CN107359912B

CN107359912B - Maximum likelihood detector

Info

Publication number: CN107359912B
Application number: CN201610302079.2A
Authority: CN
Inventors: 张仲尧; 林专湖; 杨易洵
Original assignee: Realtek Semiconductor Corp
Current assignee: Realtek Semiconductor Corp
Priority date: 2016-05-09
Filing date: 2016-05-09
Publication date: 2020-06-16
Anticipated expiration: 2036-05-09
Also published as: CN107359912A

Abstract

The invention relates to a maximum likelihood detector and a detection method. A maximum likelihood detector includes a search value selection circuit and a maximum likelihood detection circuit. The maximum likelihood detection circuit is configured to perform at least the following steps: selecting K first layer candidate signal values according to the received signal or its derivative, the channel estimation signal or its derivative, and the first layer search value; calculating K second-layer candidate signal values according to the K first-layer candidate signal values; judging whether to increase P second-layer supplementary candidate signal values according to the K second-layer candidate signal values to generate a judgment result; if so, calculating a log-approximate ratio according to the K first-layer candidate signal values, the K second-layer candidate signal values, the P first-layer complementary candidate signal values and the P second-layer complementary candidate signal values; and when the judgment result is negative, calculating a logarithmic approximation ratio according to the K first-layer candidate signal values and the K second-layer candidate signal values.

Description

Maximum likelihood detector

Technical Field

The present invention relates to wireless signal reception, and more particularly, to wireless signal reception with maximum likelihood detection.

Background

In wireless communication applications, the user's demand for high data transmission rate is rising, and how to effectively increase the bandwidth utilization efficiency and further increase the system Throughput (TP) in the limited system bandwidth is always an issue to be researched by each of the practitioners. In the market trend, Spatial Multiplexing (Spatial Multiplexing) transmission scheme under MIMO (Multiple-Input Multiple-Output) technology can greatly increase system throughput without increasing bandwidth, and thus the transmission scheme has attracted considerable attention in recent years.

The spatial multiplexing transmission method mainly utilizes a plurality of antennas at a transmitting end to transmit mutually independent signals at the same time and the same frequency band, and a receiving end also adopts a plurality of antennas to receive and detect the signals. In order to achieve better demodulation performance, the receiving end may use Maximum Likelihood (ML) method. The maximum likelihood method can be understood as an optimization-pursuing algorithm that estimates the best solution of the transmitted signal among all possible solutions from the received signal in the form of an Exhaustive Search (explicit Search). However, exhaustive search is not an efficient search method, since it takes long Computation time delay (processing latency), Complexity (Complexity) and Computation Power (Computation Power) to compute all possible solutions.

More prior art can be found in the following documents: chinese patent application publication No. CN 101582748A; and IEEE, Massimiliano Siti, Michael P.Fitz, "A Novel Soft-Output layer and organic layer Detector for Multiple Antenna Communications," IEEE International Conference on Communications (ICC), 2006.

Disclosure of Invention

In view of the deficiencies of the prior art, it is an object of the present invention to provide a maximum likelihood detector and a detection method for improving the prior art.

The invention discloses a maximum likelihood detector for performing a maximum likelihood detection according to a received signal or a derivative thereof and according to a channel estimation signal or a derivative thereof. One embodiment of the maximum likelihood detector includes a search value selection circuit and a maximum likelihood detection circuit. The search value selection circuit is used for determining a first layer search value, and the first layer search value is not less than a preset threshold value. The maximum likelihood detection circuit is configured to perform at least the following steps: selecting K first layer candidate signal values according to the received signal or a derivative thereof, the channel estimation signal or a derivative thereof, and the first layer search value, wherein K is a positive integer greater than 1; calculating K second-layer candidate signal values according to the K first-layer candidate signal values; judging whether to increase P second-layer supplementary candidate signal values according to the K second-layer candidate signal values, and generating a judgment result according to the P second-layer supplementary candidate signal values, wherein P is a positive integer; if the judgment result is yes, adding the P second-layer supplementary candidate signal values, selecting P first-layer supplementary candidate signal values according to the P second-layer supplementary candidate signal values, and calculating a log-approximate ratio according to the K first-layer candidate signal values, the K second-layer candidate signal values, the P first-layer supplementary candidate signal values and the P second-layer supplementary candidate signal values; and when the judgment result is negative, calculating a logarithmic approximation ratio according to the K first-layer candidate signal values and the K second-layer candidate signal values.

Another embodiment of the maximum likelihood detector comprises a search value selection circuit and a maximum likelihood detection circuit. The search value selection circuit is used for determining a first search value and a second search value, and the first search value and the second search value are not smaller than a preset threshold value. The maximum likelihood detection circuit is configured to perform at least the following steps: selecting K1 first-tier first candidate signal values according to the received signal or a derivative thereof, the channel estimation signal or a derivative thereof, and the first search value, wherein K1 is a positive integer greater than 1; calculating K1 second candidate signal values according to the K1 first candidate signal values; selecting K2 first-layer second candidate signal values according to the received signal or its derivative, the converted signal of the channel estimation signal or its derivative, and the second search value, wherein K2 is a positive integer greater than 1; calculating K2 second-level first candidate signal values according to the K2 first-level second candidate signal values; a log-approximate ratio is calculated according to the K1 first-tier first candidate signal values, the K1 second-tier second candidate signal values, the K2 first-tier second candidate signal values, and the K2 second-tier first candidate signal values.

The features, operation and function of the present invention will be described in detail with reference to the drawings.

Drawings

FIG. 1 is a schematic diagram of an embodiment of a wireless signal receiver employing maximum likelihood detection in accordance with the present invention;

FIG. 2 is a tree diagram of an exhaustive search;

FIG. 3 is a tree diagram of the architecture of a hierarchical orthogonal lattice detector;

FIG. 4 is a schematic diagram of one embodiment of a maximum likelihood detector of the present invention;

FIG. 5 is a schematic diagram of one embodiment of the signal detection circuit of FIG. 4;

FIG. 6 is a constellation diagram for an implementation of the maximum likelihood detector of FIG. 4;

FIG. 7 is a tree diagram corresponding to FIG. 6;

FIG. 8 is a schematic diagram of one embodiment of the signal detection circuit of FIG. 4;

FIG. 9 is a diagram illustrating an optimal path corresponding to a minimum distance;

FIG. 10 is a diagram illustrating an implementation concept of the candidate signal value supplementing unit of FIG. 8;

FIG. 11 is a diagram illustrating insufficient second tier candidate signal values;

FIG. 12 is a schematic diagram of a supplemental path corresponding to a minimum distance;

FIG. 13 is a schematic diagram of one embodiment of the signal detection circuit of FIG. 4;

FIG. 14 is a schematic diagram of one embodiment of the signal detection circuit of FIG. 4;

FIG. 15 is a diagram illustrating an embodiment of the second detection circuit of FIG. 14.

Detailed Description

The following description is made with reference to terms commonly used in the art, and some terms are defined or explained in the specification, and the explanation of the terms in the specification is based on the description or the definition in the specification.

The disclosure includes a maximum Likelihood detector and a detection method, and a wireless signal receiver using the maximum Likelihood detection, which output a corresponding Log Likelihood Ratio (LLR) as an input value of a decoding circuit in a Soft Decision (Soft Decision) manner in units of bits (bits) to perform error correction. Some of the components of the detectors and receivers, taken alone, may be known components, and the details of the individual known components will be omitted from the following description without departing from the full disclosure and applicability of the invention; in addition, the method can be in the form of software and/or firmware, and can be implemented by hardware devices of the present invention or equivalent circuits thereof. The invention is suitable for receiving signals of multi-Dimension (Multiple-Dimension) or multi-Layer (Multiple-Layer), such as the signal receiving of a Multiple-Input Multiple-Output (MIMO) communication system. The MIMO communication technology is, for example, Long-Term Evolution (LTE) technology, Wireless Local-area network (WLAN) technology, Worldwide Interoperability for microwave access (WiMax), and the like. For convenience of understanding, the following description will be given by taking an application of the LTE communication system as an example, but the application of the present invention is not limited thereto.

Please refer to fig. 1, which is a diagram illustrating a wireless signal receiver employing maximum likelihood detection according to an embodiment of the present invention. As shown in fig. 1, the wireless signal receiver 100 includes: a Discrete Fourier Transform (DFT) circuit 110 for converting a time domain signal into a frequency domain signal; a reference signal Extraction (Extraction) circuit 120 for generating an extracted reference signal according to the frequency domain signal; a channel estimation (channel estimation) circuit 130 for generating an estimation signal according to the extracted reference signal; a data signal extracting circuit 140 for generating an extracted data signal according to the frequency domain signal; a signal detection circuit 150 for generating a detection signal according to the channel estimation signal and the extracted data signal; and a decoding circuit 160 for generating a decoding signal according to the detection signal. The signal detection circuit 150 comprises a maximum likelihood detector 152 for performing at least the following steps: determining a first layer search value; and a maximum likelihood detection step. The maximum likelihood detecting step includes: selecting K first layer candidate signal values according to the extracted data signal or a derivative thereof, the channel estimation signal or a derivative thereof, and the first layer search value, wherein K is a positive integer greater than 1; calculating K second-layer candidate signal values according to the K first-layer candidate signal values; judging whether to increase P second-layer supplementary candidate signal values according to the K second-layer candidate signal values, and generating a judgment result according to the P second-layer supplementary candidate signal values, wherein P is a positive integer; if so, adding the P second-tier complementary candidate signal values, selecting P first-tier complementary candidate signal values according to the P second-tier complementary candidate signal values, and calculating a log-likelihood ratio according to the K first-tier candidate signal values, the K second-tier candidate signal values, the P first-tier complementary candidate signal values, and the P second-tier complementary candidate signal values to serve as at least a part of the detection signal; and when the judgment result is negative, calculating a logarithm approximate ratio according to the K first-layer candidate signal values and the K second-layer candidate signal values to be used as at least one part of the detection signal. One embodiment of the decoding circuit 160 comprises a Descrambler (Descrambler) for generating a descrambled signal according to the detection signal and a Turbo Decoder (Turbo Decoder) for generating the decoded signal according to the descrambled signal. Any of the discrete fourier transform circuit 110, the reference signal extraction circuit 120, the channel estimation circuit 130, the data signal extraction circuit 140, and the decoding circuit 160, taken alone, are known or self-developed circuits.

As mentioned above, the time domain signal is discrete Fourier transformed and then is the signal Y of the nth subcarrier in the frequency domain_nCan be expressed as an N_RThe vector of x1, as follows:

Y_n＝H_nX_n+W_n(formula one)

Wherein H_nN seen by the nth sub-carrier_R×N_TChannel matrix of, X_nFor N on the nth sub-carrier_TX1 transmission signal, W_nNoise of the nth sub-carrier, N_RIs the number of receiving antennas, and N_TIs the number of transmit antennas. In the present embodiment, a signal of one subcarrier is taken as a processing unit, and for convenience of representation, in the following description, the subscript n is omitted; in addition, the present embodiment is applied to two-transmission two-reception space multiplexing transmission, that is, the number of independent spatial streams (spatial streams) at the transmitting end and the number of independent spatial streams at the receiving end are both 2 (in this case, N is equal to N)_RAnd N_TEach of at least 2), so equation one can be simplified to:

y is HX + W (formula two)

In the LTE system, since the Reference Signal (RS) and the Data Signal (Data Signal) are carried on different subcarriers, the channel estimation circuit 130 obtains the channel response matrix H ^ of the desired Data Signal by interpolation or extrapolation according to the channel estimation performed by the Reference Signal, and the Signal detection circuit 150 performs QR decomposition or equivalent operation on the matrix H ^ as follows:

h ^ QR (three formula)

Where Q is a Unitary Matrix (Unitary Matrix) and R is an Upper Triangular Matrix (Upper Triangular Matrix). After completion of QR decomposition or equivalent operation, the signal detection circuit 150 multiplies the reception signal Y (i.e., the aforementioned extracted data signal) by Q^H(Hermitian matrix of matrix Q) or performing an equivalent operation to obtain signal Z as follows:

Z≡Q^HY＝Q^H(HX+W)＝Q^H(QRx + W) ═ RX + W' (equation four)

Wherein W' ≡ Q^HW is added. Next, the maximum likelihood detector 152 calculates the log approximate ratio in units of bits from the signal Z obtained by using the formula four as follows:

wherein b is_iIs the ith bit, G (X)_bi＝0) Set of all transmitted signals equal to 0 in the ith bit, G (X)_bi＝1) The set of bits equal to 1 for all transmitted signals,

representing candidate signal values during operation, the symbols "" are used to distinguish them from the real signal X.

Equation five can be regarded as an optimization pursuit equation, and can be solved by means of Exhaustive Search (explicit Search) as described in the prior art. If the modulation technique of the transmitted signal belongs to M-QAM (quadrature amplitude modulation) (Quadratu), the size of the signal set (or the number of Constellation points on the Constellation Diagram (Constellation Diagram) corresponding to the modulation technique) is Mre Amplitude Modulation)), the probability of the solution of equation five is M in total for two independent data streams received and transmitted, and two independent data streams received and transmitted²The seed can be represented by a tree diagram, as shown in fig. 2.

Further, since R is an upper triangular matrix, the search is performed

The operation of (2) can be simplified. Firstly, the method

Can be disassembled as follows:

since the square of the absolute value is not necessarily less than zero, it is chosen

In the case of (a) in (b),

must occur when

The following relationships are satisfied:

wherein [ gamma ], [ alpha ]]Is a Quantizer (Quantizer). In other words, in the selection

After that time, the user can use the device,

is the only solution that can be obtained through equation seven. The structures represented by the six and seven are called hierarchical orthogonal lattice detectors (LORD), and the searched treemaps can be greatly simplified as shown in fig. 3. In the LORD architecture corresponding to FIG. 3, the first layer is expanded (unfolded)

All the possibilities (M), the second layer is directly obtained by the formula seven

Does not need to be expanded again as shown in FIG. 2

Thus, the overall computational complexity is to manipulate the expansion at the first level

The number of (2).

The LORD structure corresponding to FIG. 3 is utilized to solve the formula five, so as to obtain the log-likelihood ratio of each bit (i.e. the number of the log-likelihood ratios is equal to the number of bits of the transmitted signal), and then the obtained log-likelihood ratio is sent to the decoder for error correction to complete the reception. However, for M-QAM, there are M possible solutions under the LORD architecture (or the first layer expansion in FIG. 3)

M) where some of the possible solutions are less likely to be correct solutions, should be excluded to further simplify the operation. Therefore, the maximum likelihood detector 152 of fig. 1 does not directly calculate the log-likelihood ratios for M possibilities, but first selects a search value according to the frequency domain signal or its derivative signal (e.g., the extracted data signal) and performs a maximum likelihood operation according to the search value to generate the detection signal including the log-likelihood ratios, wherein the number of candidate signal values within the search range corresponding to the search value (or the constellation points corresponding to the search range in the constellation corresponding to the modulation mode) is not greater than the number of all candidate signal values of the modulation mode (or all constellation points in the constellation corresponding to the modulation mode). In other words, the maximum likelihood test of FIG. 1The detector 152 is deployed at the first layer

(or

) Only K candidate signal values (where K ≦ M) are expanded to achieve lower computational complexity and power consumption without substantial performance loss. It is noted that the maximum likelihood detector 152 may be first expanded at the first layer with the aid of a swap operation determination unit (described later)

Then spread on the second layer

(hereinafter referred to as reverse-development), however, for the sake of clarity, the following description will mostly be given with respect to the first layer developed first

Then spread on the second layer

(hereinafter referred to as forward-unfolding) for example, but the reverse unfolding method can be understood by those skilled in the art from the disclosure of the present specification.

In particular, one embodiment of the maximum likelihood detector 152 is shown in FIG. 4 and includes a search value selection circuit 410 and a maximum likelihood detection circuit 420. The search value selection circuit 410 is used to determine the first layer search value, wherein the first layer search value is not less than a predetermined threshold, and the number of first layer candidate signal values associated with the predetermined threshold is not more than or less than the number of all first layer candidate signal values of the modulation scheme, for example, when the modulation scheme is M-QAM, a non-limiting example of the first layer search value is an integer greater than M/4, or a further non-limiting example of the first layer search value is not less than 4 and not more than 4

Is an integer of (1). The maximum likelihood detection circuit 420 is used to perform the operation of the following formula eight and the following table look-up operation or the equivalent thereof to obtain the first layer candidate signal value, perform the operation of the following formula seven to obtain the second layer candidate signal value, and perform the operation of the following formula five to generate the logarithm probability ratio.

The signal detection circuit of fig. 4 is shown in detail in fig. 5, and includes: a QR decomposition unit 530 for performing the operation of equation III or its equivalent according to the estimated signal (corresponding to the channel response matrix H ^); a signal generating unit 540 for performing the operation of formula four or the equivalent thereof according to the extracted data signal (corresponding to the signal Y); the search value selection circuit 410; and the maximum likelihood arithmetic circuit 420. In fig. 5, the maximum likelihood calculation circuit 420 includes: a first layer candidate signal value determining unit 522, configured to perform an operation of formula eight or an equivalent operation thereof according to the extracted data signal to generate an operation result, and perform a table lookup operation or an equivalent operation thereof according to the operation result and the first layer search value to generate an operation result; a second-level candidate signal value determining unit 524, configured to perform the operation according to the operation result to generate a calculation result including the operation result; and a logarithmic approximation ratio calculation unit 526 for performing the operation of formula five or its equivalent according to the calculation result. Either of the QR decomposition unit 530 and the signal generation unit 540, taken alone, are known or developed-by-the-self units.

The way in which the maximum likelihood detection circuit 420 selects the search range corresponding to the first layer search value is determined by the landing point of the received signal on the constellation corresponding to the modulation scheme. Based on equation six, the maximum likelihood detection circuit 420 may include a Zero-Forcing (ZF) Equalizer (Equalizer) (not shown) to perform a Zero-Forcing operation to obtain

The drop points on the constellation are as follows:

using the characteristics of the constellation diagram, each constellation point and the drop point

The distance relationship of (a) can be known in advance.

Take 64-QAM of LTE as an example, according to the drop point

Position of (i.e. drop point)

Corresponding values) and constellation points, 14 intervals can be defined as shown in table 1, and in each interval, the left-to-right sequence in table 1 represents the constellation point and the drop point

From near to far; by means of the constellation diagram of figure 6,

drop points on the constellation diagram

It is more clearly shown that the symbol "x" in the numerical value designations "1 x1x1 x", "x 0x1x 1", etc., may be 0 or 1. In the example of figure 6, it is shown,

the Real Part (Real-Part) of the constellation point falls in an interval 2 which is more than or equal to S < 3, the Real parts of the constellation points from near to far are 3, 1, 5, -1, 7, -3, -5 and-7 in sequence, and the probability is 8; in the same way, the method for preparing the composite material,

the Imaginary Part (Imaginary-Part) of (A) falls in the interval of-3 ≦ S < -2, and the Imaginary parts of the constellation points from near to far are sequentially-3, -1, -5, 1, -7, 3, 5 and 7, and the total number of possibilities is 8. Hypothetical searchThe search value determined by the search value selection circuit 410 is 25 (i.e., the maximum likelihood detection circuit 420 is spread out at the first level)

Only 25 first-tier candidate signal values are expanded), the maximum likelihood detection circuit 420 sequentially extracts the falling points

The nearest real part and imaginary part are 5 in each case to match 25 possibilities (or 25 candidate signal values) whose physical meaning is represented by the dotted points in fig. 6

The search area (e.g., rectangular or other shaped area) is centered, and the constellation points falling within the search area are the most likely detection circuit 420 to spread out in the first layer

Time-computed constellation points (e.g. drop points)

The nearest constellation point), the rest of the constellation points are not considered in the region, so that the present embodiment can achieve the purpose of reducing the computation complexity, as shown in the tree diagram of fig. 7. It is noted that the embodiment may query the pre-established contents (e.g., the contents of table 1) by a table lookup operation to obtain the combination of the real part and the imaginary part, thereby allowing the maximum likelihood detection circuit 420 to calculate the log-likelihood ratio according to the formula five more efficiently.

TABLE 1

S＜-6	-7	-5	-3	-1	1	3	5	7
									-6≤S＜-5	-5	-7	-3	-1	1	3	5	7
-5≤S＜-4	-5	-3	-7	-1	1	3	5	7
									-4≤S＜-3	-3	-5	-1	-7	1	3	5	7
-3≤S＜-2	-3	-1	-5	1	-7	3	5	7
									-2≤S＜-1	-1	-3	1	-5	3	-7	5	7
-1≤S＜0	-1	1	-3	3	-5	5	-7	7
									0≤S＜1	1	-1	3	-3	5	-5	7	-7
1≤S＜2	1	3	-1	5	-3	7	-5	-7
									2≤S＜3	3	1	5	-1	7	-3	-5	-7
3≤S＜4	3	5	1	7	-1	-3	-5	-7
									4≤S＜5	5	3	7	1	-1	-3	-5	-7
5≤S＜6	5	7	3	1	-1	-3	-5	-7
									6≤S	7	5	3	1	-1	-3	-5	-7

As can be seen from fig. 6, the size of the search area determines the complexity of the operation, and when the area is large enough to include all constellation points (e.g., 64 constellation points in fig. 6), the complexity of the operation of the embodiment is comparable to that of the aforementioned lor architecture. Of course, the size of the search area may be predetermined or adjusted according to the communication condition. When the size of the area is predetermined, the first-level search value is determined by the search value selection circuit 410 according to the predetermined threshold (e.g., when the search value selection circuit 410 finds that the first-level search value corresponding to the current communication condition will be smaller than the predetermined threshold, the first-level search value is made equal to the predetermined threshold). When the size of the area is determined according to the communication condition, the indicator of the communication condition (i.e. the communication indicator) may be at least one of a signal-to-noise ratio, a received signal energy strength of a subcarrier, a channel energy strength, a channel correlation, a channel estimation accuracy, an interference energy strength, and the like.

As mentioned above, when the size of the search area is determined according to the communication condition, the size of the search area is determined by the maximum likelihood detection circuit 420 performing the zero forcing operation to obtain the zero forcing result

Point of impact on constellation diagram

The Signal-to-Noise Ratio (SNR) of time, which can be defined as γ, can be expressed as follows:

wherein

For noise energy, the estimation method is known in the art, and R is₁₁Is one of the elements of the upper triangular matrix R. The greater γ, the more reliable the result of the zero forcing operation of the maximum likelihood detection circuit 420, so that the smaller search value (or smaller search area) can be selected by the search value selection circuit 410; conversely, as γ decreases, the search value selection circuit 410 may select a larger search value (or larger search area)). According to the above concept, the present embodiment can define several thresholds T_g(G ═ 1,2, …, G), where T is_g<T_g+1When the value of gamma exceeds a certain threshold value T_gThe search value selection circuit 410 may use a smaller search value as shown in Table 2, where K is_G+1<K_G<…<K₂<K₁。

TABLE 2

In addition to using the snr (as shown in equation nine) as the communication indicator, the search value selection circuit 410 can select the first level search value (or the size of the search area) according to other communication indicators, as described above. For example, search value selection circuitry 410 may select a search value based on a channel correlation p, where p may be expressed as:

in the formula ten, h₁Is row 1 (Column), H of the channel response matrix H ^ H₂Is row 2 of H. The maximum likelihood detection circuit 420 is configured to detect when ρ is larger (i.e., 1/ρ is smaller)

The less reliable the result of the zero forcing operation, so that the search value selection circuit 410 selects the larger search value (or larger search area); conversely, as ρ is smaller (i.e., 1/ρ is larger), the search value selection circuit 410 selects a smaller search value (or smaller search area). To sum up, the first layer search value is a first search value when the communication pointer (which may be γ, 1/ρ or other indicators) is higher than a first threshold, the first layer search value is a second search value when the communication pointer is lower than the first threshold, the first search value is smaller than the second search value, and the communication status represented by the communication pointer higher than the first threshold is better than the communication status represented by the communication pointer lower than the first threshold.

As mentioned above, when the first-level search value is large enough, the nth bit set of all the first-level candidate signal values in the search range includes a set with a bit value of 1 and a set with a bit value of 0, so that the maximum likelihood detector 152 does not miss the first-level candidate signal values to be considered (as shown in fig. 6); in other words, if the nth bit Set lacks the Set with bit value 0 or the Set with bit value 1, the maximum likelihood detector 152 finds that the Set with bit value 0 in the nth bit Set does not exist but is a null Set (Empty Set) or the Set with bit value 1 does not exist but is a null Set when calculating the log-likelihood ratio of formula five, so that the maximum likelihood detector 152 cannot obtain correct information when calculating the log-likelihood ratio (the spirit of log-likelihood ratio is to compare the probability of bit 1 in each bit Set with the probability of bit 0), thereby requiring additional operation and further causing performance loss. However, even if the first layer candidate signal values are not omitted, since the second layer candidate signal values are obtained according to the first layer candidate signal values (as shown in formula seven), the distribution of the second layer candidate signal values may be irregular or concentrated in a small range, and there may still be a case where the second layer candidate signal values are omitted (i.e. there is the aforementioned empty set in the second layer candidate signal values). Therefore, as mentioned above, in the case that K first layer and K second layer candidate signal values have been found, the maximum likelihood detector 152 may further perform the maximum likelihood detecting step.

In the maximum likelihood detecting step, one embodiment of the step of determining whether to increase the P second-layer supplemental candidate signal values comprises: determining whether the bit value of the nth bit of each of the K second-level candidate signal values is the same, wherein n is an integer between 0 and (m-1), and m is the number of bits per second-level candidate signal value; and if the bit value of the nth bit of each of the K second-layer candidate signal values is judged to be the same, increasing the P second-layer supplementary candidate signal values. For example, as shown in fig. 8, the signal detecting circuit 150 may further include a candidate signal value supplementing unit 810 for determining whether to add the supplementary candidate signal values, an embodiment of the candidate signal value supplementing unit 810 may sum the nth bits of each candidate signal value to obtain a sum, and determine whether the nth bits have the same value according to the sum, where the sum represents that the nth bits have 0 when the sum is 0, and represents that the nth bits have 1 when the sum is K. Another embodiment of the candidate signal value supplementing unit 810 compares a predetermined bit value with the value of the nth bit of each of the candidate signal values, and determines whether the values of the nth bits are the same according to whether the comparison results are the same, all of which belong to the implementation category of the candidate signal value supplementing unit 810.

When the candidate signal value supplementing unit 810 determines that the candidate signal values should be supplemented, the candidate signal value supplementing unit 810 may further supplement the candidate signal values by: calculating K distances according to the K first-layer candidate signal values and the K second-layer candidate signal values (e.g., substituting each corresponding set of first-layer candidate signal values and second-layer candidate signal values into formula six to obtain the K distances); selecting a second-layer reference candidate signal value from the K second-layer candidate signal values according to the K distances, wherein the second-layer reference candidate signal value corresponds to a first-layer reference candidate signal value in the K first-layer candidate signal values; increasing the P second-layer supplementary candidate signal values according to the second-layer reference candidate signal value; and making the P first-layer supplementary candidate signal values all be the first-layer reference candidate signal values. For example, the step of selecting the second-layer reference candidate signal value comprises: calculating K distances according to the K first-layer candidate signal values and the K second-layer candidate signal values; the path formed by the first layer and the second layer candidate signal values that can find the minimum of the K distances is considered as an optimal path (as shown in fig. 9, where u is 1 or 0,

is the inverse value of u); and using the second layer candidate signal value under the best path as the second layer reference candidate signal value (as shown in FIG. 10)

) Corresponding to the firstThe layer reference candidate signal value is the first layer candidate signal value under the best path. For another example, the step of increasing the P second-layer supplementary candidate signal values according to the second-layer reference candidate signal values comprises: performing a table lookup operation to search pre-stored data according to the second-tier reference candidate signal values to increase the P second-tier supplemental candidate signal values, wherein each of the P second-tier supplemental candidate signal values is located in a horizontal or vertical direction (as shown in fig. 10) of the second-tier reference candidate signal values on a constellation map associated with a modulation scheme corresponding to the received signal, and the P value is not greater than the number of bits of the second-tier reference candidate signal values.

The step of adding the P second layer supplemental candidate signal values is further described below. Please refer to fig. 9 and 10, all of which are

(

The ith bit, which may be 0 or 1), is j for example (where j is 0 or 1), observing the data belonging to the best path

Drop points on the constellation diagram

Find the ith bit as

And is away from

(or with)

Nearest constellation point) as a second layer supplemental candidate signal value. The above search process is related to the design of the constellation diagram. Taking the constellation diagram of 64-QAM as an example in FIG. 10

Is located in a section

And is

Is located in the interval

When (at this time leave)

The bit value of the nearest constellation point is 010000) in order from the most significant bit to the least significant bit, and

the nearest constellation point with the highest inverted bit value is denoted b₀To and from

The nearest constellation point with the next highest value in the reverse direction is denoted b₁…, and go

The nearest constellation point with the inverted lowest bit value is denoted b₅The constellation points with inverted bit values can be used as the supplementary candidate signal values and must be located away from

Within the cross range centered on the nearest constellation point

The positional relationship of (A) is shown in tables 3 to 5 (in tables 3 to 5)

Is 1, however this is merely an example). Thus, by a table look-up operation or equivalentThe candidate signal value complementing unit 810 can find the ith bit as

And is away from the landing point

(or with the landing point)

The nearest constellation point).

TABLE 3

TABLE 4

TABLE 5

As mentioned above, for example, as shown in FIG. 11, if there are 9 possible second layer candidate signal values

Is located in a section

Is located in the interval

And at the point of falling

For all constellation points (or candidate signal values) in the search range of the reference, the highest bit is 0 and the second highest bit is 1, and the point is dropped by looking up the table 3

Constellation point (bit value 110010) and departure point with nearest and highest bit value of 1

The most recent constellation point (bit value 000001) with the next highest value of 0 can be found as b of FIG. 10₀And b₁As shown. After finding the two nearest constellation points (or the second layer of supplemental candidate signal values), the tree diagram of fig. 9 can have two more supplemental paths as shown in fig. 12 to compensate for the possibility of the absence of 1 and 0 in the highest and second highest bits, respectively. Please note that the second layer candidate signal value

May not be so regular as in fig. 11.

Referring to fig. 13, the maximum likelihood detector 152 may perform reverse expansion (first expansion at the first layer) with the aid of a Swap Operation (Swap Operation) determining unit 1310

Then spread on the second layer

). The swapping operation determining unit 1310 determines whether to perform a swapping operation according to the channel estimation signal, and outputs the channel estimation signal to the QR decomposition unit 530 when determining not to perform the swapping operation, and outputs a swapping signal of the channel estimation signal to the QR decomposition unit 530 when determining to perform the swapping operation. More specifically, one embodiment of the swapping operation determining unit 1310 determines the order of signal detection according to the channel response matrix H ^ of the channel estimation signal, for example, according to the energy of the matrix H ^ to determine the order. For example, assume that the first-to-expand signal is

(

Or

) The post-developed signal is

(

Or

) When the square of the absolute value of the 1 st column of the matrix H ^ is less than the square of the absolute value of the 2 nd column of the matrix H ^ the swapping operation determining unit 1310 swaps the 1 st column and the 2 nd column of the matrix H ^ to output the swapped signal of the channel estimation signal, so that the signal to be spread first is the swapped signal of the channel estimation signal

When the square of the absolute value of the 1 st row of the matrix H ^ is larger than the square of the absolute value of the 2 nd row of the matrix H ^ the swapping operation determining unit 1310 outputs the matrix H ^ so that the signal to be spread first is

In short, the swapping operation determining unit 1310 detects the signal with smaller channel energy to make the first layer search value larger, thereby reducing the possibility of missing the candidate signal value. However, if the signal detection circuit 150 is fixed for the first detection

(or the equal structure of the signal detection circuit 150 is fixed and the first detection is performed

) The swap operation determining unit 1310 is not necessary.

Please refer to fig. 14, which is a diagram illustrating another embodiment of the signal detection circuit 150. In this embodiment, the signal detection circuit 150 performs forward expansion through the first detection circuit 1410, and performs backward expansion through the second detection circuit 1420, wherein the first detection circuit 1410 is implemented by

One embodiment of the maximum likelihood detector 152, second detection circuit 1420 of fig. 5 is shown in fig. 15. In FIG. 15, the swapping unit 1510 swaps the 1 st and 2 nd rows of the channel response matrix H ^ to output the swapped signal of the channel estimation signal. It should be noted that the first level search value generated by the first detection circuit 1410 may be the same as or different from the first level search value generated by the second detection circuit 1420, for example, the first level search value K generated by the first and

second detection circuits

1410, 1420 are both the predetermined threshold or greater, or the search values K1, K2 generated according to the communication pointer, respectively, neither of the two search values K1, K2 is smaller than the predetermined threshold, and neither of the search values K, K1, K2 is smaller than the predetermined threshold, so that there is no need to supplement the candidate signal value, which is not limited.

Compared to the embodiment of fig. 14, which needs to calculate the log-likelihood ratio for (K + K) (or (K1+ K2)), the embodiment of fig. 8 only needs to calculate the log-likelihood ratio for (K + P) possibilities. Generally, (K + P) must be less than (K + K) (or (K1+ K2)), so the embodiment of FIG. 8 can further reduce the complexity of the operation and reduce the circuit area.

Since one of ordinary skill in the art can readily appreciate details and variations of the embodiments of fig. 14 and 15 from the foregoing description, repeated and redundant descriptions are omitted herein without affecting the disclosed requirements and the feasibility of the embodiments of fig. 14 and 15. It is noted that, although steps may be described herein without limitation to the order of execution and while steps may be described as possible, the steps may be comprised of sub-steps, and redundant descriptions are omitted herein since such features are known to those skilled in the art and will be apparent from the disclosure herein.

In summary, compared to the prior art, the present invention provides a low complexity solution, thereby achieving the features of low time delay, low computation complexity, low computation power, low circuit area, and no substantial performance loss.

Although the embodiments of the present invention have been described above, these embodiments are not intended to limit the present invention, and those skilled in the art can make variations on the technical features of the present invention according to the explicit or implicit contents of the present invention, and all such variations may fall within the scope of the patent protection sought by the present invention.

Description of the symbols

100 wireless signal receiver

110 discrete Fourier transform circuit (DFT circuit)

120 reference signal extraction circuit

130 channel estimation circuit

140 data signal extraction circuit

150 signal detection circuit

152 maximum likelihood detector

160 decoding circuit

Candidate signal value

410 search value selection circuit

420 maximum likelihood detection circuit

522 first layer candidate signal value decision unit

524 second layer candidate signal value determining unit

526 logarithm approximate ratio calculation unit

530 QR decomposition unit

540 Signal generating Unit

Point of impact on constellation diagram

810 candidate signal value supplementing unit

Candidate signal value

Point of impact on constellation diagram

b₀、b₁、b₂、b₃、b₄、b₅Nearest constellation point with inverted bit value

1310 swap operation determination unit

1410 first detection circuit

1420 second detection circuit

1510 swap units.

Claims

1. A maximum likelihood detector for performing a maximum likelihood detection based on a received signal or its derivative and based on a channel estimation signal or its derivative, comprising:

a search value selection circuit for determining a first layer search value, wherein the first layer search value is not less than a preset threshold value; and

a maximum likelihood detection circuit for performing at least the following steps:

selecting K first layer candidate signal values according to the received signal or a derivative thereof, the channel estimation signal or a derivative thereof, and the first layer search value, wherein K is a positive integer greater than 1;

calculating K second-layer candidate signal values according to the K first-layer candidate signal values;

judging whether to increase P second-layer supplementary candidate signal values according to the K second-layer candidate signal values, and generating a judgment result according to the P second-layer supplementary candidate signal values, wherein P is a positive integer;

if the judgment result is yes, adding the P second-layer supplementary candidate signal values, selecting P first-layer supplementary candidate signal values according to the P second-layer supplementary candidate signal values, and calculating a log-approximate ratio according to the K first-layer candidate signal values, the K second-layer candidate signal values, the P first-layer supplementary candidate signal values and the P second-layer supplementary candidate signal values; and

if the result is negative, calculating the approximate logarithm ratio according to the K first-layer candidate signal values and the K second-layer candidate signal values.

2. The maximum likelihood detector of claim 1 wherein the step of determining whether to increase the P second layer supplemental candidate signal values comprises:

determining whether the bit value of the nth bit of each of the K second layer candidate signal values is the same according to the K second layer candidate signal values, wherein n is an integer between 0 and (m-1), and m is the bit number of each second layer candidate signal value; and

if the bit value of the nth bit of each of the K second-layer candidate signal values is determined to be the same, the P second-layer complementary candidate signal values are added.

3. The maximum likelihood detector of claim 1 wherein the steps of adding the P second layer supplemental candidate signal values and selecting the P first layer supplemental candidate signal values comprise:

calculating K distances according to the K first-layer candidate signal values and the K second-layer candidate signal values;

selecting a second-layer reference candidate signal value from the K second-layer candidate signal values according to the K distances, wherein the second-layer reference candidate signal value corresponds to a first-layer reference candidate signal value in the K first-layer candidate signal values;

increasing the P second-layer supplementary candidate signal values according to the second-layer reference candidate signal value; and

the P first-tier supplemental candidate signal values are all the first-tier reference candidate signal values.

4. The maximum likelihood detector of claim 3, wherein the value of P is not greater than the number of bits of the second tier reference candidate signal values.

5. The maximum likelihood detector of claim 3, wherein the step of selecting the second layer reference candidate signal values comprises: the second-tier reference candidate signal value is selected from the K second-tier candidate signal values according to the smallest of the K distances.

6. The maximum likelihood detector of claim 3 wherein the step of increasing the P second tier supplemental candidate signal values based on the second tier reference candidate signal values comprises: a look-up table is performed to look up pre-stored data according to the second tier reference candidate signal values, thereby increasing the P second tier supplemental candidate signal values.

7. The maximum likelihood detector of claim 3 wherein each of the P second layer supplemental candidate signal values is located horizontally or vertically with respect to the second layer reference candidate signal value on a constellation map associated with the modulation scheme corresponding to the received signal.