KR100858476B1 - A Hybrid Detection Method for Multiple Input Multiple Output Systems - Google Patents

Abstract

Multiple inputs Multiple output systems have high frequency efficiency and are most similar. The best performance when using a detector. However, as the number of signals of the transmitting antenna and the constellation increases, the detector is most similar exponentially complicated. As a result, the performance of the most similar, but less complex than the zero detector has been proposed. This invention relates to a new multi-stage detection technique that is less complex than most similar detectors and performs better than zero-generation detectors. The proposed technique is more effective when the signal to noise ratio is high. In other words, they are less complex than the most similar detectors and have about the same performance.

Multiple inputs Multiple outputs, blast, zero making detector, most similar detector

Description

A Hybrid Detection Method for Multiple Input Multiple Output Systems

1 shows several antenna transmission systems with N _T transmit antennas and N _R receive antennas.

2 shows signal constellations in a transmitting antenna when sending signals with octal phase shift keying.

Figure 3 shows the signal constellations at the transmitting antenna when sending signals with hexadecimal orthogonal amplitude modulation.

Figure 4 shows the performance of three detectors when sending signals with quaternary phase shift, octal phase shift, and hexadecimal quadrature amplitude modulation in a system where N _T = 2, N _R = 2.

FIG. 5 shows the performance of three detectors when sending signals with quaternary phase shift, octal phase shift, and hexadecimal quadrature amplitude modulation in a system where N _T = 2, N _R = 3.

FIG. 6 shows the performance of three detectors when signaling with quaternary phase shift in a system where N _T = 2, 4 and N _R = 4.

Fig. 7 shows the average number of multiplications of the three detectors when a signal is transmitted with a quaternary phase shift key, an octal phase shift key, and a hexadecimal quadrature amplitude modulation in a system where N _T = 2 and N _R = 2.

FIG. 8 shows the average number of multiplications of the three detectors when a signal is transmitted with a quaternary phase shift, an octal phase shift, and a hexadecimal quadrature amplitude modulation in a system where N _T = 2, N _R = 3.

FIG. 9 shows the average number of multiplications of the three detectors when a signal is sent with a quaternary phase shift in a system where N _T = 2, 4 and N _R = 4.

When wireless channel characteristics are bad, the diversity techniques of multiple input multiple output (MIMO) systems can be used to reduce the effects of attenuation or interference. Multiple inputs Multiple output systems are more frequency efficient and more resistant to interference than single input single output (SISO) systems. On the other hand, when paying attention to the frequency efficiency of a communication system, several input and multiple output systems are used to increase the data rate. For example, Bell (Bell Laboratories Layered Space-Time (BLAST)) systems are known to be easy to implement and have high data rates.

Many studies have proposed detectors suitable for multiple input and multiple output systems. Theoretically, the maximum likelihood (ML) detector provides the best performance in multiple input and multiple output systems, but the transmit antenna and constellation (the constellation is the constellation of the sending signal, and the complex signal is in-phase (in- It is expressed as a point on a two-dimensional plane divided into a phase component and a qurature component, and the set of points is called a signal constellation. Therefore, zero forcing (ZF) and ordered successive interference cancellation (OSIC) algorithms, which are less complex than the most similar detectors, have been proposed.

The zero-making detector can multiply the signal vector received by the pseudo inverse of the channel matrix to eliminate the interference between the received signals, and based on this, the signal can be detected independently for each antenna. In the sequential sequential interference cancellation detector, the signal vector received the pseudo inverse of the channel matrix is multiplied to eliminate the interference between the received signals, and the signals are sequentially detected according to the channel. These zero-making and sequential sequential interference canceling detectors are the closest. In addition, there are several detection schemes suitable for multiple input and multiple output systems, some of which are most similar, simpler than detectors, and better than others.

In this invention, we propose a new multi-stage detection technique that is less complex than the most similar detector and performs better than zero-made detectors, and obtains the crystal domain of the proposed detector. In addition, by changing the modulation scheme and the number of antennas, the Monte Carlo simulations are compared and compared with zero-making, most similar, and the proposed block error rate performance and average multiplication times.

FIG multiple antenna transmission system shown in Figure 1 is an N _T × N _R N _T transmit antenna system used by one, two receive antennas N _R. The transmitter divides the data into N _T pieces and sends them to the wireless channel by signing each data. In a slow attenuation wireless channel environment, the receiving antenna receives a combination of signals from the N _T transmitters. In this case, it is assumed that the channel is estimated by a short training sequence before detecting the signal.

In the transmitting antenna, the set S = {s ⁽¹⁾ , s ⁽²⁾ , ..., which has M signals as elements. , s ^(M) } to send the information. In the signal set S, the set of all repetitive permutations that select N _T signals is called V, and the element of V

[Equation 1]

은 아래와 같이 나타낼 수 있다.Let's say Where the superscript T represents the vector transpose. Received signal vector

Can be written as

[Equation 2]

Here, r _j is a signal received from the jth receiving antenna, and the channel matrix H is

[Equation 3]

은 독립이고 분포가 같은 복소 확률 잡음 벡터이다. 이때, n_j는 j째 수신 안테나에서 받은 잡음이며, 채널 행렬 H의 원소 h_ji는 i째 송신 안테나에서 j째 수신 안테나 사이의 복소 채널 전달 계수를 나타내고, {h_ji}는 평균이 0이고 분산이 1인 독립 복소 정규 확률변수들이라 하자.ego,

Is an independent and equally complex probability noise vector. Where n _j is the noise received from the jth receive antenna, and the element h _ji of the channel matrix H represents the complex channel transfer coefficient between the jth receive antenna and the jth receive antenna, and {h _ji } has an average of 0 and variance Suppose that these 1 independent complex normal random variables.

-진 가설검정 문제로 모형화할 수 있다. 여기서, 가설

는 아래와 같다.Now, the detection problem is when there is noise

Can be modeled as a true hypothesis test problem. Where hypothesis

Is shown below.

[Equation 4]

이다. 관측 모형 수학식 4에서

과

의 결합 확률밀도함수

는 아래와 같다.From above, s _{i, k} ∈ S,

to be. In observation model equation 4

and

Combined probability density function of

Is shown below.

[Equation 5]

과

은 각각 실수와 허수 부분을 나타내고,

은

의 공통 확률밀도함수이며,

는

의 공통 확률밀도함수이다.here,

and

Represent the real and imaginary parts, respectively.

silver

Is the common probability density function of

Is

The common probability density function of.

Using the similarity criterion, we obtain the decision region of the most similar detector as shown below.

[Equation 6]

Most similar in a normal noise environment.

[Equation 7]

은 유클리드 거리이다. 이론적으로 가장 비슷함 검파기를 쓰면 성능이 가장 뛰어나지만, 송신 안테나와 별자리의 신호수가 많아질수록 가장 비슷함 검파기는 지수적으로 복잡해진다. 보기를 들어, 4×4 시스템에서 16진 직교진폭변조(quadrature amplitude modulation: QAM)를 하여 신호를 보낼 때, 신호를 검파하려면 65536번 견주어보고 2097152번 곱해야 한다.to be. here,

Is at the Euclidean street. Theoretically, the closest detector offers the best performance, but the higher the number of signals in the transmit antenna and constellation, the closer the detector becomes exponentially complex. For example, when sending a signal with hexadecimal quadrature amplitude modulation (QAM) in a 4x4 system, the signal must be compared to 65536 times and multiplied by 2097152 times.

The zero-making detector can multiply the signal vector received by the pseudo inverse of the channel matrix to eliminate the interference between the received signals, and based on this, the signal can be detected independently for each antenna. This spirit making detection is divided into two stages. first,

을 얻는다. 여기서,

이고, 채널 행렬 H의 의사역행렬 V는 아래와 같으며,
[수학식 9]

[Equation 8]

Get here,

The pseudo inverse V of the channel matrix H is
[Equation 9]

delete

delete

delete

delete

은 켤레 복소수 전치를 나타낸다. 이때, 행렬 V는 여러 방법들을 써서 간단히 얻을 수 있다. 채널의 감쇄가 느리게 나타날 때, 채널 행렬 H는 심벌 주기 동안에 바뀌지 않는다고 할 수 있으므로 행렬 V를 한번만 계산한다. 여기서, VH=I이므로 수신 안테나는 다른 신호들의 간섭 없이 송신 안테나에서 보낸 신호를 검파할 수 있다. 수학식 8을 바탕으로 y를 얻은 다음 아래의 결정 영역을 써서 신호를 검 파한다.

Denotes the complex conjugate transpose. In this case, the matrix V can be obtained simply by using several methods. When the attenuation of the channel is slow, the channel matrix H is not changed during the symbol period, so the matrix V is calculated only once. Here, since VH = I, the receiving antenna can detect a signal sent from the transmitting antenna without interference of other signals. Based on Equation 8, y is obtained and the signal is detected using the following decision region.

[Equation 10]

Meanwhile, in the normal noise environment, Equation 10 is

[Equation 11]

This can be calculated simply by using a slicer.

We now propose a new multi-stage detection technique that is less complex than the most similar detector and performs better than the zero-creation detector. The proposed technique can be divided into two stages. In the first step, we use the zero-making algorithm to estimate the sent signal, and in the second step, we find the sent signal based on the detection of the closest approximation in the neighboring constellation. At this time, if the first estimate is the same as found in the second step, the first (and second) estimated signal is determined as the final output. Otherwise, the constellation is extended with the remaining signals to form a larger constellation to continue detecting the most similar.

를 얻는다.In order to detect the signal while reducing the complexity, we first use the zero-making detection technique. At this time, the receiving antenna is estimated signal using the decision region equation (11) of the zero making detector

Get

는 가장 비 슷함 검파 기법으로 추정한 신호에 가까울 것이다. 그렇지 않고, 행렬 V의 원소가 1보다 아주 크다면, 수학식 8의 잡음 성분 Vn는 수학식 2의 잡음 성분 n보다 매우 크고, 수학식 8에서 잡음 성분이 추정신호

에 주는 영향도 클 것이다. 다시 말해서, 추정신호

에 가까운 심벌들 가운데, 보낸 심벌과의 거리가

보다 짧은 심벌이 적어도 하나 있다. 그러므로, 처음 추정한 심벌과 이웃한 별자리에서 가장 비슷함 검파로 보낸 신호를 찾는다. 처음 추정한 신호와 이 신호와 이웃한 별자리를 '줄인 별자리'라 부르자. 도 2와 3은 각각 8진 위상편이변조(phase shift keying: PSK)와 16진 직교진폭변조하여 신호를 보낼 때, 송신 안테나에서의 신호 별자리를 나타낸다.If the elements of matrix V are similar to 1, then the signal estimated in step 1

Will be close to the signal estimated by the most similar detection technique. Otherwise, if the element of the matrix V is much larger than 1, the noise component V n of Equation 8 is much larger than the noise component n of Equation 2, and the noise component of Equation 8 is the estimated signal.

It will also have a big impact. In other words, the estimated signal

Among the symbols close to, the distance from the sent symbol

There is at least one shorter symbol. Therefore, find the signal sent by the first similarity detection in the neighboring constellation. Let's call the first estimate of the signal and the constellation contiguous with the signal a reduced constellation. 2 and 3 show signal constellations in a transmitting antenna when transmitting signals with octal phase shift keying (PSK) and hexadecimal quadrature amplitude modulation, respectively.

는 아래와 같다.Second estimated signal

Is shown below.

[Equation 12]

가 줄인 별자리에서 가장 비슷함 검파 기법으로 찾은

와 같다면, 곧, k=q이면

를 마지막 출력으로 결정한다. 한편, k≠q이면, 셋째 단계를 쓴다.Where S _q is the set of signals in the reduced constellation and V _q is the set of all repetitive permutations that select N _T signals from S _q . Initially estimated signal

Most similar in constellations with

Is equal to k = q

Is determined as the last output. On the other hand, if k ≠ q, the third step is written.

와 둘째 단계에서 찾은 신호

가 같지 않으면, 줄인 별자리를 늘려 가장 비슷함 검파를 하고 수학식 7을 써서 마지막 출력

를 얻는다. 이때, 단계 i에서 추정한 신호가 단계 i-1에서 추정한 것과 같을 때까지 한 단계씩 별자리를 늘리면서 둘째 단계를 되풀이한다.Initially estimated signal

And signals found in the second step

If is not equal, increase the reduced constellation to detect the most similar and use Equation 7 to output

Get At this time, the second step is repeated while increasing the constellation step by step until the signal estimated in step i is equal to that estimated in step i-1.

쯤이고 곱셈 횟수는

쯤이다. 견줌 횟수에서

는 수학싯 7과 수학식 11에서 얻었고 남은 1은 모의실험 결과에서 얻었다. 이와 비슷하게, 곱셈 횟수에서

는 수학식 7과 11에서 얻었고 남은 2N_RN_T는 모의실험 결과에서 얻었다.Tables 1 and 2 show the zero making, the most similar, and the average number of times of multiplication and multiplication of the proposed detector when the signal to noise ratio is 30dB. Determinants of Equations 7 and 11 are the most similar to zero making. The average number of times of multiplication and multiplication of the detector can be obtained, respectively. In this case, the average number of times of multiplication and the number of multiplications of the proposed detector depend on the signal-to-noise ratio. When the signal-to-noise ratio is 30dB and the M-phase phase shift signal is sent, the average frequency of the proposed detector is

And the number of multiplications

About In the number of times

Were obtained from Equations 7 and 11, and the remaining 1 was obtained from the simulation results. Similarly, in the multiplication count

Were obtained from Equations 7 and 11 and the remaining 2N _R N _T was obtained from the simulation results.

For example, when the signal-to-noise ratio is 30 dB, and the signal is sent with hexadecimal orthogonal amplitude modulation in a 4x4 system, to detect the signal, it must be 64 times compared to 128 times in a zero-creation detector and multiplied by 128 times. It should be multiplied by 2097152 times, compared with 642 times in the proposed detector, and multiplied by 20160 times. In other words, the average number of times of multiplication and multiplication of detectors is 102 times and 104 times that of the proposed number of times of multiplication and multiplication. On the other hand, when the signal-to-noise ratio is less than 30dB and the M-phase phase shift signal is sent, the average number of times of multiplication and the number of multiplications of the proposed detector change according to the number of signals M of the constellation.

Finally, if N _T = 1 or M = 2,3, the proposed detector is the most similar since the constellation is the same as the reduced constellation of the detector. The average number of times of multiplication and multiplication is more than that of the detector. Thus, the proposed detector is useful when N _T ≥ 1 and M ≥ 4.

TABLE 1

At a signal-to-noise ratio of 30 dB, zero-making, most similar, and average number of times the proposed detector is observed in an N _T × N _R system

TABLE 2

At a signal-to-noise ratio of 30 dB, we make zero, most similar, and mean multiplication of the proposed detector in an N _T × N _R system.

Now, we simulate 10 to ⁶ times with Monte Carlo method in a normal noise environment, and compare and obtain the block error rate (BLER) performance and average multiplication times of the proposed detector. The average number of times of multiplication and multiplication of the detector is often used to indicate the complexity of the detector. Here, the number of observations is much smaller than the number of multiplications, so it is not considered. Table 3 shows the simulation environment covered in this invention.

Figure 4 shows the zero-making, most similar, and proposed detectors when sending signals with quaternary phase shift, octal phase shift, and hexadecimal quadrature amplitude modulation in a system where N _T = 2 and N _R = 2. Performance characteristics are shown. The similarity and the proposed detector are almost the same, and the performance is better than the zero detector. On the other hand, as the number of signals in the constellation M increases, zero making, most similar, and the performance of the proposed detector decreases. Fig. 5 shows zero-making, most similar, and proposed detectors when sending signals with quaternary phase shift, octal phase shift, and hexadecimal quadrature amplitude modulation in a system where N _T = 2 and N _R = 3. Show performance characteristics. In this result, similar characteristics to those shown in FIG. 4 can be seen. 4 and 5, as the number of receiving antennas increases, the performance difference of the detector decreases, and the performance difference of the zero generator and the proposed detector decreases.

In the system of N _T = 2, 4 and N _R = 4, when the signal is transmitted by the quaternary phase shift, the zero-making, the most similar, and the block error rate performance characteristics of the proposed detector are shown. As the number of transmitting antennas increases, zero making, most similar, and the performance of the proposed detector decreases.

7 shows a zero-making, most similar, and proposed detector when sending signals with quaternary phase shift, octal phase shift, and hexadecimal quadrature amplitude modulation in a system where N _T = 2 and N _R = 2. The average number of multiplications of each is shown. When the signal-to-noise ratio is high, as the number of signals in the constellation increases, the zero-producing and the proposed detectors are the most similar. The average number of multiplications is smaller than the detector.

8 shows zero making, the most similar, and the average number of multiplications of the proposed detector when the number of receiving antennas is increased to 3 in FIG. Here, similar results to those seen in FIG. 7 can be seen. 7 and 8, the average number of multiplications of the zero making detector does not change even if the number of receiving antennas increases, but it is similar. The number of multiplications of the detector and the proposed detector increases as the number of receiving antennas increases. Also, when the signal-to-noise ratio is high, the maximum number of receiving antennas is most similar. The detector has more average multiplication times than the proposed detector. Fig. 9 shows that when the signal-to-noise ratio is high, the number of transmitting antennas increases, the zero making and the proposed detector are the most similar.

TABLE 3

Simulation table

The proposed technique was more effective when the signal to noise ratio was high. The proposed detector is most similar when the signal-to-noise ratio is high. Less complexity and nearly identical performance than the detector.

Claims (8)

In the method of detecting multiple input multiple output communication system, Estimating a signal sent by using a zero making detector; A second step of finding a sent signal based on detection of the most similarity in the first estimated symbol and the neighboring constellation; If the signal estimated in the first step and the signal found in the second step is the same, this signal is determined as the last output, and if it is different, the third step of continuing the second step by increasing the constellation with the remaining signals; Multiple input Multiple output communication system detection method. In the method of detecting multiple input multiple output communication system, A first step of estimating a signal sent by using a zero making detector using a determination region of the following equation; A second step of finding a sent signal based on the most similar detection in a signal estimated in the first step and a constellation representing a constellation adjacent to the estimated signal; If the estimated signal in the first step and the signal found in the second step is the same, this signal is determined as the last output, and if it is different, the third step of continuing the second step by increasing the reduced constellation; Input detection method for multiple output communication systems.

(Equation 11) here,

(8) and the received signal vector

Is,

The pseudo inverse V of the channel matrix H is

(9), and the signal sent

(Equation 1),

Is,

Is at the Euclidean street. In the method of detecting multiple input multiple output communication system, Input of the estimated signal and the signals in the constellations, which represent the constellations adjacent to the estimated signal, and input signals using the most similar detector using the following decision region. Detection method for multiple output communication systems.

(Equation 7) Where the received signal vector

Is,

H _ji is the element of the channel matrix H and the signal sent

(Equation 1),

Is,

Is at the Euclidean street. The method of claim 2, wherein the continuing of the second step is to repeat the detection of the most similarity by increasing the constellation by one step until the signal estimated in step i is equal to that estimated in step i-1. Input detection method for multiple output communication systems. In the detector of multiple input multiple output communication system, Using a zero making detector to estimate the signal sent; Find the signal sent on the basis of the most similar detection in the first estimated symbol and neighboring constellations; If the estimated signal and the found signal are the same, this signal is determined as the last output, and if it is different, the constellation is increased by remaining constellations to continue the search for the sent signal based on the most similar detection. Detector of output communication system. In the detector of multiple input multiple output communication system, Estimating a signal sent by using a zero making detector using a decision region of the following equation; Finding a signal sent based on the most similar detection in the estimated signal and a constellation in which the estimated signal is adjacent to the constellation indicating the constellation; If the estimated signal and the found signal are the same, this signal is determined as the last output, and if it is different, a larger constellation is created to include the reduced constellation and the neighboring constellation, and based on the most similar detection above for this constellation, A detector of multiple input multiple output communication systems characterized by continuing to find the signal sent.

(Equation 11) here,

(8) and the received signal vector

Is,

The pseudo inverse V of the channel matrix H is

(9), and the signal sent

(Equation 1),

Is,

Is at the Euclidean street. In the detector of multiple input multiple output communication system, Input of the estimated signal and the signals in the constellations, which represent the constellations adjacent to the estimated signal, and input signals using the most similar detector using the following decision region. Detector for multiple output communication systems.

(Equation 7) Where the received signal vector

Is,

H _ji is the element of the channel matrix H and the signal sent

(Equation 1),

Is,

Is at the Euclidean street. 6. The multiple input multiple output communication according to claim 5, wherein the continuing is to repeat the most similar detection while increasing the constellation step by step until the signal estimated in step i is equal to that estimated in step i-1. Detector of the system.

Images