A method for separating gigabit ethernet mixed signals

Abstract

The application discloses a method for separating mixed signals of a gigabit Ethernet, which comprises the following steps: selecting a first signal collection position for collecting signals at a master device end and a second signal collection position for collecting signals at a slave device end between the master device and the slave device; the master device separately sends signals to obtain a channel transfer function at the 0th signal collection; the master device and the slave device simultaneously send signals to obtain a channel transfer function at the 1st signal collection; judging whether the channel transfer functions at the 2nd signal collection are consistent; if yes, separating the sending signals of the master device and the sending signals of the slave device according to the signals collected at the two signal collection positions and the channel transfer function at the 1st signal collection; if not, re-executing until the channel transfer functions are consistent. The application utilizes the irrelevance of the signals sent by the devices at two sides of a wired network to provide a stable and effective mixed signal separation method for gigabit Ethernet devices, and further extracts the device fingerprints of the gigabit Ethernet.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of information security and computer network technology, and particularly relates to a method for separating mixed signals of a gigabit Ethernet. BACKGROUND

[0002] In the communication process of a gigabit Ethernet, communication devices are physically connected and communicated through fixed physical cables, and the signals of communication can only be sent by the sender and reach the receiver through fixed lines, so the security performance of the wired network is relatively good. However, wired networks still have some common security vulnerabilities, such as address spoofing of media access control (MAC) and Internet Protocol (IP), which can make unauthorized network devices access the network and thus pose a threat to network security. In order to make up for the security vulnerabilities of the data link layer and the network layer, a physical layer security method is proposed, which can identify the data of the attacker at the physical layer and distinguish it from normal devices, so as to prevent illegal devices from accessing the network.

[0003] As a classic and most widely used wired network, researchers have begun to use device fingerprints for identity recognition in the research on identity authentication of Ethernet devices at home and abroad. However, since the gigabit Ethernet generally operates in full-duplex communication mode, the data collected on the line is usually a mixed signal of both ends of the communication, which brings great challenges to the detection of the intrusion of illegal devices. Based on this, a method for separating the mixed signals of the gigabit Ethernet is proposed, which separates the signals sent by the master device and the access device through the separation of the mixed signals, and then obtains the device fingerprint of the access device. By extracting the device fingerprint in the signal sent by the access device and comparing it with the existing legal device fingerprint in the local area network, it is determined whether the access device is legal to decide whether to allow the access device to access the network. By separating the mixed signals, the device fingerprint of the access device is obtained, so as to determine whether the access device is allowed to access the network, and thus effectively solve the security problem of address spoofing of the network layer and the data link layer.

[0004] The document "Gerdes, R. M, Mina, et al. Physical-Layer Identification of Wired Ethernet Devices [J]" first extracts the device fingerprint of the 10M standard Ethernet standard based on the signal preamble, and uses the method of matching filter and adaptive threshold to classify different devices. However, in gigabit Ethernet, the random scrambling sequence reaches 33 bits, the determined preamble sequence and the random scrambling sequence are exclusive or, and then the signal coding is performed, so that the transmission signal in the actual communication is random code for a long time, which makes the method of extracting device fingerprint based on preamble introduce data interference, increase the difficulty of extracting device fingerprint, and the accuracy is not enough. The patent "Device fingerprint information extraction method in wired network card signal" discloses a method of obtaining adaptive filter parameters as wired network card device fingerprint by restoring ideal signal. For half-duplex 100M wired network card, the method is also applicable, but for full-duplex devices such as gigabit Ethernet and vehicle-mounted gigabit / gigabit Ethernet, the full-duplex signals sent by the devices at both ends of the communication line are mixed together, it is difficult to restore the correct ideal signal, thereby causing the extracted device fingerprint to be unstable. Therefore, in order to solve the device fingerprint extraction and access authentication problem of gigabit Ethernet full-duplex device, an effective and data-independent technical means is needed to separate the signals sent by the master device side and the access device side in the mixed signal. SUMMARY

[0005] The purpose of the application is to solve the problems in the prior art. The application discloses a method for separating gigabit Ethernet mixed signals.

[0006] The technical scheme is as follows:

[0007] A method for separating gigabit Ethernet mixed signals, comprising the following steps:

[0008] S1, connecting two communication devices by wire, and selecting two signal collection positions on the connection line between the two communication devices, the two communication devices being a master device and a slave device, and the two signal collection positions being a first signal collection position for collecting master device end signal and a second signal collection position for collecting slave device end signal;

[0009] S2, the master device sends signals to the slave device alone, and the 0th signal collection is performed at the first signal collection position and the second signal collection position, and the channel transfer function at the 0th signal collection is calculated after the signals collected at the two signal collection positions are subjected to autocorrelation operation and frequency domain transformation respectively;

[0010] S3, the master device and the slave device simultaneously send signals to each other, the first signal collection is performed at the first signal collection position and the second signal collection position simultaneously, the channel transfer function at the first signal collection is calculated after the cross-correlation operation and the frequency domain transformation are performed on the signals collected at the two signal collection positions and the self-correlation operation and the frequency domain transformation are respectively performed on the signals collected at the two signal collection positions;

[0011] S4, it is judged whether the channel transfer function at the 0th signal collection is consistent with the channel transfer function at the 1st signal collection, if consistent, the channel transfer function at the 0th signal collection is taken as the final channel transfer function, and the sending signal of the master device and the sending signal of the slave device at the 1st signal collection are separated according to the signals collected at the two signal collection positions at the 1st signal collection and the channel transfer function; if not consistent, the step S2 and the step S3 are re-executed until the channel transfer function at the 0th signal collection is consistent with the channel transfer function at the 1st signal collection.

[0012] Preferably, in the step S2, the frequency domain H0(k) of the channel transfer function at the 0th signal collection is expressed as:

[0013]

[0014] wherein, represents the value after the self-correlation operation and the frequency domain transformation are performed on the signal v 10 (n) collected at the first signal collection position at the 0th signal collection, represents the value after the self-correlation operation and the frequency domain transformation are performed on the signal v 20 (n) collected at the second signal collection position at the 0th signal collection.

[0015] Preferably, in the step S3, the frequency domain H1(k) of the channel transfer function at the 1st signal collection is expressed as:

[0016]

[0017] wherein, represents the value after the self-correlation operation and the frequency domain transformation are performed on the signal v 11 (n) collected at the first signal collection position at the 1st signal collection, represents the value after the self-correlation operation and the frequency domain transformation are performed on the signal v 21 (n) collected at the second signal collection position at the 1st signal collection, 11 represents the value after the self-correlation operation and the frequency domain transformation are performed on the signal v 21 (n) and v 11 (n) at the 1st signal collection. 21(n) is the value after cross-correlation and frequency domain transformation.

[0018] Preferably, in step S4, the separated master device transmission signal and slave device transmission signal are respectively:

[0019]

[0020]

[0021] Among them, X 11 (k) represents the frequency domain signal of the master device's transmitted signal, X 21 (k) represents the frequency domain signal of the signal transmitted from the device, V 11 (k) represents the signal v acquired at the first signal acquisition position during the first signal acquisition. 11 The frequency domain signal of (n), V 21 (k) represents the signal v acquired at the second signal acquisition position during the first signal acquisition. 21 H(k) represents the frequency domain signal of the final channel transfer function.

[0022] Preferably, the following steps are also included:

[0023] S5. Replace the network card of different slave devices. The master device and slave device send signals to each other at the same first signal acquisition position and the same second signal acquisition position. Simultaneously perform the i-th signal acquisition, i = 2, 3, ... Repeat the calculation process in steps S3 and S4 for the signals acquired at the two signal acquisition positions. Calculate the frequency domain of the channel transfer function, the transmitted signal of the master device and the transmitted signal of the slave device at the time of the i-th signal acquisition.

[0024] Compare the channel transfer function during each signal acquisition to see if they are consistent. If they are consistent, it means that the modeling is successful and the separation of the mixed signals is successful and effective.

[0025] Compare whether the transmitted signals of the master device separated in each signal acquisition are consistent, and compare whether there are differences in the transmitted signals of different slave devices separated in each signal acquisition. If the above conditions are met, it means that the modeling is successful and the separation of the mixed signals is successful and effective.

[0026] Beneficial effects: Compared with the prior art, the present invention has the following significant beneficial effects:

[0027] The method described in this invention utilizes the lack of correlation between the signals transmitted by full-duplex devices on both sides of a wired network to provide a stable and effective method for separating mixed signals for Gigabit Ethernet devices. This method can then be used to extract device fingerprints for Gigabit Ethernet, thereby achieving security for Gigabit Ethernet network access. Attached Figure Description

[0028] Figure 1 The flow chart of the method for separating the gigabit Ethernet hybrid signal according to the present application;

[0029] Figure 2 The schematic diagram of the connection mode of the master device and the slave device;

[0030] Figure 3 The waveform diagram and the amplitude spectrum diagram of the signal collected at the first signal collection position in the 0th signal collection;

[0031] Figure 4 The waveform diagram and the amplitude spectrum diagram of the signal collected at the first signal collection position in the 1st signal collection;

[0032] Figure 5 The time domain sending signal of the separated slave device and the amplitude spectrum diagram thereof. DETAILED DESCRIPTION

[0033] The present application will be further described below in combination with the drawings.

[0034] The present application discloses a method for separating a gigabit Ethernet hybrid signal, as shown in the following steps: Figure 1

[0035] Step S1, two full duplex communication devices equipped with gigabit Ethernet cards are connected by wire, one of which is a master device and the other is a slave device;

[0036] Two signal collection positions are selected on the connection line between the master device and the slave device, which are the first signal collection position and the second signal collection position, respectively, wherein the first signal collection position collects the signal at the master device end of the connection line, and the second signal collection position collects the signal at the slave device end of the connection line.

[0037] In this embodiment, the master device M and the slave device S are both 1000base-T gigabit Ethernet devices, and the master device M and the slave device S are connected by wire such as twisted pair, and the first signal collection position A and the second signal collection position B are provided on the twisted pair, and the signals are collected at the two signal collection positions by a signal collection device such as an oscilloscope, and the connection mode is as shown in the following figure: Figure 2

[0038] Step S2, the master device and the slave device are powered on respectively, and the master device sends signals to the slave device alone, and the signals sent by the master device are collected at the first signal collection position and the second signal collection position at the same time, that is, the signals in one direction, as the 0th signal collection, wherein the signal collected at the first signal collection position is denoted as v 10 (n), and the signal collected at the second signal collection position is denoted as v 20 (n). ​​

[0039] In this embodiment, the signal acquisition device acquires the differential signals at the first signal acquisition location A and the second signal acquisition location B at a sampling rate of 500Msps. This is considered the 0th signal acquisition, and the acquisition is performed only once during the entire separation process of the mixed signal. The number of sampling points is N = 7,000,000. The symbol rate on a pair of twisted pairs in a Gigabit Ethernet network is 125MHz, approximately four times the sampling rate.

[0040] The signal v acquired at the first signal acquisition location 10 Waveform and amplitude spectrum V of (n) 10 (k) such as Figure 3 As shown.

[0041] During the 0th signal acquisition, the signals v acquired at the first and second signal acquisition positions are... 10 (n) and v 20 (n) are respectively:

[0042] v 10 (n)=x 10 (n) (1)

[0043] v 20 (n)=x 10 (n)*h(n) (2)

[0044] Where n = 0, 1, ..., N-1, x 10 h(n) represents the time-domain transmitted signal of the master device corresponding to the 0th signal acquisition (including the device fingerprint signal of the master device), h(n) represents the channel transfer function containing channel characteristics and noise, and "*" represents the linear convolution operation.

[0045] For the signal v during the 0th signal acquisition 10 (n) and v 20 (n), perform autocorrelation operation and frequency domain transformation respectively to obtain the corresponding frequency domain signal. and The frequency domain H0(k) of the channel transfer function at this time is calculated.

[0046] For v 10 (n) and v 20 (n) Perform autocorrelation operations on each, as follows:

[0047]

[0048]

[0049] For v 10 (n), v 20 Autocorrelation function of (n) Transform them to the frequency domain, and they are expressed as follows:

[0050]

[0051]

[0052] where k = 0, 1,..., N-1, represents the value after the self-correlation operation and the frequency domain transformation of the signal v 10 (n) is sequentially performed, which can be used to represent the power spectrum of v 10 (n), represents the value after the self-correlation operation and the frequency domain transformation of the signal v 20 (n) is sequentially performed, which can be used to represent the power spectrum of v 20 (n),FT(·) represents an N-point frequency domain transformation, including but not limited to a discrete Fourier transform, a discrete cosine transform, etc.;

[0053] At the same time, the self-correlation operation is performed on x 10 (n), which is represented as:

[0054]

[0055] The self-correlation function of x 10 (n) is transformed into the frequency domain, which is represented as:

[0056]

[0057] where, represents the value after the self-correlation operation and the frequency domain transformation of the time domain transmission signal x 10 (n) of the master device at the 0th signal acquisition, according to the Wiener-Sin theorem, which can be used to represent the power spectrum of x 10 (n) ;

[0058] The self-correlation operation is performed on the channel transfer function h(n), which is represented as:

[0059]

[0060] The time domain convolution is converted into the frequency domain and multiplied. Since the spectrum is a symmetric even function, the conversion into the frequency domain is calculated as follows:

[0061] R h (k) = H(k)H(-k) = H 2 (k)

[0062] Since the characteristics of the channel and the noise remain basically constant, the value Rh (k) is equal to H 2 (k).

[0063] Combining equations (1) and (2), it is thus satisfied that:

[0064]

[0065]

[0066] The frequency domain of the channel transfer function at the 0th signal acquisition is solved as:

[0067]

[0068] Step S3, the master device and the slave device simultaneously send signals to each other, and simultaneously collect the mixed signals of the gigabit Ethernet at the first signal acquisition position and the second signal acquisition position as the 1st signal acquisition. The mixed signals collected are superimposed by the signals sent by the master device and the slave device. At this time, the signal collected at the first signal acquisition position is denoted as v 11 (n), and the signal collected at the second signal acquisition position is denoted as v 21 (n).

[0069] In this embodiment, the signal acquisition device collects the differential signals at the first signal acquisition position A and the second signal acquisition position B at a sampling rate of 500 Msps as the 1st signal acquisition, and the number of sampling points N = 7000000. The symbol rate on a pair of twisted pairs of the gigabit Ethernet is 125 MHz.

[0070] The waveform diagram and the amplitude spectrum V 11 (k) of the signal v 11 (n) collected at the first signal acquisition position are as shown in Figure 4 .

[0071] Unlike the signal v 10 (n) containing only the single-ended signal of the master device and the signal v 20 (n) of the 0th signal acquisition in step S2, the signals v 11 (n) and v 21 (n) are full-duplex mixed signals superimposed by the signals of the master device and the slave device. At the 1st signal acquisition, the signals v 11 (n) and v 21 (n) collected at the first signal acquisition position and the second signal acquisition position are respectively:

[0072] v 11 (n) = x 11 (n) + x 21 (n) * h(n) (6)

[0073] v21 (n) = x 11 (n) * h(n) + x 21 (n) (7)

[0074] where n = 0, 1,..., N - 1, x 11 (n) and x 21 (n) represent the time-domain transmitting signals of the master device and the slave device, respectively, in the first signal collection.

[0075] where the signal x 11 (n) contains the device fingerprint signal of the master device, and the signal x 21 (n) contains the device fingerprint signal of the slave device, that is:

[0076] x 11 (n) = s 11 (n) * p 11 (n) (8)

[0077] x 21 (n) = s 21 (n) * p 21 (n) (9)

[0078] where s 11 (n) and s 21 (n) represent the time-domain ideal transmitting signals of the master device and the slave device, respectively, in the first signal collection, p 11 (n) and p 21 (n) represent the time-domain expressions of the device fingerprints of the master device and the slave device.

[0079] The cross-correlation operation and the frequency-domain transformation are performed on the signals v 11 (n) and v 21 (n) in the first signal collection, and the autocorrelation operation and the frequency-domain transformation are performed on the signals v 11 (n) and v 21 (n), respectively, to calculate the frequency-domain H1(k) of the channel transfer function at this time.

[0080] The cross-correlation operation is performed on v 11 (n) and v 21 (n), and is represented as:

[0081]

[0082] where n = 0, 1,..., N - 1, N represents the number of sampling points, and represent the autocorrelation functions of x 11 (n) and x 21 (n), respectively, and x 11(n) and x 21 (n) represent the signals sent by the two devices, which have almost no correlation, therefore x 11 (n)Corr.x 21 (n) can be approximated as 0 and ignored. Approximation operations can be performed during the calculation process.

[0083] v 11 (n) and v 21 The cross-correlation function γ(v) of (n) 11 ,v 21 Transform it to the frequency domain, and it will be represented as:

[0084]

[0085] Where k = 0, 1, ..., N-1, This represents the time-domain transmission signal x of the master device during the first signal acquisition. 11 (n) The value after sequential autocorrelation and frequency domain transformation can be used to represent x. 11 The power spectrum of (n), This indicates that the signal x was sent from the device in the time domain during the first signal acquisition. 21 (n) The value after sequential autocorrelation and frequency domain transformation can be used to represent x. 21 The power spectrum of (n).

[0086] For v 11 (n) is subjected to autocorrelation, and is represented as:

[0087]

[0088] Among them, due to x 11 (n) and x 21 (n) represent the signals sent by the two devices, which have almost no correlation, therefore the cross-correlation term r(x) is... 11 (n),x 21 (n) can be approximated as 0 and ignored. Approximation operations can be performed during the calculation process.

[0089] Similarly, for v 21 (n) is subjected to autocorrelation, and is represented as:

[0090]

[0091] For v 11 Autocorrelation function of (n) Transformed to the frequency domain, it can be represented as:

[0092]

[0093] and:

[0094]

[0095]

[0096] H(k) = FT(h(n))

[0097] Thus, we have:

[0098]

[0099] Similarly, the autocorrelation function of v 21 (n) is transformed to the frequency domain and is denoted as:

[0100]

[0101] where n = 0, 1,..., N - 1, k = 0, 1,..., N - 1, and N represents the number of sampling points, denotes the value obtained by sequentially performing autocorrelation operation and frequency domain transformation on the signal v 11 (n) collected at the first signal collection position during the first signal collection, and can be used to represent the power spectrum of v 11 (n). denotes the value obtained by sequentially performing autocorrelation operation and frequency domain transformation on the signal v 21 (n) collected at the second signal collection position during the first signal collection, and can be used to represent the power spectrum of v 21 (n).

[0102] Through formulas (14) and (15), we can obtain:

[0103]

[0104] Through formulas (11) and (16), we can obtain:

[0105]

[0106] Let then the frequency domain of the channel transfer function during the first signal collection can be solved as:

[0107]

[0108] i.e.:

[0109]

[0110] ​Step S4, comparing the frequency domain H0(k) of the channel transfer function at the 0th signal acquisition obtained in step S2 and the frequency domain H1(k) of the channel transfer function at the 1st signal acquisition obtained in step S3, judging whether the data modeling under the two signal acquisitions has a problem.

[0111] In this embodiment, the frequency domain H0(k) of the channel transfer function at the 0th signal acquisition calculated in the foregoing formula (5) and the frequency domain H1(k) of the channel transfer function at the 1st signal acquisition calculated in formula (18) are compared, respectively, to judge whether the modeling results of the single-ended signal at the 0th signal acquisition and the mixed signal at the 1st signal acquisition are close or consistent. If formula (18) still holds after H0(k) calculated in formula (5) is brought into H1(k) in formula (18), it indicates that H0(k) and H1(k) are consistent, which means that the modeling is successful, proving that H0(k) is correct and effective, and the frequency domain H(k) of the channel transfer function is H0(k). If formula (18) does not hold after H0(k) calculated in formula (5) is brought into H1(k) in formula (18), it indicates that the modeling is not successful, and steps S2 and S3 are re-executed until the modeling is successful.

[0112] According to the signal v 11 (n) collected at the first signal acquisition position at the 1st signal acquisition and the signal v 21 (n) collected at the second signal acquisition position, combined with the channel transfer function H(k), the transmission signal of the master device and the transmission signal of the slave device at the 1st signal acquisition are separated out.

[0113] In this embodiment, for the signals v 11 (n) and v 21 (n) collected in step S4, according to the modeling formulas (6) and (7), the time domain signals are transformed into the frequency domain, which are represented as:

[0114] V 11 (k)=X 11 (k)+X 21 (k)H(k) (19)

[0115] V 21 (k)=X 11 (k)H(k)+X 21 (k) (20)

[0116] Wherein, V 11 (k) represents the frequency domain signal of the signal v 11 (n), V 21 (k) represents the frequency domain signal of the signal v 21 (n), X 11 (k) represents the frequency domain signal of the signal x 11The frequency domain signal of (n), X 21 (k) represents the signal x 21 The frequency domain signal of (n):

[0117] V 11 (k) = FT(v 11 (n))

[0118] V 21 (k) = FT(v 21 (n))

[0119] X 11 (k) = FT(x 11 (n))

[0120] X 21 (k) = FT(x 21 (n))

[0121] Solving the binary linear equations composed of formulas (19) and (20) obtains:

[0122]

[0123]

[0124] The inverse transformation is performed on X 11 (k) and X 21 (k) calculated in formulas (21) and (22) respectively, so that the time domain sending signal x 11 (n) of the master device and the time domain sending signal x 21 (n) of the slave device are obtained.

[0125] The waveform diagram and the amplitude spectrum V 21 (k) of the separated time domain sending signal x 21 (n) of the slave device are shown as follows. Figure 5

[0126] The method for separating the gigabit Ethernet hybrid signal can further include the following steps for verifying the effectiveness of the separated sending signal of the master device and the sending signal of the slave device:

[0127] In step S5, the network card of the slave device is replaced, and the full duplex hybrid signal is collected simultaneously at the same first signal collection position and second signal collection position as the i-th (i=2, 3,...) signal collection, and the calculation process in step S3 is repeated.

[0128] ​In this embodiment, by testing different gigabit network cards of the slave device multiple times, multiple signal separations are performed to further test the modeling process. The network card of the slave device is replaced, and the ith signal collection is performed at the first signal collection position and the second signal collection position. At this time, the signal collected at the first signal collection position is denoted as v 1i (n), and the signal collected at the second signal collection position is denoted as v 2i (n).

[0129] At the ith signal collection, the signals collected at the first signal collection position and the second signal collection position are v 1i (n) and v 2i (n) respectively.

[0130] v 1i (n) = x 1i (n) + x 2i (n) * h(n) (23)

[0131] v 2i (n) = x 1i (n) * h(n) + x 2i (n) (24)

[0132] wherein n = 0, 1,..., N-1, x 1i (n) and x 2i (n) respectively represent the time-domain transmission signals of the master device and the slave device corresponding to the ith signal collection.

[0133] The cross-correlation operation is performed on v 1i (n) and v 2i (n), and is transformed into the frequency domain, and is denoted as:

[0134]

[0135] wherein k = 0, 1,..., N-1, denotes the value after the autocorrelation operation and the frequency domain transformation of x 1i (n) of the master device at the ith signal collection, which can be used to represent the power spectrum of x 1i (n). denotes the value after the autocorrelation operation and the frequency domain transformation of x 2i (n) of the slave device at the ith signal collection, which can be used to represent the power spectrum of x 2i (n).

[0136] The autocorrelation operation is performed on v 1i (n) and v 2i (n) respectively, and after being transformed into the frequency domain, the following can be obtained:

[0137]

[0138] By formula (25) and (26), can be obtained

[0139]

[0140] Let Then the frequency domain of the channel transfer function of the i-th signal acquisition can be solved:

[0141]

[0142] Solve:

[0143]

[0144] Directly determine whether a i is equal to a1, that is, whether H(k) is consistent with H i (k), and test the correctness of the model multiple times.

[0145] For the signal v 1i (n) and v 2i (n) of the i-th signal acquisition, according to the modeling formula (23) and (24), the time domain signal is transformed into the frequency domain, and is expressed as:

[0146] V 1i (k) = X 1i (k) + X 2i (k)H(k) (29)

[0147] V 2i (k) = X 1i (k)H(k) + X 2i (k) (30)

[0148] Wherein, V 1i (k) represents the frequency domain signal of signal v 1i (n), V 2i (k) represents the frequency domain signal of signal v 2i (n), X 1i (k) represents the frequency domain signal of signal x 1i (n), X 2i (k) represents the frequency domain signal of signal x 2i (n).

[0149] Solve the binary linear equation composed of formula (29) and (30), and obtain:

[0150]

[0151]

[0152] Multiple tests were conducted on different slave devices to separate the transmission signals of the master device and different slave devices, verifying the correctness of the model and simultaneously separating the transmission signals of different devices.

[0153] Compare the values ​​obtained by cross-correlation and frequency domain transformation of the signals acquired at the two acquisition locations at each signal acquisition point with the sum of the values ​​obtained by autocorrelation and frequency domain transformation of the signals acquired at the two acquisition locations at each acquisition point (i.e., the ratio of 'a' calculated at each signal acquisition point). i Whether (i=1,2,...)) are consistent, that is, to determine whether H(k) and H i (k) Whether they are consistent. If they are consistent, it means that the modeling is successful, and the method described in this invention is successful and effective in separating mixed signals in Gigabit Ethernet. Furthermore, this value a can be used to further demonstrate the modeling process. i (i=1,2,...) yields a relatively accurate channel transfer function h(n), which also demonstrates the stability of the physical channel of Gigabit Ethernet.

[0154] When different slave devices communicate with the same master device, the master device's transmitted signal should remain unchanged, while the transmitted signals of the different slave devices should have certain differences.

[0155] Based on this, we determine whether the main device signal separated from multiple tests is stable.

[0156] Compare the frequency domain signal of the master device's transmitted signal separated during each signal acquisition (i.e., the X calculated during each signal acquisition). 1i Are (k)(i=1,2,...)) consistent? (If there are occasional deviations in the transmitted signals of a few separated master devices, these can be considered as negligible error data.) Compare the frequency domain signals of the transmitted signals of different slave devices separated during each signal acquisition (i.e., the X calculated during each signal acquisition). 2i If there is a certain difference between (k)(i=1,2,...)), and the above conditions are met, it indicates that the modeling is successful and the method described in this invention is successful and effective in separating mixed signals in Gigabit Ethernet.

[0157] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for separating mixed signals in Gigabit Ethernet, characterized in that, Includes the following steps: S1. Two communication devices are connected by a wire. Two signal acquisition positions are selected on the connection line between the two communication devices. The two communication devices are a master device and a slave device, and the two signal acquisition positions are a first signal acquisition position for acquiring signals from the master device and a second signal acquisition position for acquiring signals from the slave device. S2. The master device sends a signal to the slave device separately, and performs the 0th signal acquisition at the first signal acquisition position and the second signal acquisition position. After performing autocorrelation operation and frequency domain transformation on the signals acquired at the two signal acquisition positions, the channel transfer function at the time of the 0th signal acquisition is calculated. S3. The master and slave devices simultaneously send signals to each other, performing the first signal acquisition simultaneously at the first and second signal acquisition positions. Cross-correlation and frequency domain transformation are performed on the signals acquired at the two positions, and autocorrelation and frequency domain transformation are performed on the signals acquired at each position respectively. The channel transfer function at the time of the first signal acquisition is then calculated. The frequency domain expression H1(k) of the channel transfer function at the time of the first signal acquisition is as follows: in, This represents the signal v acquired at the first signal acquisition position during the first signal acquisition. 11 (n) The values ​​after performing autocorrelation and frequency domain transformation sequentially. This represents the signal v acquired at the second signal acquisition position during the first signal acquisition. 21 (n) is the value after sequential autocorrelation and frequency domain transformation, Γ(v) 11 (n),v 21 (n) represents the signal v during the first signal acquisition. 11 (n) and v 21 (n) is the value after cross-correlation and frequency domain transformation; S4. Determine whether the channel transfer function during the 0th signal acquisition is consistent with the channel transfer function during the 1st signal acquisition. If they are consistent, use the channel transfer function during the 0th signal acquisition as the final channel transfer function. Based on the signals and channel transfer functions acquired at the two signal acquisition locations during the 1st signal acquisition, separate the master device's transmit signal and the slave device's transmit signal during the 1st signal acquisition. If they are inconsistent, repeat steps S2 and S3 until the channel transfer function during the 0th signal acquisition is consistent with the channel transfer function during the 1st signal acquisition.

2. The method for separating mixed signals in Gigabit Ethernet according to claim 1, characterized in that, In step S2, the frequency domain H0(k) of the channel transfer function during the 0th signal acquisition is expressed as: in, This represents the signal v acquired at the first signal acquisition position during the 0th signal acquisition. 10 (n) The values ​​after performing autocorrelation and frequency domain transformation sequentially. This represents the signal v acquired at the second signal acquisition position during the 0th signal acquisition. 20 (n) The value after performing autocorrelation and frequency domain transformation in sequence.

3. The method for separating mixed signals in Gigabit Ethernet according to claim 1, characterized in that, In step S4, the separated master device transmission signal and slave device transmission signal are respectively: Among them, X 11 (k) represents the frequency domain signal of the master device's transmitted signal, X 21 (k) represents the frequency domain signal of the signal transmitted from the device. V 11 (k) represents the signal v acquired at the first signal acquisition position during the first signal acquisition. 11 The frequency domain signal of (n), V 21 (k) represents the signal v acquired at the second signal acquisition position during the first signal acquisition. 21 H(k) represents the frequency domain signal of the final channel transfer function.

4. The method for separating mixed signals in Gigabit Ethernet according to claim 1, characterized in that, It also includes the following steps: S5. Replace the network card of different slave devices. The master device and slave device send signals to each other at the same first signal acquisition position and the same second signal acquisition position. Simultaneously perform the i-th signal acquisition, i = 2, 3, ... Repeat the calculation process in steps S3 and S4 for the signals acquired at the two signal acquisition positions. Calculate the frequency domain of the channel transfer function, the transmitted signal of the master device and the transmitted signal of the slave device at the time of the i-th signal acquisition. Compare the channel transfer function during each signal acquisition to see if they are consistent. If they are consistent, it means that the modeling is successful and the separation of the mixed signals is successful and effective. Compare whether the transmitted signals of the master device separated in each signal acquisition are consistent, and compare whether there are differences in the transmitted signals of different slave devices separated in each signal acquisition. If the transmitted signals of the master device separated in each signal acquisition are consistent, and the transmitted signals of different slave devices separated in each signal acquisition are consistent, then the modeling is successful, and the separation of the mixed signals is successful and effective.

Images