexible use as a method of increasing the efficiency of radio waves exclusively used. As a result, a frequency is allocated but is not actually used and is detected as an empty frequency. The concept of cognitive radio (CR) has been proposed so that it can be effectively shared and used.Wireless cognitive technology can make use of the spectrum much more flexibly to meet the diverse needs of commercial, military and public customers. It is expected to be able to satisfy all.

On the other hand, the radio recognition technology is based on SDR (Software Defined Radio: technology that can implement a variety of communication standards in one terminal) to meet the changing communication protocol. SDR is a technology that changes the method of supporting high frequency (RF) from hardware in base station and terminal to hardware. It is a service in a multi-mode, multi-band, multi-environment wireless communication environment with one terminal regardless of place and time. Means the technology to provide wireless devices and services by the operation of software to provide economically.

On the other hand, since a personal wireless device, i.e., a wireless cognitive terminal for applying such a wireless cognitive technology, must be operated in a multi standard / multi mode, the wireless cognitive terminal should search for all frequency bands that can be allocated to the system and mode. Available standards and modes should be recognized.

However, current personal wireless devices are basically implemented to operate in a single standard single mode, which mode of operation is based on the search for beacon signals in relation to a particular standard. That is, since the signal search of the existing personal wireless device is performed only in the available mode or does not consider all the multiple modes, a new frequency search method for applying the radio recognition technology is required.

In addition, since the frequency recognition or network recognition is a very important issue and a core technology in the wireless cognitive terminal, the operation for the frequency recognition or the network recognition in the wireless cognitive terminal can be performed quickly and efficiently at low power. There is a need for a frequency recognition method.

In addition, future terminals, that is, radio-aware terminals, require support of multiple modes, multiple standards, and at the same time, low power consumption and small size should be the basis, and these require reconfigurable and programmable hardware platforms. The most flexible method for vision can be the gradual application of software radio technology, or Software Defined Radio (SDR) technology.

However, the current semiconductor technology does not provide a platform for signal processing on a digital processing processor (DSP) and a central processing unit of a reduced instruction set computer (RISC) type. The realization of the treatment is the most important.

Meanwhile, a rake receiver is a receiver capable of separating two signals having a time difference (delay) from each other, and requires a tape amplitude and delay estimator (TAD) technique for estimating the amplitude and delay of a tap. . However, the existing TADE algorithm is very complex and difficult to implement in a RISC processor or a DSP. Accordingly, a new TADE algorithm is required, which can be implemented based on software or firmware on the DSP.

Accordingly, an object of the present invention is to provide an iterative frequency recognition method for efficiently utilizing radio resources in a multi-standard / multi-mode environment and implementing a seamless service.

In particular, the present invention provides an efficient frequency recognition method for a wireless cognitive terminal that can have excellent performance while reducing the complexity of implementation of a wireless cognitive terminal requiring a frequency or network recognition capability for available services and system recognition. The purpose.

In addition, another object of the present invention is to provide a software or firmware based tap amplitude and delay estimation (TADE) algorithm.

에 대해 추정된 CIR(채널 임펄스 응답) 벡터를 가리키는

를 구하여, 결합된 CIR 벡터

를 생성하는 단계와; B)

를 상기 A)단계에서 추정된 CIR 벡터의 인덱스라 하고, p번째 경로가 가장 강한 경로라 할 때, P개의 가장 강한 경로들에 대한 추정값

을 결정하고, 추정값들을 이용하여 추정된 채널 진폭 벡터

를 형성하는 단계와; C)

에 근거하여 1부터 P까지의 p에 대해 지연값

를 결정하는 단계; 를 포함한다.In order to achieve the above object, the present invention provides a tap amplitude and delay estimation method for efficient repetitive frequency recognition in a wireless recognition terminal requiring frequency or network recognition capability for recognizing available wireless services and systems. Assuming N _S samples per period, for all n _S from 1 to N _S ,

Pointer to the estimated channel impulse response (CIR) vector for

To obtain the combined CIR vector

Generating a; B)

Is the index of the CIR vector estimated in step A), and when the p-th path is the strongest path, the estimated values for the P strongest paths.

And estimate the channel amplitude vector using the estimates.

Forming a; C)

Delay value for p from 1 to P based on

Determining; It includes.

Details of the above object and technical configuration of the present invention and the resulting effects thereof will be more clearly understood from the following detailed description based on the accompanying drawings.

The user equipment for the radio recognition application, i.e., the radio recognition terminal, is a multi-standard / multi-mode, in which radio frequency terminals must be searched for all frequency bands potentially assigned to the systems and modes to which the terminal can connect, This should identify possible standards and modes.

The frequency recognition method proposed by the present invention is performed on a subscriber basis, and this operation step is included in a radio recognition operation of the master terminal controller, that is, an operation of recognizing an available frequency or network.

In addition, the frequency recognition algorithm of the present invention may be performed in the analog domain center, the digital domain center, or the mixed mode center, and may be implemented in a serial version or a parallel version.

In the present invention, a description will be given on the assumption that the wireless recognition terminal is provided with only one receiving antenna, and thus the effect of the antenna arrangement for the execution of the TADE is not discussed in the present invention.

In the present invention, the baseband signal will be represented by a complex value. In the following, '*' refers to continuous-time convolution. Complex-valued variables will be underlined, and '(.) ^* ' Represents the logical product of a complex number.

Also, (.) ^T denotes a matrix or vector binomial, (.) ^H denotes a Hermitian matrix binomial, a vector denotes lowercase italic bold, and a matrix denotes uppercase italic bold.

First, the mathematical system model of the present invention will be described.

로부터의 세팅을 통해 만들어질 수 있다.In a WCDMA downlink system, TADE may be based on a Primary Synchronization Code (PSC) that deploys a unique CDMA code based on a generalized hierarchical Golay Sequence, which is hierarchical Golay. Sequence is a 16-bit subsequence

Can be made via settings from.

,

인 바이너리 칩(c_q)들로 구성되는 실수값의 PSC 시퀀스는 다음과 같이 정의된다.

,

The PSC sequence of the real value composed of the binary chips c _q is defined as follows.

와 곱해져, 다음과 같은 복소수 값의 PSC 시퀀스 c 를 형성한다. 여기서,

이다.Prior to transmission, the PSC sequence c defined as in Equation 2 is

And multiply by to form a complex valued PSC sequence c as here,

to be.

The energy values included in the PSC sequence c of Equation 2 and the PSC sequence c of Equation 3 are as shown in Equation 4, respectively.

Each chip c _q or c _q has a chip period of equation (5).

, 그리고 수학식 3의 PSC 시퀀스 c 에 의해, 복소수 값의 PSC 신호는 수학식 6과 같이 표현된다.T represents the actual time, and the Dirac Delta Function

And, by the PSC sequence c of Equation 3, a complex valued PSC signal is expressed as in Equation 6.

This complex valued PSC signal consists of a linear overlap of the time delayed version of the delta function, weighted by the appropriate complex valued chips c _q , q = 1,.

1 is a conceptual diagram of a downlink system model used in an embodiment of the present invention.

As shown, the downlink system model according to an embodiment of the present invention is a transmitter RRC filter (Transmit RRC Fiter, 10), Mobile Radio Channel (20), Additive White Gaussian Noise (AWGN) noise of the transmitter. It includes an adder (30) and the receiver (40, 50, 60) to overlap the.

The transmitter has a transmitting Root Raised Cosine (RRC) filter 10 having a Roll-Off Factor, and through the Excess Delay Parameter τ, the transmitting RRC filter 10 determines its real value. It is specified as follows by the impulse response with.

The receiver includes a receive RRC filter 40 having the same characteristics as the transmit RRC filter 10 and includes an AD converter (ADC) 50 and a tap amplitude and delay estimator 60.

The PSC signal c (t) having the complex value of Equation 6 passes through the transmission RRC filter 10 of the transmitter to generate the transmission signal s (t) having the complex value of Equation 8.

The transmission signal s (t) is weighted by the PSC chips c _q , q = 1, ..., Q, with appropriate complex values, and of the time delay replicas of the impulse response h _RRC (t) with a real value. It consists of a linear overlap.

The s (t) is provided as a reference signal for channel estimation, and this reference signal s (t) is transmitted through the mobile radio channel 20. In the following, the mobile radio channel 20 is modeled as a tap delay line and consists of P paths called taps (TAPs).

Here, each tap is identified by the radix p, p = 1, .., P, and has a tap coefficient a _p (t) that is variable over time and has a complex value, and a specific tap delay τ _p that varies over time. has (t)

In a CDMA system with a carrier frequency around 2 GHz, which is the center frequency (f ₀ ), the channel interference time T _coh is required in the order of 1 ms for a travel speed of 280 km per hour or less. Therefore, T _coh should be much larger than 67 dB, which is the duration of the reference signal s (t) transmitted from the transmitter. The maximum multipath spread T _M is up to 20 μs.

Accordingly, a _p (t) and τ _p (t) may be considered to not change with time during the duration of e (t), hereinafter not dependent on time t.

The complex envelope of the channel impulse response (CIR) is given by

The output signal e _u (t) of the mobile wireless channel 20 is expressed by Equation 10.

By both sides (Double-Sided) noise power spectral density N _0/2 Zero-Mean white Gaussian noise samples of the complex envelope function with a n (t), the received signal is given by equation (11).

At the input of the receiver, e (t) is filtered by the receiving RRC filter 40 to produce a signal having a complex value which is the impulse response of the Raised Cosine (RC) filter.

The RC filter has an ideal low pass characteristic in which the radio frequency band is 1 / T _C and is weighted by a function adjusted by the aforementioned roll-off factor α.

Although the RRC and RC filters each have infinite duration impulse bands, the largest portion of the energy contained in these impulse responses is the excess delay interval of T _RC , which is a finite duration, and τ around 0 seconds. Are focused on.

The connection of the transmitter RRC filter 10 of the transmitter, the mobile radio channel 20, and the receiver RRC filter 40 of the receiver form an efficient transmission channel specified by the combined CIR.

This efficient transport channel manages the performance of the WCDMA downlink system. It is therefore conceivable to incorporate the downlink receiver into equation (14) above.

Assuming that the impulse response of the RC filter has a finite duration T _RC , b (t) by Equation 14 is also finite duration (T _RC + T _M ).

The covariance function of noise is given by

For digital signal processing, the eRRC (t) of equation (12) is converted to a digital sequence represented by a filtered received vector r having a complex value, using the AD converter 50.

(

, N_S는 칩주기 T_c 당 샘플 수)에 의해 샘플링되어 시간상 분리되는 샘플들을 생성하는, 일반적인 샘플링을 가정할 것이다. N_S 인자에 의한 오버샘플링은 불연속 시간 TADE 알고리즘의 순간적 레졸루션(Resolution)을 향상시키기 위해 행해진다. In the present invention, temporarily spaced at equal intervals, adjacent samples are sampled

(

, N _S is the number of samples per chip period T _c ) and will assume a normal sampling, producing samples that are separated in time. Oversampling by the N _S factor is done to improve the instantaneous resolution of the discontinuous time TADE algorithm.

An efficient transmission channel has a combined CIR vector with W components as shown in equation (16).

In order to simplify the following mathematical technique, as in Equation 17, W is an integer multiplied by N _W and N _S.

An efficient transport channel can be thought of as a channel with W paths. The components of b defined in (16) are samples of b (t) in (14), which are obtained at each instant τ '.

Here, Equation 19 may be derived.

After AD converting, the noise vector n is expressed by Equation 20.

Through the complex-valued PSC sequence c defined in Equation 3, a reduced complex valued PSC matrix such as Equation 21 can be generated, and correspondingly, a PSC matrix of a real value such as Equation 22 is found.

, 그리고 크로네커 곱(Kronecker Product)을 의미하는

을 이용하여,

PSC 매트릭스를 정의할 수 있다. C in Equation 21, which is the unit matrix of N _S × N _S

, And Kronecker Product

Using

PSC matrix can be defined.

Then, with b defined in Equation 16, n in Equation 20, and C in Equation 21, a complex reception filter vector can be developed as in Equation 24.

For further signal processing in TADE, samples of r may be allocated to N _S individual sample set vectors.

각각에 포함된 샘플들은 칩 주기 T_c에 의해 시간상에서 분리된다. 따라서, 수학식 26의 N_S개의 개별 샘플 셋 노이즈 벡터들

을 형성하는 특정 벡터

에 포함된 노이즈 샘플들은 수학식 15의 변수 N₀를 갖는 부가적인 백색 가우시안 노이즈(AWGN)의 i.i.d. 샘플들이다. N _S individual sample set vectors

The samples included in each are separated in time by the chip period T _c . Thus, N _S individual sample set noise vectors of equation (26).

To form a specific vector

The noise samples included in are iid samples of additional white Gaussian noise (AWGN) with the variable N ₀ of equation (15).

에 의해 수학식 25는 수학식 28과 같이 표현된다.Combined Channel Impulse Vector of Equation 21, Equation 26, and Equation 27 Separated by N _S

Equation 25 is expressed by Equation 28.

This expression allows the introduction of parallel execution in the CITADE (Correlation based Iterative TADE) algorithm proposed by the present invention, which supports the compiler of the C6000 DSP to provide good pipelining efficiency. Since the huge parallelism of algorithms requires a change in the hardware architecture, the present invention does not address the development of an in-depth manual on parallel execution. However, the development of hardware accelerators for parallel execution will increase processing speed, compared to the cost of increasing hardware complexity.

Next, the TADE algorithm will be described.

Various publications on the TADE algorithm have been published, but most do not mention running the TADE algorithm over discrete time. However, as mentioned above, the implementation on discrete time is of significant importance, and thus the present invention will focus on the CITADE algorithm on discrete time.

The CITADE process proposed by the present invention is as follows.

는

, 에 대해 추정된 CIR 벡터를 가리킨다.In the following,

Is

, And CIR vectors estimated for.

를 구하며, 그 결과값은 결합된 CIR 벡터

이다.First, for every n _S from 1 to N _S

, And the result is the combined CIR vector

to be.

,

, 는 수학식 29에서 추정된

의 인덱스를 나타내며, 이는 p번째 최대 에너지 값

을 갖는다. 이 추정값은 p번째의 가장 강한 경로와 관련된다.

,

, Is estimated by Equation 29

Represents the index of, where p is the maximum energy value

Has This estimate is associated with the strongest path in the pth.

을 결정한다. P는 일반적으로 N_W보다 작은 것으로 추정되는 것이 바람직하며, 이러한 추정값들은 추정된 채널 진폭 벡터

를 형성한다. 각 인덱스

는 특정 딜레이(지연)와 연관된다.Next, P estimates for the P strongest paths

Determine. It is preferred that P is generally estimated to be less than N _W , and these estimates are estimated channel amplitude vectors

To form. Each index

Is associated with a specific delay (delay).

를 수학식 31에 근거하여 결정한다.Next, P delays for p from 1 to P

Is determined based on Equation 31.

에 포함된

는 수학식 9 및 수학식 9의 이전에 언급한, 1부터 P까지의 p에 대한 연속 시간값

의 불연속 추정값들로 생각될 수 있으며, 이 관계를 명확히하기 위해, 본 발명에서는 이를

로 칭한다.Of Equation 31

Included in

Is the continuous time value for p from 1 to P, previously mentioned in Equations 9 and 9

Can be thought of as discrete estimates of, and to clarify this relationship,

It is called.

의 추정값을 활용한다.

추정을 위한 추정기는 상관 추정기(Correlative Estimator) 및 CIATDE의 코어를 형성하는 코릴레이션 기반 접근법을 기초로 한다. 상관 추정기는 수학식 32의 추정 매트릭스에 의해 형성되며, 추정된 CIR 벡터는 수학식 33과 같이 주어진다.The TADE algorithm described above applies to all n _S 1 through N _S.

Use the estimate of.

The estimator for estimation is based on a correlation estimator and a correlation based approach that forms the core of CIATDE. The correlation estimator is formed by the estimation matrix of Eq. 32, and the estimated CIR vector is given by Eq.

On the other hand, the first part of the CITADE process of Equations 29 and 33 will require a high cost with respect to signal processing. The second and third steps can be performed by efficient algorithms such as the Quick Sort Algorithm (QSORT). Therefore, the present invention only discusses the first step of the CITADE process.

의 하나의 요소

를 계산하기 위한 프로세싱 에포트를 고려할 것이다. In order to measure the processing effort required in one tap delay estimation process, k is calculated from 1 to N _W.

One element of

We will consider the processing effort to compute.

를 결정하기 위해 512개의 실수값 샘플들

및

을 로딩해야 하고, 그 다음은 MAC(Multiply Accumulate) 동작이 실행되어야 한다. MAC 동작은 두 데이터 값을 곱하는 동작과, 그 결과를 이전에 획득/저장된 값과 더하는 동작으로 이루어진다. DSP에서, 하나의 MAC 동작은 보통 하나의 프로세서 사이클에서 실행된다. first,

512 real-time samples to determine

And

Next, the Multiply Accumulate (MAC) operation must be executed. The MAC operation consists of multiplying two data values and adding the result to a previously obtained / stored value. In a DSP, one MAC operation is usually executed in one processor cycle.

Slots in a UMTS / WCDMA system have a period consisting of fields containing bits, which will hereinafter be referred to as a time slot. The time slot has 2560 chip lengths. There are 15 time slots per radio frame of a 10 ms period, and channel estimation may be performed by using pilot symbols included in each time slot.

를 가중시키기 위해서는 4번의 부가적인 MAC 동작이 필요하다. 결국, 두개의 실수값을 갖는 결과 샘플들인

와

는 반드시 저장되어야 한다. 그리고, 15개의 타임 슬롯으로 구성되는 각각의 무선 프레임에 대해, 상기한 동작을 N_W·N_S·15번 즉, 무선 프레임당 2100번을 수행하여야 한다.Assuming that N _S is 2 and the maximum excess delay is 7 ms, it can be seen that N _W is 70. A review of the executed code shows that in this case it requires 1300 MAC operations per time slot. then,

Four additional MAC operations are needed to weight the. In the end, the resulting samples with two real values

Wow

Must be stored. For each radio frame consisting of 15 time slots, the above operation should be performed N _W · N _S · 15 times, that is, 2100 times per radio frame.

The algorithm described above is not desirable because it requires a processing time of approximately 5.3 ms, i.e. 8.0 time slots, on a TI TMS320C6416 DSP running at a 720 MHz clock when directly implemented as a formula.

를 도출한다.Accordingly, the circulated filtered reception vector of Equation 35 using Equation 34 below.

To derive

Using Equation 35 reduces the number of MAC operations, which requires about 3.4ms of processing time on the TI TMS320C6416 DSP and requires 5.1 time slots. But still, this method is not the best on one DSP.

The following discusses optimal processing.

Many current DSPs, such as the TI C6000 DSP group, are designed to handle packet data, and improving the hierarchical structure of 16-length PSC sequences can realize potential reductions in processing effort.

Now assume that a packet with two data elements should be processed.

Regarding the 16-length PSC sequence, the packet-related algebra operation of Equations 36 and 37 can be proposed.

The TI C6000 DSP group allows these operators to be used in the form of optimized single instructions sadd2, ssub, dotp2 using a single processor cycle. Other DSP groups will have similar instructions for similar processing.

을 구할 수 있다.Now, by using Equation 38 to improve the periodic characteristics, by using Equation 2

Can be obtained.

은 TI 명령어 LDDW를 사용함에 의해 DSP 상으로 로딩될 수 있다. 부가적인 데이터 로드나 파이프라인의 플러쉬(Flush)의 페널티를 갖는 불필요한 곱셈 또는 조건적 브랜치들을 피하기 위해, 조건적 실행이 적용될 수 있다.Each element

Can be loaded onto the DSP by using the TI instruction LDDW. Conditional execution can be applied to avoid unnecessary multiplication or conditional branches with a penalty of additional data load or flushing of the pipeline.

The processing optimization techniques mentioned so far only require 809k cycles, require processing time of 1.1 ms (1.7 time slots), and approximately 0.8 ms (1.2 time slots) when using the recently introduced 1 GHz version of the TI TMS320C6416. It can be implemented simply on a single TI TMS320C6416 DSP.

Another important point of optimization is efficient memory access and proper use of pipelining. However, memory consumption is not considered in the present invention.

Therefore, according to the frequency recognition method for a wireless recognition terminal of the present invention, it is possible to implement a seamless service while efficiently utilizing a radio resource in a multi-standard / multi-mode environment.

In addition, according to the present invention, the implementation complexity of the wireless cognitive terminal requiring the frequency or network recognition capability for the available service and system recognition can be reduced and have excellent performance, and software or firmware-based tap amplitude and delay estimation ( TADE), so its utilization can be expected.

Claims (8)

A tap amplitude and delay estimation method for efficient repetitive frequency recognition in a wireless recognition terminal requiring frequency or network recognition capability for recognition of an available wireless service and system, A) For all n _S from 1 to N _S , assuming N _S samples per chip period,

Pointer to the estimated channel impulse response (CIR) vector for

To obtain the combined CIR vector

Generating a; B)

Is estimated in step A)

If the pth path is the strongest path, then the estimated value of P strongest paths

And estimate the channel amplitude vector using the estimates.

Forming a; C)

Delay value for p from 1 to P based on

Determining; Frequency recognition method for a wireless recognition terminal comprising a. The method of claim 1, Step A) is Correlation Based Estimation Matrix

Generating a CIR vector using the frequency recognition method for a wireless cognitive terminal. The method of claim 2, In step A), N _S individual sample set vectors

When the CIR vector is expressed as

Frequency recognition method for a wireless recognition terminal, characterized in that generated by. A tap amplitude and delay estimation method for frequency recognition in a wireless recognition terminal requiring a frequency or network recognition capability for recognition of an available wireless service and system, A) For all n _S from 1 to N _S , assuming N _S samples per chip period,

Estimated CIR Vectors for

Obtaining a combined CIR vector; B) when the p-th path is the strongest path, determining an estimate for the P strongest paths and using the estimates to form an estimated channel amplitude vector; C) a delay value for p from 1 to P using the result of step B) above

Determining; Frequency recognition method for a wireless recognition terminal comprising a. The method of claim 4, wherein Step A) is Correlation Based Estimation Matrix

Generating a CIR vector by using a frequency recognition method for a wireless cognitive terminal. The method of claim 5, In step A), N _S individual sample set vectors

When the CIR vector is expressed as

Frequency recognition method for a wireless recognition terminal, characterized in that generated by. The method of claim 5, Step A) is N _S individual sample set vectors

The real value samples of the sample set vector

And

When we say Equation

And Equation

Circular filtering through vector

Deriving; Generating a CIR vector using the cyclically filtered reception vector; Frequency recognition method for a wireless recognition terminal comprising a. The method of claim 7, wherein Step A) is Equation

Improving the periodic characteristics of the PSC sequence using the PSC sequence; Frequency recognition method for a wireless recognition terminal further comprises.

Images