MIMO detection maximum likelihood estimation method and Isin-MIMO detector

Abstract

The invention discloses an MIMO (Multiple Input Multiple Output) detection maximum likelihood estimation method, which comprises the following steps of: acquiring a complex-valued transmitting signal and a complex-valued channel matrix between a receiving antenna and a transmitting antenna to obtain a complex-valued signal at a receiving end of a base station; according to the complex value transmitting signal, the complex value signal of the base station receiving end and the complex value channel matrix between the receiving antenna and the transmitting antenna, an optimal estimation signal is obtained, and MIMO maximum likelihood detection is mapped to an Isin model; mIMO maximum likelihood estimation is carried out through an Isin-MIMO detector, and a user emission signal is restored. The invention also discloses an Isin-MIMO (Multiple Input Multiple Output) detector. According to the method, the technical problem of how to accurately carry out MIMO detection maximum likelihood estimation in real time and output a result is solved.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of communication, in particular to a MIMO detection maximum likelihood estimation method and an Ising-MIMO detector. BACKGROUND

[0002] A signal detection device in a multiple-input multiple-output (MIMO) detector communication system is mainly used to improve the capacity and reliability of a wireless communication system. The MIMO detector receives and detects an infinite signal through a plurality of antenna arrays, can acquire channel parameters such as a multipath time delay power spectrum, supports pre-processing of the signal to optimize the detection performance, decomposes a data stream into a plurality of sub-streams, and combines the signal at a receiving end by using a specific algorithm such as maximum likelihood detection and spherical detection, so that the transmission efficiency and anti-interference capability are significantly improved. Traditional digital algorithms such as ML detection have an exponential increase in complexity with the number of antennas, and it is difficult to meet the real-time demand of large-scale MIMO. Existing analog calculation methods such as an analog circuit based on BP detection have the problems of limited circuit size, high energy consumption and low mapping flexibility. Therefore, a MIMO detection maximum likelihood estimation method and an Ising-MIMO detector are urgently needed to solve the technical problem of how to perform real-time and accurate MIMO detection maximum likelihood estimation and output the result. SUMMARY

[0003] The main purpose of the application is to provide a MIMO detection maximum likelihood estimation method and an Ising-MIMO detector, which aims to solve the technical problem of how to perform real-time and accurate MIMO detection maximum likelihood estimation and output the result.

[0004] To achieve the above purpose, the application provides a MIMO detection maximum likelihood estimation method, which comprises the following steps:

[0005] S1, acquiring a complex-valued transmission signal and a complex-valued channel matrix between receiving antennas and transmitting antennas to obtain a base station receiving end complex-valued signal;

[0006] S2, acquiring an optimal estimated signal according to the complex-valued transmission signal, the base station receiving end complex-valued signal and the complex-valued channel matrix between the receiving antennas and the transmitting antennas, and mapping the MIMO maximum likelihood detection to an Ising model;

[0007] S3, performing MIMO maximum likelihood estimation by using an Ising-MIMO detector to restore the user transmission signal.

[0008] In one of the preferred schemes, the base station receiving end complex-valued signal is:

[0009] y=Hx+n

[0010] wherein, y is the complex-valued signal received by the base station receiver, H is the complex-valued channel matrix between the receiving antenna and the transmitting antenna, x is the complex-valued transmitting signal, and n is the complex-valued noise vector.

[0011] In one preferred solution, the step S2 obtains the best estimated signal according to the complex-valued transmitting signal, the complex-valued signal received by the base station receiver, and the complex-valued channel matrix between the receiving antenna and the transmitting antenna, and specifically comprises:

[0012]

[0013] wherein, is the best estimated signal, y is the complex-valued signal received by the base station receiver, H is the complex-valued channel matrix between the receiving antenna and the transmitting antenna, and x is the complex-valued transmitting signal.

[0014] In one preferred solution, the step S2 maps the MIMO maximum likelihood estimation into the Ising model based on the full-connection spin network topology.

[0015] In one preferred solution, the step S2 maps the MIMO maximum likelihood detection into the Ising model, and specifically comprises:

[0016]

[0017] wherein, H(σ) is the system energy of the Ising model, σ is the quantum spin of the Ising model, J is the coupling coefficient of the Ising model, h is the local field coefficient, i and j are spin indexes, and N is the total number of spins, J ij is the coupling coefficient between the spin index i and the spin index j, σ i is the i-th quantum spin, σ j is the j-th quantum spin.

[0018] In one preferred solution, the step S3 specifically comprises:

[0019] S31, obtaining the random number in the voltage domain according to the cross array of the memristor, and changing the conductance value of the memristor;

[0020] S32, randomizing the voltage of each row of the cross array of the memristor, that is, starting iteration with a random initial spin state through the Ising model, and the spin value is represented by the voltage of each row of the cross array of the memristor;

[0021] S33, calculating the spin local energy field contribution and the energy difference caused by spin flipping of the cross array of the memristor in the simulated current domain;

[0022] S44, calculating the spin flipping probability, and making flipping judgment according to the random number in the voltage domain;

[0023] S45, set the number of iterations and temperature threshold, when reaching the number of iterations or temperature threshold, output the final spin state, and get the estimated signal of MIMO detection maximum likelihood estimation through linear transformation.

[0024] In one preferred embodiment, the spin local energy field contribution is:

[0025]

[0026] Wherein, LF i is the local energy field contribution value of spin index i, i, j is the spin index, J ij is the coupling coefficient between spin index i and spin index j, σ i is the i-th quantum spin, h i is the conductance value of the last row of the RRAM array.

[0027] In one preferred embodiment, the energy difference value caused by spin flip is:

[0028] ΔE i = -2σ i LF i

[0029] Wherein, ΔE i is the energy difference value caused by spin flip.

[0030] In one preferred embodiment, the step S44 is specifically:

[0031] Calculate the self-flip probability; the self-flip probability is:

[0032]

[0033] Flip_Prob is the self-flip probability, e is the natural constant, and T is the temperature value of the current iteration state.

[0034] Flip judgment is made according to the voltage domain random number; specifically:

[0035]

[0036] Wherein, Flip is the judgment result of spin flip, and r is the voltage domain random number.

[0037] An Ising-MIMO detector comprising the MIMO detection maximum likelihood estimation method comprises a cross array of memristors, a column driving circuit, a row driving circuit, a transimpedance amplifier, a level switching circuit, a control circuit, a random number generator, an energy difference calculation circuit, an adaptive calibration circuit, a flip circuit and an output buffer circuit; the cross array of memristors is connected with the column driving circuit, the row driving circuit, the transimpedance amplifier and the level switching circuit respectively, the control circuit is connected with the level switching circuit, the transimpedance amplifier and the flip circuit respectively, the transimpedance amplifier is connected with the adaptive calibration circuit, the random number generator and the energy difference calculation circuit respectively, the random number generator is connected with the flip circuit, the flip circuit is connected with the level switching circuit and the energy difference calculation circuit respectively, and the energy difference calculation circuit is connected with the output buffer circuit.

[0038] In the technical scheme, the MIMO detection maximum likelihood estimation method comprises the following steps: obtaining a complex-valued transmitting signal and a complex-valued channel matrix between receiving antennas and transmitting antennas to obtain a base station receiving end complex-valued signal; obtaining an optimal estimated signal according to the complex-valued transmitting signal, the base station receiving end complex-valued signal and the complex-valued channel matrix between the receiving antennas and the transmitting antennas, and mapping the MIMO maximum likelihood detection to an Ising model; and performing MIMO maximum likelihood estimation through an Ising-MIMO detector to restore a user transmitting signal. The technical scheme solves the technical problem of how to perform real-time and accurate MIMO detection maximum likelihood estimation and output results.

[0039] In the technical scheme, the MIMO detection maximum likelihood estimation method comprises the following steps: obtaining a complex-valued transmitting signal and a complex-valued channel matrix between receiving antennas and transmitting antennas to obtain a base station receiving end complex-valued signal; obtaining an optimal estimated signal according to the complex-valued transmitting signal, the base station receiving end complex-valued signal and the complex-valued channel matrix between the receiving antennas and the transmitting antennas, and mapping the MIMO maximum likelihood detection to an Ising model; and performing MIMO maximum likelihood estimation through an Ising-MIMO detector to restore a user transmitting signal. The technical scheme solves the technical problem of how to perform real-time and accurate MIMO detection maximum likelihood estimation and output results.

[0040] In the technical scheme, the MIMO detection maximum likelihood estimation method comprises the following steps: obtaining a complex-valued transmitting signal and a complex-valued channel matrix between receiving antennas and transmitting antennas to obtain a base station receiving end complex-valued signal; obtaining an optimal estimated signal according to the complex-valued transmitting signal, the base station receiving end complex-valued signal and the complex-valued channel matrix between the receiving antennas and the transmitting antennas, and mapping the MIMO maximum likelihood detection to an Ising model; and performing MIMO maximum likelihood estimation through an Ising-MIMO detector to restore a user transmitting signal. The technical scheme solves the technical problem of how to perform real-time and accurate MIMO detection maximum likelihood estimation and output results. BRIEF DESCRIPTION OF DRAWINGS

[0041] In order to more clearly illustrate the technical scheme in the embodiments of the present application or the prior art, the drawings needed in the following description of the embodiments or the prior art will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained according to the structures shown in the drawings without creative labor for those skilled in the art.

[0042] Figure 1A schematic diagram of a MIMO detection maximum likelihood estimation method according to an embodiment of the present application;

[0043] Figure 2 A schematic diagram of an Ising-MIMO detector according to an embodiment of the present application;

[0044] Figure 3 A communication schematic diagram of an Ising-MIMO detector according to an embodiment of the present application;

[0045] Figure 4 A schematic diagram of a random number generator according to an embodiment of the present application;

[0046] Figure 5 A schematic diagram of the relationship between bit error rate and signal-to-noise ratio under different architectures according to an embodiment of the present application.

[0047] The implementation, functional features and advantages of the present application will be further described with reference to the accompanying drawings. DETAILED DESCRIPTION

[0048] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0049] It should be noted that all directional indications (such as up, down, …) in the embodiments of the present application are only used to explain the relative positional relationship, movement condition, etc. between components in a certain posture (as shown in the drawings), and if the certain posture changes, the directional indications will also change accordingly.

[0050] In addition, the description such as "first", "second" and the like in the present application is only for the purpose of description, and cannot be understood as indicating or implying the relative importance of the indicated technical features or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first", "second" can explicitly or implicitly include at least one of the features.

[0051] In addition, the technical solutions of each embodiment of the present application can be combined with each other, but it must be based on the fact that a person skilled in the art can realize it, and when the combination of technical solutions appears contradictory or unachievable, it should be considered that the combination of technical solutions does not exist, and is not within the scope of protection required by the present application.

[0052] Referring to Figures 1-5 According to an aspect of the present application, a MIMO detection maximum likelihood estimation method is provided, wherein the MIMO detection maximum likelihood estimation method comprises the following steps:

[0053] S1. Obtain the complex-valued transmitted signal and the complex-valued channel matrix between the receiving antenna and the transmitting antenna to obtain the complex-valued signal at the base station receiver.

[0054] S2. Based on the complex-valued transmitted signal, the complex-valued signal at the base station receiver, and the complex-valued channel matrix between the receiving antenna and the transmitting antenna, obtain the optimal estimated signal and map the MIMO maximum likelihood detection to the Ising model.

[0055] S3. Perform MIMO maximum likelihood estimation using the Ising-MIMO detector to reconstruct the user's transmitted signal.

[0056] Specifically, in this embodiment, the complex-valued signal received by the base station is:

[0057] y = Hx + n

[0058] Where y is the complex-valued signal received by the base station receiver, H is the complex-valued channel matrix between the receiving antenna and the transmitting antenna, x is the complex-valued transmitted signal, and n is the complex-valued noise vector; the present invention reconstructs the user's transmitted signal by performing symbol detection on the complex-valued signal.

[0059] Specifically, in this embodiment, the MIMO maximum likelihood detection problem is transformed into a QUBO problem and mapped to the Ising model based on a fully connected spin network topology. Maximum likelihood detection is a theoretically optimal signal detection algorithm that obtains the optimal estimated signal by comparing the Euclidean distance between the H and x signals obtained after all possible signal combinations x are transmitted through the channel and the received y signal. Step S2 obtains the optimal estimated signal based on the complex-valued transmitted signal, the complex-valued signal at the base station receiver, and the complex-valued channel matrix between the receiving antenna and the transmitting antenna. Specifically:

[0060]

[0061] in, For the optimal estimated signal, y is the complex-valued signal received by the base station receiver, H is the complex-valued channel matrix between the receiving antenna and the transmitting antenna, and x is the complex-valued transmitted signal;

[0062] The maximum likelihood detection problem can be mapped to the Ising model through transformation. Step S2 maps MIMO maximum likelihood detection to the Ising model, specifically as follows:

[0063]

[0064] Where H(σ) is the system energy of the Ising model, σ is the quantum spin of the Ising model, J is the coupling coefficient of the Ising model, h is the local field coefficient, i and j are spin indices, N is the total spin number, and J ijis the coupling coefficient between spin index i and spin index j, σ i is the i-th quantum spin, σ j is the j-th quantum spin; wherein, J and h are calculated from MIMO complex-valued channel matrix H and complex-valued signal y received at the base station receiver, J∝H T H, h∝-2y T H.

[0065] Specifically, in the embodiment, the step S3 is specifically:

[0066] S31, obtaining a random number in the voltage domain according to the memristor cross array, and changing the conductance value of the memristor;

[0067] S32, randomizing the voltage of each row of the memristor cross array, that is, starting iteration with a random initial spin state through the Ising model, and the spin value is represented by the voltage of each row of the memristor cross array;

[0068] S33, calculating the spin local energy field contribution of the memristor cross array in the analog current domain and the energy difference caused by spin flipping;

[0069] S44, calculating the spin flipping probability and making flipping judgment according to the random number in the voltage domain;

[0070] S45, setting the iteration number and the temperature threshold value, and outputting the final spin state when the iteration number or the temperature threshold value is reached, and obtaining the estimated signal of MIMO detection maximum likelihood estimation through linear transformation.

[0071] Specifically, in the embodiment, the step S31 is specifically: obtaining a random number in the voltage domain according to the memristor cross array, and the conductance value of RRAM before programming can be considered as randomly distributed due to process deviation; first, optionally selecting a row of the RRAM array, while grounding the other rows, applying a voltage to the selected row, and then optionally selecting two columns; due to the asymmetry caused by process deviation, the cumulative current of the selected two columns is inconsistent, and a stable unclonable output response can be obtained through the current comparator, and different responses can be output when different rows are selected, and the random number in the voltage domain can be obtained through the transimpedance amplifier to convert the current signal; the output current of each column in the memristor cross array is:

[0072]

[0073] wherein, I j is the output current of the j-th column of the memristor cross array, V i is the voltage value of the i-th row, G ij is the conductance value of the Memristor in the i-th row and the j-th column, V h is the voltage of the last row, G hMemristor conductance value of the last row and the jth column;

[0074] The coupling coefficient J and the local field coefficient h of the Ising model are calculated according to the complex value signal y received by the MIMO base station receiver end and the complex value channel matrix H, voltage pulse signals are applied to the RRAM array by the column driving circuit and the row driving circuit to program the RRAM conductance value, the conductance step is coarsely adjusted by adjusting the pulse width or amplitude, and fine adjustment can be realized by multiple small amplitude short pulses, the conductance value is increased by applying a forward pulse, and the conductance value is reduced by applying a reverse pulse, wherein the conductance values of 1-n rows of the RRAM array are set as the coupling coefficient J between the spin index i and the spin index j ij , and the conductance value of the last row is set as h j .

[0075] Specifically, in the embodiment, the step S1 further includes: setting an initial temperature, clearing the iteration number, and setting a stop condition including temperature, iteration number and energy as threshold values.

[0076] Specifically, in the embodiment, the Ising model starts iteration with a random initial spin state, and the spin value (+1, -1) is represented by the row voltage; wherein +VDD represents the spin value +1, -VDD represents the spin value -1, a random initial level (±VDD) is applied to 1-n rows before iteration starts, +VDD level is set for the last row, and initial iteration temperature T0 and iteration number are cleared; a spin is randomly selected, and the local energy field contribution is realized in the analog current domain by using the memristor cross array; the local energy field in the Ising model is defined as the contribution of a single spin to the Hamiltonian of the system, and the spin local energy field contribution is:

[0077]

[0078] wherein LF i is the local energy field contribution value of spin index i, i and j are spin indexes, J ij is the coupling coefficient between spin index i and spin index j, σ i is the i-th quantum spin, h i is the conductance value of the last row of the RRAM array, since the level of this row is set to +VDD, VDD is taken as 1V for simplifying calculation, therefore, the output current LF i of any column is the local field energy of spin i;

[0079] By using the analog domain calculation method, the Hamiltonian calculation is realized in the memristor cross array, the difference between the current state and the previous state system energy is obtained, and any spin σ iThe energy change in the system caused by the flip from +VDD to -VDD or from -VDD to +VDD, i.e. the energy difference caused by the spin flip, is:

[0080] ΔE i = -2σ i LF i

[0081] where ΔE i is the energy difference caused by the spin flip; the calculation of the energy change value is completed by the transimpedance amplifier and the energy difference calculation circuit, and the output current LF i of the RRAM array is converted into a voltage signal by the transimpedance amplifier and then enters the energy difference calculation circuit for calculation.

[0082] Specifically, in the present embodiment, the step S44 specifically comprises:

[0083] calculating the spin flip probability; the spin flip probability is:

[0084]

[0085] Flip_Prob is the spin flip probability, e is the natural constant, and T is the temperature value of the current iteration state, but it is difficult to directly perform exponential calculation in the circuit, so the exponential calculation part can be linearly approximated according to the accuracy requirement, and then a simple linear operation circuit is realized through an operational amplifier, for example:

[0086]

[0087] wherein, since the temperature T > 0, if the energy change value ΔE i is negative, i.e. the system energy will decrease after the spin flip, the spin flip probability is 1; when the energy change is positive, the spin flip probability is calculated by the above formula, i.e. the system is allowed to jump in the direction that may increase the system energy, so as to jump out of the local minimum;

[0088] performing flip determination according to the voltage domain random number; specifically:

[0089]

[0090] wherein, Flip is the determination result of the spin flip, and r is the voltage domain random number.

[0091] Specifically, in the embodiment, whether spin flip is needed is determined according to the random number of the voltage domain, if spin flip is needed, only the level value of the row needs to be changed through the control circuit, and the level is switched between +VDD and -VDD through the level switching circuit and the control circuit; the change of temperature and the control of the iteration process are controlled through the control circuit to ensure the normal work of the circuit, after a certain number of iterations or the temperature reaches the preset threshold, the iteration process is ended, the final spin state is output, and the user transmission signal is restored through a simple linear transformation; if the stop condition is not reached, the process returns to step S32 and continues iteration; the output spin configuration is reconstructed into MIMO symbols through linear mapping, specifically:

[0092]

[0093] wherein σ is a spin configuration vector, Q is a linear mapping matrix, for BPSK modulation, Q is a unit matrix, and for M-QAM modulation, the linear mapping matrix is:

[0094]

[0095]

[0096] wherein I 2K is a unit matrix of size 2K, q is an amplitude mapping coefficient vector, M is the order of quadrature amplitude modulation, and K is the number of MIMO transmitting end users;

[0097] Finally, the received symbol is compared with the user transmission signal x', and the symbol error rate (SER) and the bit error rate (BER) are calculated to evaluate the performance of the detector.

[0098] Specifically, in the embodiment, referring to Figure 5 shows the horizontal comparison of the bit error rates of various algorithms when the number of antennas of the receiving end and the transmitting end is configured as 16*16, wherein under BPSK modulation, the bit error rate performance obtained by using the MIMO detection maximum likelihood estimation method of the application is better than that of the traditional detection algorithms ZF and MMSE, and when the SNR increases to 12 dB, the bit error rate is reduced to 0; under QPSK modulation, the bit error rate of the MIMO detection maximum likelihood estimation method of the application is close to that of the MMSE algorithm in a low SNR scenario, and is slightly worse than that of the MMSE algorithm in a high SNR scenario, but is still better than that of the ZF detection.

[0099] According to another aspect of the application, referring to Figure 2The application provides an Ising-MIMO detector, which comprises a memristor cross array, a column driving circuit, a row driving circuit, a transimpedance amplifier, a level switching circuit, a control circuit, a random number generator, an energy difference calculation circuit, an adaptive calibration circuit, a flip circuit and an output buffer circuit; the memristor cross array is connected with the column driving circuit, the row driving circuit, the transimpedance amplifier and the level switching circuit respectively; the control circuit is connected with the level switching circuit, the transimpedance amplifier and the flip circuit respectively; the transimpedance amplifier is connected with the adaptive calibration circuit, the random number generator and the energy difference calculation circuit respectively; the random number generator is connected with the flip circuit; the flip circuit is connected with the level switching circuit and the energy difference calculation circuit respectively; and the energy difference calculation circuit is connected with the output buffer circuit; the memristor cross array is used for storing coupling coefficients and realizing high-parallel multiply-accumulate operation; the column driving circuit and the row driving circuit are used for writing coupling coefficients; the transimpedance amplifier is used for converting local energy field current signals of each column into a transimpedance amplifier of a voltage signal; the control circuit is used for controlling spin flip, temperature reduction and state iteration; the random number generator is used for generating random numbers in combination with a memristor process deviation; the energy difference calculation circuit is used for calculating Hamiltonian change; the adaptive calibration circuit is used for compensating for memristor device drift error online; the flip circuit is used for assisting Hamiltonian iteration to jump out of a local minimum state; and the output buffer circuit is used for outputting a final spin state.

[0100] The above is only the preferred embodiment of the application, and does not limit the patent scope of the application, and any equivalent structural transformation made under the inventive concept of the application, or direct / indirect application in other related technical fields is included in the patent protection scope of the application.

Claims

1. A method of MIMO detection maximum likelihood estimation, characterized in that, The method comprises the following steps: S1, obtaining a complex-valued transmitting signal and a complex-valued channel matrix between receiving antennas and transmitting antennas to obtain a base station receiving end complex-valued signal; S2, obtaining an optimal estimated signal according to the complex-valued transmitting signal, the base station receiving end complex-valued signal and the complex-valued channel matrix between the receiving antennas and the transmitting antennas, and mapping the MIMO maximum likelihood detection to an Ising model; S3, performing MIMO maximum likelihood estimation through the Ising-MIMO detector to restore the user transmitting signal.

2. The method of claim 1, wherein, The base station receiving end complex-valued signal is: y = Hx + n Wherein, y is a complex-valued signal received by the base station receiving end, H is a complex-valued channel matrix between receiving antennas and transmitting antennas, x is a complex-valued transmitting signal, and n is a complex-valued noise vector.

3. The method of claim 1 or 2, wherein, The step S2 obtains an optimal estimated signal according to the complex-valued transmitting signal, the base station receiving end complex-valued signal and the complex-valued channel matrix between the receiving antennas and the transmitting antennas, and specifically comprises: wherein is the best estimate signal, y is the complex received signal at the base station receiver, H is the complex channel matrix between the receive antennas and the transmit antennas, and x is the complex transmitted signal.

4. The method of claim 1-2, wherein, The step S2 maps the MIMO maximum likelihood estimation to the Ising model based on the full-connection spin network topology.

5. The method of claim 1-2, wherein, The step S2 maps the MIMO maximum likelihood detection to the Ising model, and specifically comprises: where H(σ) is the system energy of the Ising model, σ is the quantum spin of the Ising model, J is the coupling coefficient of the Ising model, h is the local field coefficient, i and j are spin indexes, and N is the total number of spins, J ij is the coupling coefficient between the spin index i and the spin index j, σ i is the i th quantum spin, and σ j is the j th quantum spin.

6. The method of claim 1-2, wherein, The step S3 specifically comprises: S31, obtaining random numbers in a voltage domain according to the memristor cross array, and changing the conductance value of the memristor; S32, randomizing the voltage of each row of the memristor cross array, that is, starting iteration with a random initial spin state through the Ising model, and the spin value is represented by the voltage of each row of the memristor cross array; S33, calculating the spin local energy field contribution of the memristor cross array in the analog current domain and the energy difference value caused by spin flipping; S44, calculating the spin flipping probability and performing flipping judgment according to the random numbers in the voltage domain; S45, setting the iteration number and the temperature threshold value, and outputting the final spin state when the iteration number or the temperature threshold value is reached, and obtaining the estimated signal of the MIMO detection maximum likelihood estimation through linear transformation.

7. The method of claim 6, wherein, The spin local energy field contribution is: wherein, LF i is the local energy field contribution value for spin index i, i, j are spin indices, J ij is the coupling coefficient between spin index i and spin index j, σ i is the i-th quantum spin, h i is the conductance value of the last row of the RRAM array.

8. The method of claim 7, wherein, The energy difference value caused by spin flipping is: ΔE i = -2σ i LF i where ΔE is the energy difference caused by spin flipping. i is the energy difference caused by spin flipping.

9. The method of claim 7, wherein, The step S44 specifically comprises: Calculating the self-flipping probability; the self-flipping probability is: Flip_Prob is the self-flipping probability, e is a natural constant, and T is the temperature value of the current iteration state; Performing flipping judgment according to the random numbers in the voltage domain; specifically: Wherein, Flip is the judgment result of spin flipping, and r is the random number in the voltage domain.

10. An Ising-MIMO detector comprising a MIMO detection maximum likelihood estimation method according to any one of claims 1-9, characterized in that, It comprises: The memristor cross array, the column driving circuit, the row driving circuit, the transimpedance amplifier, the level switching circuit, the control circuit, the random number generator, the energy difference value calculation circuit, the adaptive calibration circuit, the flipping circuit and the output buffer circuit; the memristor cross array is connected with the column driving circuit, the row driving circuit, the transimpedance amplifier and the level switching circuit respectively, the control circuit is connected with the level switching circuit, the transimpedance amplifier and the flipping circuit respectively, the transimpedance amplifier is connected with the adaptive calibration circuit, the random number generator and the energy difference value calculation circuit respectively, the random number generator is connected with the flipping circuit, the flipping circuit is connected with the level switching circuit and the energy difference value calculation circuit respectively, and the energy difference value calculation circuit is connected with the output buffer circuit.

Images