RFID-based carrier cancellation search method and RFID reading device

Abstract

The present application relates to the field of reader, specifically to a carrier cancellation search method based on RFID and RFID reading equipment. The method comprises the following steps: a logic unit receives a control component and judges whether the reflected power is less than a set threshold; if the reflected power is less than the set threshold, the reflected power is continuously received, otherwise, the adjustable capacitance value of the control component is adjusted and recorded, and the adjusted reflected power is obtained. Through marking and resetting the front and rear reflected power, it is determined whether to continue searching, and the initial value of the capacitance that cannot converge can be exited in time, avoiding invalid search and reducing search time; and an improved momentum method is proposed, so that the adjustable capacitance value updated each time converges at the same speed towards the direction that is most easily converged, and the adjustable capacitance value that meets the conditions can be found more stably.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of RFID technology, in particular to a carrier wave elimination search method based on RFID and an RFID reading device. BACKGROUND

[0002] The reflected power is the sum of all the power output by the device to the capacitive plate or inductive coil, including the effective power eventually absorbed by the load and the reflected power not absorbed. When the load is not in ideal matching state with the device, part of the incident power cannot be absorbed by the load, but is consumed in the output circuit of the device, which is called reflected power. The carrier wave is a radio wave carrying RFID information, which can be emitted by the antenna connected to the RFID reader or the antenna connected to the tag, and its function is to transmit information through radio waves.

[0003] When the UHF RFID reader reads the signal of the RFID tag, the carrier wave is transmitted from the transmitting end, and the receiving end receives the response signal sent by the RFID tag. At this time, the carrier wave signal of the transmitting end will leak to the receiving end circuit, and the signal strength is often greater than that of the tag response signal, thereby interfering with the signal of the RFID tag response and reducing the receiving sensitivity of the reader. In order to suppress the carrier leakage, the phase and amplitude of the reflected signal need to be controlled by adjusting the adjustable capacitance value in the circuit. When the cancellation signal and the reflected signal have the same amplitude and a phase difference of 180°, the influence of the leaked carrier wave can be theoretically eliminated, thereby achieving the purpose of carrier suppression.

[0004] The method of carrier wave elimination is generally divided into two kinds. One is to change the internal circuit of the reader and increase or redesign the structure to achieve the purpose of suppressing carrier leakage, but this often causes the circuit structure of the reader to be complicated and increases the hardware cost. The other is to adjust the adjustable capacitance value in the circuit to generate a cancellation signal before reading the tag signal, such as the traversal method or the bisection method, to achieve the purpose of carrier elimination. However, the traversal search or bisection method often has a large amount of calculation, and each search takes more than a few hundred milliseconds, which is slow in response and poor in real-time, making it difficult to effectively receive signals in a high dynamic environment. Or adjust the adjustable capacitance to change the reflected power one by one to find the combination with the most power offset, but the hit probability is low.

[0005] In the prior art, CN 111178103 A "A method and system for eliminating the carrier of an ultra-high frequency RFID reader" eliminates the carrier by obtaining the carrier cancellation control signal data corresponding to the ultra-high frequency RFID reader at different transmission powers, antenna port load impedances and temperatures. It has high real-time performance, but the collected data covers a limited scene and cannot cope with environmental changes once the environment changes; CN 107786218 A "An ultra-high frequency RFID radio frequency front-end carrier suppression circuit" designs a carrier suppression circuit composed of an automatic gain circuit, a carrier cancellation circuit and a phase-locked loop circuit. The circuit can effectively reduce the influence of the inverse frequency mixing, reduce the direct current bias of the demodulator, and improve the receiving sensitivity of the reader receiver. However, the additional circuit increases the cost of the equipment, and the stability is difficult to guarantee. SUMMARY

[0006] The present application solves the problems of complex reader circuit structure, low search method hit rate, or large amount of calculation leading to long time consumption. The present application proposes a carrier cancellation search method based on RFID and an RFID reading device without changing the original circuit.

[0007] The technical scheme of the present application is as follows:

[0008] A carrier cancellation search method based on RFID includes the following calculations:

[0009] A1: Set power threshold r, incremental number threshold n p , decremental number threshold n q and cycle threshold c;

[0010] A2: Adjust the adjustable capacitance value of the control component, the adjustable capacitance value is used to adjust the reflected power, the adjustable capacitance value is generated by three variable capacitors, and the three adjustable capacitance values are c in , c len and c out , set Y group of capacitance initial values

[0011] A3: Take the adjustable capacitance value as a three-dimensional coordinate, c v is a coordinate point, and take the capacitance initial value as the coordinate origin, calculate the gradient by the function get_g(), obtain the gradient vector g v =[g in ,g len ,g out ], when searching in a loop, the update of the gradient vector g v guides the update of the adjustable capacitance value;

[0012] A4: The reflected power measurement formula is: p = get_power(c v), and determine whether the reflected power is less than the power threshold r:

[0013] A41: If the reflected power p now at the current time is less than the power threshold r, the logic unit stops searching for the reflected power, adjusts the adjustable capacitance value to the reflected power p now , and ends the process; otherwise, step A42 is performed. vnow

[0014] A42: According to the magnitude of the reflected power p now at the current time and the reflected power p last at the last time, comparison is made and the number of times is counted to determine whether the logic unit stops searching for the reflected power, specifically including:

[0015] When p now > p last , marking is performed, the number of times of marking is k p , k p must be continuously increased, otherwise k p is reset to zero, and the number of cycles m is increased by 1; and when k p is greater than the increment threshold n p times, the search is stopped and step A3 is re-executed; if the search is not stopped, step A5 is performed.

[0016] When p now < p last , marking is performed, the number of times of marking is k q , k q must be continuously increased, otherwise k q is reset to zero, and the number of cycles m is increased by 1; and when k q is greater than the decrement threshold n q times, the search is stopped and step A3 is re-executed; if the search is not stopped, step A5 is performed.

[0017] When the number of cycles m is greater than the cycle threshold c, the search is stopped and step A3 is re-executed; if the search is not stopped, step A5 is performed.

[0018] A5: The adjustable capacitance value c v and the gradient vector g v are updated by momentum, specifically as follows:

[0019] A51: v is the momentum at the current time, v last represents the momentum at the last time, l is the attenuation rate, and η is the learning rate, so the current momentum is:

[0020]

[0021] norm() represents the length of the vector.​ represents the value of g v is normalized;

[0022] A52: update the adjustable capacitance value c v , and then input it into the function get_g() to obtain the gradient

[0023] vector g v :

[0024] c v ← c v -v

[0025] g v = get_g(c v )

[0026] A53: repeat the steps of A4.

[0027] Further, S1: continuously receive the reflected power on the integrated circuit, and determine whether the reflected power is less than the set power threshold r;

[0028] S2: if the reflected power is less than the power threshold r, repeat S1, otherwise execute S3;

[0029] S3: execute the above-mentioned A1-A5 method steps.

[0030] Further, the adjustable capacitance value satisfies the following formula:

[0031] randomly select a set of capacitor initial values from the Y group of capacitor initial values Initialize c in , c len and c out , and let c v represent the set of c in , c len and c out , that is, c v = [c in , c len , c out ], then:

[0032]

[0033] wherein random() represents a function of random selection.

[0034] Further, the function get_g() includes the following formula:

[0035] g in = get_power([c in + L, clen , c out ])-get_power([c in -L, c len , c out ])

[0036] g len =get_power([c in , c len +L, c out ])-get_power([c in , c len -L, c out ])

[0037] g out =get_power([c in , c len , c out +L])-get_power([c in , c len , c out -L])

[0038] where get_power() means the measured control component reflected power according to the tunable capacitance value, L means the search step of the gradient investigation, used for the gradient calculation.

[0039] Further, the tunable capacitance value can also be greater than 3 or less than 3 generated by the variable capacitor. Depending on the device structure, but the method principle does not change.

[0040] Further, the calculation method of the gradient is not limited to the center gradient scheme of get_g(), which can be the existing forward gradient or backward gradient calculation method; without changing the existing principle.

[0041] The A5 step updates the tunable capacitance value c v and the gradient vector g v , which can be through the general momentum method v = l * v last + η * g v , or Newton method, without changing the existing principle.

[0042] The get_power() of the measured reflected power value depends on the device itself properties. Without changing the existing principle.

[0043] An RFID reading device, comprising: a monitoring unit, a control component and a logic unit; wherein,

[0044] The monitoring unit receives the reflected power on the integrated circuit and transmits it to the control component;

[0045] The control component includes a carrier cancellation circuit composed of three adjustable capacitors; controls the monitoring unit and the logic unit, receives the reflected power of the monitoring unit and sends it to the logic unit;

[0046] The logic unit receives the reflected power of the control component, performs carrier cancellation operation, controls the three variable capacitors through the interface, and causes the reflection of the radio frequency signal at the port.

[0047] The component performs the above search method. As follows:

[0048] The above components perform steps A1-A5:

[0049] A1: set power threshold r, increment threshold n p , decrement threshold n q , and cycle threshold c;

[0050] A2: adjust the adjustable capacitance value of the control component, which is used to adjust the reflected power, and the adjustable capacitance value is generated by three variable capacitors, and the three adjustable capacitance values are c in , c len and c out , set the initial value of Y group capacitor

[0051] A3: take the adjustable capacitance value as three-dimensional coordinates, c v as the coordinate point, and take the initial value of the capacitor as the coordinate origin, calculate the gradient by the function get_g(), and obtain the gradient vector g v =[g in , g len , g out ], when searching in a loop, the update of the gradient vector g v guides the update of the adjustable capacitance value;

[0052] A4: the reflected power measurement formula is p = get_power(c v ), and it is judged whether the reflected power is less than the power threshold r:

[0053] A41: if the current reflected power p now is less than the power threshold r, the logic unit stops searching for the reflected power, adjusts the adjustable capacitance value to the adjustable capacitance value c now corresponding to the reflected power p vnow , and ends the process; otherwise, step A42 is performed;

[0054] A42: according to the current reflected power p now and the last reflected power p lastThe size of the logical unit is compared and counted, and it is determined whether the logical unit stops searching for reflected power, specifically comprising:

[0055] When p now > p last , mark, the number of times of marking is k p , k p must be continuously increased, otherwise k p is reset to zero, and the number of cycles m is increased by 1; and when k p is greater than the increment threshold n p times, stop searching, and re-execute step A3; if the search is not stopped, execute step A5.

[0056] When p now < p last , mark, the number of times of marking is k q , k q must be continuously increased, otherwise k q is reset to zero, and the number of cycles m is increased by 1; and when k q is greater than the decrement threshold n q times, stop searching, and re-execute step A3; if the search is not stopped, execute step A5.

[0057] When the number of cycles m is greater than the cycle threshold c, stop searching, and re-execute step A3; if the search is not stopped, execute step A5.

[0058] A5: Update the adjustable capacitance value c v and the gradient vector g v , specifically as follows:

[0059] A51: v is the current momentum, v last represents the momentum at the last moment, l is the attenuation rate, and η is the learning rate, so the current momentum is:

[0060]

[0061] norm() represents the length of the vector, which means that g v is normalized;

[0062] A52: Update the adjustable capacitance value c v after the current momentum value, and then input the function get_g() to obtain the gradient vector g v :

[0063] c v ← c v -v

[0064] g v = get_g(cv )

[0065] A53: Repeat the step A4.

[0066] If the reflection phase of the carrier cancellation circuit is 180° and the amplitude is unchanged, the leaked interference signal will be cancelled.

[0067] Further, the carrier cancellation circuit of the control component is provided with a directional coupler, i.e. a directional coupler is arranged between the monitoring unit and the adjustable capacitor, the directional coupler receives the reflected power and transmits to the monitoring unit. The reflected power includes the reflected power generated by the carrier cancellation circuit, and can also include the reflected power of the interference signal received in other lines.

[0068] Further, the directional coupler is connected with the antenna to receive the transmitting power of the antenna.

[0069] Further, the logic unit is arranged on the mainboard assembly or the RFID module.

[0070] The present application has the following beneficial effects:

[0071] 1. The carrier cancellation search method based on RFID and the RFID reading device search the adjustable capacitance value in three dimensions at the same time, which is shorter than the traditional dichotomy or traversal method, and is more likely to hit the qualified reflected power compared with the method of adjusting the variable capacitance one by one. The current time reflected power p now and the comparison result of the reflected power p last of the last time are marked and reset to determine whether to continue searching, and the initial value of the capacitance that cannot converge can be exited in time, avoiding invalid search and reducing search time.

[0072] 2. The carrier cancellation search method based on RFID and the RFID reading device propose an improved momentum method, so that the adjustable capacitance value updated each time converges at the same speed in the direction that is most likely to converge, and the qualified adjustable capacitance value can be found more stably.

[0073] 3. The carrier cancellation search method based on RFID and the RFID reading device are optimized at the algorithm and process level, without the need to increase additional components, effectively reducing the hardware complexity and cost. BRIEF DESCRIPTION OF DRAWINGS

[0074] Figure 1 is a flowchart of the carrier cancellation search method based on RFID;

[0075] Figure 2A search flowchart of a carrier cancellation search method based on RFID of the present application is shown in the figure.

[0076] Figure 3 A schematic diagram of an RFID reading device of the present application is shown in the figure.

[0077] Figure 4 An embodiment effect diagram of the present application is shown in the figure.

[0078] Reference numerals: 10 - monitoring unit; 20 - logic unit; 30 - control component; 31 - directional coupler; 40 - antenna; 50 - reader IC ST25RU3993. DETAILED DESCRIPTION

[0079] The present application will be described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0080] A carrier cancellation search method based on RFID and an RFID reading device are shown in the figure, taking reader IC ST25RU3993 as the hardware platform, the carrier cancellation system includes: a monitoring unit 10, a control component 20 and a logic unit 30, wherein, Figures 1-3

[0081] The monitoring unit 10 receives the reflected power on the integrated circuit and transmits it to the control component 20.

[0082] The control component 20 includes a carrier cancellation circuit composed of three adjustable capacitors; controls and manages the monitoring unit 10 and the logic unit 30, receives the reflected power of the monitoring unit 10 and sends it to the logic unit 30.

[0083] The logic unit 30 receives the reflected power of the control component 20, performs carrier cancellation operation, controls the three variable capacitors through the interface, and causes the radio frequency signal to reflect at the port. If the reflection phase shift of the carrier cancellation circuit is 180° and the amplitude is unchanged, the leaked interference signal will be cancelled.

[0084] The control unit is provided with a directional coupler 30, which receives the reflected power and transmits it to the monitoring unit 10. The directional coupler 30 is connected to the antenna 40 and receives the transmitted power of the antenna 40. The logic unit 30 is arranged on the mainboard assembly or the RFID module. The following steps are performed:

[0085] S1: Continuously receive the reflected power on the integrated circuit and determine whether the reflected power is less than the set power threshold r.

[0086] ​S2: If the reflected power is less than the power threshold r, repeat S1, otherwise perform S3. When the reflected power is greater than the power threshold r, it is assumed that the reception has been affected, and the capacitors need to be adjusted to reduce the reflected power.

[0087] S3: Control the three variable capacitors to reduce the reflected power.

[0088] A1: Set the power threshold r, the increment threshold n p , the decrement threshold n q , and the cycle threshold c.

[0089] A2: Adjust the adjustable capacitance value of the control component 20, which is used to adjust the reflected power, and the adjustable capacitance value is generated by three variable capacitors, and the three adjustable capacitance values are c in , c len , and c out , set the Y group capacitor initial value The adjustable capacitance value satisfies the following formula:

[0090] Randomly select a group of capacitor initial values from the Y group capacitor initial value Initialize c in , c len , and c out , and let c v represent the set of c in , c len , and c out , that is, c v = [c in , c len , c out ], then:

[0091]

[0092] Where, random() represents a random selection function.

[0093] A3: For the problem of fast convergence, the fast descent method is generally used for mathematical solution. The three capacitors are regarded as three-dimensional coordinates, c v is a coordinate point, and the gradient is calculated with the capacitor initial value as the coordinate origin. The relationship between the capacitor value and the power is a hyper-surface graph with multiple extreme values, but since the carrier cancellation model cannot be accurately obtained, the derivative of the analytical expression cannot be calculated. In addition, due to the uncertainty of the model, there are many extreme points without rules, which will lead to the fact that it is extremely easy to fall into a local minimum value in the process of finding the minimum value. Therefore, a simplified way to solve the gradient is proposed, and the fast descent method is improved.

[0094] The search for the low-order value of the reflected power is performed to converge the function. The adjustable capacitor value is converted into the corresponding reflected power. The search direction and magnitude of the adjustable capacitor value are determined using a fast descent method: the adjustable capacitor value is used as a three-dimensional coordinate, c v Using the capacitor as the initial coordinate point and the capacitor as the origin, the gradient is calculated using the function get_g() to obtain the gradient vector g. v =[g in g len g out When the search loop continues, the temporal gradient vector g v The update guides the update of the adjustable capacitor value, and the update of the adjustable capacitor value also affects the gradient vector g. v Update; the function get_g() contains the following formula:

[0095] g in =get_power([c in +L, c len c out ])-get_power([c in -L,c len c out ])

[0096] g len =get_power([c in c len +L, c out ])-get_power([c in c len -L,c out ])

[0097] g out =get_power([c in c len c out +L])-get_power([c in c len c out -L])

[0098] Where get_power() means obtaining the reflected power of the measurement control unit based on the adjustable capacitor value, and L represents the search step size for gradient investigation, which is a constant value used to calculate the gradient.

[0099] A4: The formula for measuring reflected power is: p = get_power(c v ), determine whether the reflected power is less than the power threshold r:

[0100] A41: If the current reflection power pnow If the power threshold r is less than the reflected power p, the logic unit 30 stops searching for the reflected power, adjusts the adjustable capacitance value to the reflected power p now The corresponding adjustable capacitance value c vnow ; otherwise, step A42 is executed.

[0101] A42: To prevent over-searching, the threshold is set to a power threshold r, an incremental number threshold n p , a decremental number threshold n q , and a cycle number threshold c, k p represents the number of times p now > p last , k q represents the number of times p now < p last , according to the size of the reflected power p now at the current time and the reflected power p last at the previous time, comparison is made and the number is counted, to determine whether the logic unit 30 stops searching for the reflected power, which specifically includes:

[0102] When p now > p last , it indicates that the adjustable capacitance value does not search in the direction of decreasing reflected power at this time, and the adjustable capacitance value is searched incrementally, p now > p last is marked, and the number of times of marking is k p , k p must be continuously increased, otherwise k p is reset to zero, and the cycle number m is increased by 1; and when k p is greater than the incremental number threshold n p times, the search is stopped, and step A3 is re-executed; if the search is not stopped, step A5 is executed.

[0103] When p now < p last , it indicates that the adjustable capacitance value searches in the direction of decreasing reflected power at this time, and the adjustable capacitance value is searched decrementally, p now < p last is marked, and the number of times of marking is k q , k q must be continuously increased, otherwise k q is reset to zero, and the cycle number m is increased by 1; and when k q is greater than the decremental number threshold n q times, the search is stopped, and step A3 is re-executed; if the search is not stopped, step A5 is executed.

[0104] When the cycle number m is greater than the cycle number threshold c, the search is stopped, and step A3 is re-executed; if the search is not stopped, step A5 is executed.

[0105] A5: The fast descent method in the prior art is prone to fall into a local minimum value, resulting in an increased probability of search failure. The momentum method can improve the problem, but both methods are prone to search for a minimum value that meets the condition due to the too large modulus value of the gradient g v . To this end, an improved momentum method optimization algorithm is proposed, which makes it easier to jump out of a local optimal solution that does not meet the condition and hit a minimum value that meets the condition during the descent process. Through momentum, the adjustable capacitance value c v and the gradient vector g v are updated, reducing the search loop times, as follows:

[0106] A51: v is the current momentum, v last represents the momentum of the last moment, l is the attenuation rate, and η is the learning rate, so the current momentum is:

[0107]

[0108] norm() represents the modulus of the vector, and g v is normalized.

[0109] A52: The adjustable capacitance value c v is updated according to the current momentum value, and then the gradient vector g v is obtained by bringing the function get_g() into the function.

[0110] c v ← c v -v

[0111] g v = get_g(c v )

[0112] The value of c v -v is re-assigned to c v ; the update of g v guides the update of the adjustable capacitance value. After the gradient g v is normalized, the update of c v is limited in the same step, i.e., the convergence is equal. When searching in the direction of the fastest descent, it is easier to hit the minimum value that meets the condition.

[0113] A53: Repeat step A4.

[0114] The embodiment is the hit rate and time performance of the present application under different capacitance initial values (64 groups and 27 groups). The power threshold r for jumping out is -20dbm, the incremental number threshold n p is 3, and the decremental number threshold n qc is 3, and the cycle number threshold value c is 16.

[0115] Embodiments Illustration Number of sets of initial values of capacitances Embodiment one Figure 4 (top left) 64 Embodiment two Figure 4 (Top right) 64 Embodiment three Figure 4 (bottom left) 27 Embodiment four Figure 4 (Bottom right) 27

[0116] Figure two The test results comparison chart in Table 1 shows that the average fold line on the upper side is the test result of the present application, and the lower side is the conventional test method. As shown in Examples 1 and 3, the actual hit rate of the present application is higher than that of the improved momentum method and gradient descent method; as shown in Examples 2 and 4, the average cycle time is less than that of the improved momentum method and gradient descent method.

[0117] The skilled person in the art can adjust the following values or similar replacements according to the present application, which all belong to the concept of the present application, such as adjusting the adjustable capacitance value to be greater than 3 or less than 3 to generate the variable capacitor. The calculation method of the gradient can be the forward gradient or backward gradient method in addition to the central gradient of the present application. The A5 step updates the adjustable capacitance value c v and the gradient vector g v , which can be obtained by the general momentum method v = l * v last + η * g v , or the existing technology method such as Newton method. None of them changes the principle of the present application.

[0118] Forward gradient and backward gradient are commonly used in the fields of mathematics and computer science. They are used to determine the local gradient or slope of a function to help optimization algorithms find the minimum or maximum value of the function. These two methods are also known as forward difference and backward difference.

[0119] Forward gradient (Forward Gradient) or forward difference (Forward Difference):

[0120] Forward gradient is a method of calculating gradient, which uses a point and a slightly forward point near the point to estimate the local gradient of the function.

[0121] Specifically, for a one-dimensional function f(x), the forward gradient can be expressed as:scssCopy code

[0122] f'(x)≈(f(x+h)-f(x)) / h

[0123] Where h is a small positive number representing the distance between points. By calculating the difference between the function at the current point and the slightly forward point, and dividing by h, we can estimate the gradient of the function f(x) at the point x.

[0124] Backward gradient (Backward Gradient) or backward difference (Backward Difference):

[0125] Backward gradient is another method of calculating the gradient, which uses a point and a slightly backward point near the point to estimate the local gradient of the function.

[0126] For a one-dimensional function f(x), the backward gradient can be represented as:scssCopy code

[0127] f'(x)≈(f(x)-f(x-h)) / h

[0128] Similarly, h is a small positive number representing the distance between points. By calculating the difference of the function at the current point and the slightly backward point, and dividing by h, we can estimate the gradient of the function f(x) at the point x.

[0129] The choice of which method to use usually depends on the nature of the specific problem and the requirements of numerical stability.

[0130] The above is only an embodiment of the present application, and the selection of the embodiment scheme is only to better understand the content of the application, and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation using the content of the present application specification and drawings, or direct or indirect application in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. An RFID-based carrier-canceling search method, characterized by, Comprising the following steps: A1: setting a power threshold r, an increment number threshold n p , a decrement number threshold n q and a cycle number threshold c; A2: adjusting the adjustable capacitance value of the control component (20) for adjusting the reflected power, the adjustable capacitance value being generated by three variable capacitors, three adjustable capacitance values being c in , c len and c out , a set of c v , c in , c len and c out being represented by c v =[c in , c len , c out ], setting the initial value of the Y group of capacitors randomly selecting a group of capacitor initial values from among them in , c len and c out are initialized; A3: taking the adjustable capacitance value as a three-dimensional coordinate, c v is a coordinate point, and taking the capacitance initial value as a coordinate origin, a gradient vector g is obtained by a function get_g() to calculate a gradient v = [g in , g len , g out ], when a search loop is searched, the update of the gradient vector g v guides the update of the adjustable capacitance value; the function get_g() includes the following formula: g in = get_power([c in -L, c len , c out ]) - get_power([c in + L, c len , c out ]) g len = get_power([c in , c len + L, c out ]) - get_power([c in , c len - L, c out ]) g out = get_power([c in ,c len ,c out + L]) - get_power([c in ,c len ,c out - L]) Wherein get_power() represents the measured reflection power of the control component (20) according to the adjustable capacitor value, L represents the search step of gradient examination, and is a constant value for calculating the gradient; A4: The reflection power measurement formula is: p = get_power(c v ) and it is determined whether the reflection power is less than the power threshold r: A41: If the current moment reflection power p now is less than the power threshold r, the logic unit (30) stops searching for the reflection power, adjusts the adjustable capacitance value to the reflection power p now corresponding to the adjustable capacitance value c vnow , and ends the process; otherwise, the step A42 is executed. A42: Based on the current reflection power p now and the reflected power p at the previous moment last The size is compared and counted to determine whether the logic unit (30) stops searching for reflection power. Specifically, this includes: When p now > p last , the number of markings is k p , k p must be continuously increased, otherwise k p is reset to zero, the number of loops m is incremented by 1; and when k p is greater than the threshold number of increments n p , the search is stopped and the A3 step is re-executed; if the search is not stopped, the A5 step is executed; When p now < p last , a mark is made, the number of marks being k q , k q must be increased consecutively, otherwise k q is reset to zero and the number of cycles m is incremented by one; and when k q is greater than a threshold number of decrements n q , the search is stopped and the A3 step is re-executed; if the search is not stopped, the A5 step is executed; When the number of cycles m is greater than the cycle threshold c, stop searching and re-execute the A3 step; if the search is not stopped, execute the A5 step; A5: updating the adjustable capacitance value c by momentum v and the gradient vector g v , in particular as follows: A51: v is the current momentum, v last denotes the momentum of the previous time, l is the decay rate, and η is the learning rate. The current momentum is then: norm() denotes taking the norm of a vector, denotes normalizing g v is normalized; A52: Update the adjustable capacitance value c according to the current momentum value v After, the function get_g() is brought in to obtain the gradient vector g v : c v ←c v -v g v = get_g(c v ) A53: repeatedly execute the A4 step.

2. The RFID-based carrier-canceling search method of claim 1, wherein, Comprising the following steps: S1: continuously receive the reflection power on the integrated circuit and determine whether the reflection power is less than the set power threshold r; S2: if the reflection power is less than the power threshold r, repeat the S1 step, otherwise execute the S3 step; S3: execute the steps A1-A5 of claim 1.

3. The RFID-based carrier-cancelled search method of claim 1 or 2, wherein, The adjustable capacitor value satisfies the following formula: Wherein random() represents a randomly selected function.

4. The RFID-based carrier sense search method of claim 1, wherein, The adjustable capacitor value is greater than 3 or less than 3 variable capacitors.

5. The RFID-based carrier-canceling search method of claim 1, wherein, The gradient calculation method is a forward gradient or backward gradient calculation method; The A5 step updates the tunable capacitance value c v and the gradient vector g v is by a general momentum method v = l * v last + η * g v or Newton's method.

6. An RFID reading device for performing the method of claim 1, characterized by Comprising: A monitoring unit (10), a control component (20) and a logic unit (30); wherein, The monitoring unit (10) receives the reflection power on the integrated circuit and transmits it to the control component (20); The control component (20) comprises a carrier cancellation circuit composed of three adjustable capacitors; controls and manages the monitoring unit (10) and the logic unit (30), receives the reflection power of the monitoring unit (10) and transmits it to the logic unit (30); The logic unit (30) receives the reflection power of the control component (20), performs carrier cancellation operation, controls the three variable capacitors through the interface, and causes the reflection of the radio frequency signal at the port; The above monitoring unit (10), control component (20) and logic unit (30) execute the steps of claim 1: A1 : setting a power threshold r, an increment number threshold n p , a decrement number threshold n q and a cycle number threshold c; A2: adjusting the adjustable capacitance value of the control component (20) for adjusting the reflected power, the adjustable capacitance value being generated by three variable capacitors, the three adjustable capacitance values being c in , c len , and c out , the set of c v , c in , c len , and c out being denoted as c v = [c in , c len , c out ], setting the initial value of the Y group of capacitors randomly selecting a group of capacitor initial values from among them; initializing c in , c len , and c out ; A3: taking the adjustable capacitance value as a three-dimensional coordinate, c v is a coordinate point, and taking the capacitance initial value as a coordinate origin, a gradient vector g is obtained by a function get_g() v = in , len , out ], when searching in a loop, the update of the gradient vector g v guides the update of the adjustable capacitance value; the function get_g() contains the following formula: g in = get_power([c in + L, c len , c out ]) - get_power([c in - L, c len , c out ]) g len = get_power([c in , c len + L, c out ]) - get_power([c in , c len - L, c out ]) g out = get_power([c in ,c len ,c out + L]) - get_power([c in ,c len ,c out - L]) Wherein get_power() represents the measured reflection power of the control component (20) according to the adjustable capacitor value, L represents the search step of gradient examination, and is a constant value for calculating the gradient; A4: The reflection power measurement formula is: p = get_power(c v ) and it is determined whether the reflection power is less than the power threshold r: A41 : If the current instant reflected power p now is less than the power threshold r, the logic unit (30) stops searching for a reflected power, adjusts the adjustable capacitance value to the reflected power p now corresponding to the adjustable capacitance value c vnow , ends the procedure; otherwise, it performs the step A42. A42: Based on the current reflection power p now and the reflected power p at the previous moment last The size is compared and counted to determine whether the logic unit (30) stops searching for reflection power. Specifically, this includes: When p now > p last , the number of markings is k p , k p must be increased continuously, otherwise k p is reset to zero, and the number of loops m is incremented by 1; and when k p is greater than the threshold number of increments n p times, the search is stopped and the A3 step is re-executed; if the search is not stopped, the A5 step is executed; When p now < p last , a mark is made, the number of marks being k q , k q must be increased consecutively, otherwise k q is reset to zero and the number of cycles m is incremented by one; and when k q is greater than the threshold number of decrements n q , the search is stopped and the A3 step is re-executed; if the search is not stopped, the A5 step is executed; When the number of cycles m is greater than the cycle threshold c, stop searching and re-execute the A3 step; if the search is not stopped, execute the A5 step; A5: updating the adjustable capacitance value c by momentum v and the gradient vector g v , in particular as follows: A51: v is the current momentum, v last denotes the momentum of the previous time, l is the decay rate, and η is the learning rate. The current momentum is then: norm() denotes taking the norm of a vector, denotes normalizing g v is normalized; A52: Update the adjustable capacitance value c according to the current momentum value v After, the function get_g() is brought in to obtain the gradient vector g v : c v ←c v -v g v = get_g(c v ) A53: repeatedly execute the A4 step.

7. An RFID reading device according to claim 6, wherein, The control component (20) is provided with a directional coupler (31), which receives the reflection power and transmits it to the monitoring unit (10).

8. An RFID reading device according to claim 7, wherein, The directional coupler (31) is connected with the antenna (40) and receives the transmission power of the antenna (40).

9. The RFID reading device of claim 6, wherein, The logic unit (30) is arranged on the mainboard assembly or the RFID module.

Images