Multi-stage detection method and multi-stage detector for bipolar pulse amplitude modulation signals

Abstract

The invention discloses a multi-stage detection method and a multi-stage detector for a bipolar pulse amplitude modulation (BPAM) signal, which are suitable for a BPAM modulation-direct detection system, especially for a short-distance high-speed data transmission scene, and aim to solve the problem of high direct detection complexity caused by distortion of various signals in the prior art. By decoupling three kinds of damage of transmitter response, diode square law detection and receiver response, step-by-step compensation is carried out in stages, the method comprises the steps of receiver response compensation, polarity recovery and transmitter-optical fiber response compensation, and the method is a low-complexity detection scheme capable of compensating multi-class distortion coupling. Decoupling and staged step-by-step compensation of complex damage are achieved, actual implementation is easy, robustness is high, and the calculation complexity of a detection module is remarkably reduced.

Description

Technical Field

[0001] The present invention relates to optical communication technologies, and more particularly to a multi-level detection technology for bipolar pulse amplitude modulation (BPAM) signals. Background Art

[0002] With the continuous development of artificial intelligence technologies, the data traffic in short-distance applications has grown explosively. To meet this demand, the transmission rate of intra-data center interconnects (intra-DCIs) is moving towards 1.6 Tb / s or even 3.2 Tb / s. Due to low cost and simple architecture, intensity modulation direct detection (IMDD) transmission schemes are favored in intra-DCIs. To support terabit-level transmission rates, wavelength division multiplexing (WDM) with 8 or 16 channels has become the mainstream solution. However, dense channel multiplexing brings integration challenges, and the increase in the number of channels leads to a proportional increase in the number of components, affecting economic efficiency. Therefore, increasing the single-channel data rate is a more promising solution.

[0003] To increase the single-channel rate, high-order complex modulation formats can be used to improve spectral efficiency. However, in IMDD systems, traditional methods only detect the signal intensity and lose phase information, resulting in low spectral efficiency. To improve spectral efficiency, bipolar pulse amplitude modulation (BPAM) and oversampling reception schemes have been proposed.

[0004] BPAM signals retain the positive and negative polarities of the signals during modulation, so that more information can be carried per symbol. Direct detection means that the receiver only detects the intensity of the optical signal, that is, the square of the amplitude of the optical electric field. To recover the signal phase information lost in direct detection, generally, the signal after direct detection needs to be oversampled by two times and processed using the maximum likelihood detection (MLD) algorithm. The complexity of MLD increases exponentially with the increase in the modulation order and the link memory length. For example, when the channel memory length of a BPAM-16 signal exceeds 8, the complexity of MLD is as high as 16 8 , which is not feasible in practical applications. To reduce the complexity, a detection scheme combining serial interference cancellation and shortened channel memory length has been proposed, which can partially alleviate the problem of excessive complexity. However, in practice, the non-ideal characteristics of transceivers and distortion effects such as the square-law detection of photodiodes are coupled together, resulting in problems such as difficult overall channel modeling, complex experimental implementation, and limited performance. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide an easy-to-implement low-complexity multi-level detection scheme for the problem that in an IMDD system for intra-data center interconnects, due to the loss of polarity of BPAM signals after passing through diodes and the influence of various distortion effects in the system, the complexity of traditional MLD on the receiver side is too high.

[0006] The technical solution adopted by the present invention to solve the above technical problems is a multi-level detection method for bipolar pulse amplitude modulation signals, which is characterized by including the following steps:

[0007] Receiver response compensation step: Input the signal sampled at twice the symbol rate from the receiver response output into the feed-forward equalizer T / 2-FFE with a twice-symbol sampling rate. After the feed-forward equalizer T / 2-FFE compensates for the receiver response, it outputs the power estimation value of the odd and even sampled symbols;

[0008] Polarity recovery step: Utilize the interpolation relationship between the power estimation values of the odd and even sampled symbols, and recover the polarity of the sampled symbols at the single-symbol sampling rate through maximum likelihood estimation to obtain the samples sampled at the single-symbol sampling rate;

[0009] Transmitter-fiber response compensation step: Input the samples sampled at the single-symbol sampling rate into the feed-forward equalizer T-FFE with a single-symbol sampling rate. The feed-forward equalizer T-FFE compensates for the transmitter-fiber response and outputs the bipolar pulse amplitude modulation electrical signal with distortion coupling compensation completed.

[0010] Specifically, before the receiver response compensation step, the measurement of the receiver frequency response and the estimation of the frequency responses of the transmitter and the fiber channel are first carried out.

[0011] The tap values of the feed-forward equalizer T / 2-FFE are obtained according to the T / 2-FFE response; the T / 2-FFE response is the inverse function of the receiver frequency response;

[0012] The tap values of the feed-forward equalizer T-FFE are obtained according to the T-FFE response; the T-FFE response is the inverse function of the frequency responses of the transmitter and the fiber channel.

[0013] At the same time, a multi-level detector for implementing the above method is provided, including a receiver response compensation unit, a polarity recovery unit, and a transmitter-fiber response compensation unit;

[0014] The receiver response compensation unit is used to input the signal sampled at twice the symbol rate from the receiver response output into the feed-forward equalizer T / 2-FFE with a twice-symbol sampling rate. After the feed-forward equalizer T / 2-FFE compensates for the receiver response, it outputs the power estimation value of the odd and even sampled symbols;

[0015] The polarity recovery unit is used to utilize the interpolation relationship between the power estimation values of the odd and even sampled symbols, and recover the polarity of the sampled symbols at the single-symbol sampling rate through maximum likelihood estimation to obtain the samples sampled at the single-symbol sampling rate;

[0016] The transmitter-fiber response compensation unit is used to input the samples sampled at the single-sampling rate into the feed-forward equalizer T-FFE with the single-symbol sampling rate. The feed-forward equalizer T-FFE compensates the transmitter-fiber response output to complete the distortion-coupling compensation of the bipolar pulse amplitude modulation electrical signal.

[0017] Specifically, the receiver response compensation unit receives the known receiver frequency response, which is obtained by measurement; the transmitter-fiber response compensation unit receives the known frequency response of the transmitter and the fiber channel, which is obtained by estimation.

[0018] The receiver response compensation unit sets the tap values of the feed-forward equalizer T / 2-FFE according to the T / 2-FFE response; the T / 2-FFE response is the inverse function of the receiver frequency response;

[0019] The transmitter-fiber response compensation unit sets the tap values of the feed-forward equalizer T-FFE according to the T-FFE response; the T-FFE response is the inverse function of the frequency response of the transmitter and the fiber channel.

[0020] The multi-stage detection scheme of the present invention decouples three main impairments, namely the transmitter-fiber response, the nonlinear effect caused by the square-law of the photodiode, and the receiver response, and compensates them stage by stage to replace the complex impairment compensation.

[0021] The beneficial effects of the present invention are as follows: realizing the decoupling and stage-by-stage compensation of complex impairments, being easy to implement in practice and having high robustness, and significantly reducing the computational complexity of the detection module. Description of the Drawings

[0022] Figure 1 It is a block diagram of the BPAM transmission system and the multi-stage detection module of the present invention.

[0023] Figure 2 It is an implementation block diagram of the BPAM transmission system of the present invention. Detailed Embodiments

[0024] The multi-stage detection scheme of the embodiment decouples three main impairments, namely the transmitter-fiber response, the nonlinear effect caused by the square-law of the photodiode, and the receiver response, and performs receiver response compensation, polarity recovery, and transmitter-fiber response compensation stage by stage.

[0025] As Figure 1 shown, at the transmitting end, the bipolar pulse amplitude modulation BPAM-16 symbol passes through the transmitter response h tx (t) and then passes through the electro-optic conversion to the double-polar optical signal to the optical fiber; at the receiving end, the double-polar optical signal {s e ,s o}The detected electrical signal {|s is obtained through direct detection of photoelectric conversion e | 2 ,|s o | 2}, {|s e | 2 ,|s o | 2} is convolved with the receiver response h rx (t) to obtain the signal {r e ,r o} sampled at twice the symbol rate. Among them, e represents the even position in the bipolar, and o represents the odd position in the bipolar. {r e ,r o} is input to the multi-stage detection module for hierarchical compensation of the multi-stage damage coupling effect of the system link.

[0026] The multi-stage detection module includes a receiver response compensation unit, a polarity recovery unit, and a transmitter-fiber response compensation unit. Before compensation, the multi-stage detection module first measures the receiver frequency response using a spectrum analyzer or a vector analyzer for receiver response compensation. The frequency responses of the transmitter and the fiber channel are estimated by the recursive least squares (RLS) method.

[0027] The multi-stage detection module specifically performs the following steps to compensate for the multi-stage damage coupling effect of the system link:

[0028] Receiver response compensation step: The receiver response compensation unit compensates the receiver response for the signal {r e ,r o} sampled at twice the symbol rate using a feed-forward equalizer T / 2-FFE with a sampling rate twice the symbol rate. Among them, r e represents the even-position sampling points of the signal sampled at twice the symbol rate, and r o represents the odd-position sampling points of the signal sampled at twice the symbol rate. T / 2-FFE means that the tap interval of the equalizer is 1 / 2 of the symbol period T, that is, a feed-forward equalizer with a sampling rate twice the symbol rate. The samples at twice the symbol rate output by T / 2-FFE are alternately divided into odd and even parts, that is, the power estimation values of the even and odd sampling symbols The response of the feed-forward equalizer T / 2-FFE is the inverse function of the measured receiver frequency response. When the response of T / 2-FFE is known, the tap values of T / 2-FFE can be calculated by performing an inverse discrete Fourier transform on the frequency response of T / 2-FFE.

[0029] The power estimation values of the odd and even sampling symbols output after equalization are corresponding to the optical signal {s e ,s oThe power estimation value of} has lost the corresponding polarity information. Therefore, polarity recovery is required.

[0030] Polarity recovery steps: The polarity recovery unit uses the interpolation relationship between the even and odd sampled symbols to estimate the polarity of the sampled symbols through the maximum likelihood algorithm. When the optical signal is sampled at twice the symbol rate, there is an interpolation relationship The interpolation parameter h p can be estimated by transmitting a known training sequence. The known training sequence transmitted is a BPAM-16 symbol x of a certain length transmitted by the transmitter. Further, it can be deduced that the interpolation relationship between the power estimation values of the even and odd sampled symbols is:

[0031]

[0032] where n is random noise, approximately following a Gaussian distribution. Using the constraint relationship satisfied between the even and odd sampled symbols in Equation (1), the polarity of either the odd or even symbol is estimated through the maximum likelihood algorithm. In the embodiment, the estimated even sampled symbol is taken as an example. The polarity of is the sequence of polarity variables c' when the objective function in Equation (2) reaches the maximum value:

[0033]

[0034] where the constant σ 2 represents the variance of the noise n, which can be set to 1 here without affecting the polarity estimation result. The error metric c' m ∈{-1,1} represents the polarity variable. k and m are sequence number variables in the symbol sequence. Equation (2) takes the sequence of polarity variables c' = [c1',…,c' m ,…] that minimizes the overall error metric (maximizes the objective function) as the polarity estimation value Using the polarity of the symbol can be recovered:

[0035]

[0036] In the embodiment, only the polarity of the even sampled symbols is recovered, and is used as the sample value for single symbol rate sampling. Similarly, only the polarity of the odd sampled symbols can be recovered, and is used as the sample value for single symbol rate sampling.

[0037] Transmitter-fiber response compensation steps: The transmitter-fiber response compensation unit compensates the samples with recovered polarity The transmitter-fiber response is compensated by a feed-forward equalizer T-FFE with a single symbol sampling rate. With the aid of the training sequence x, the frequency responses of the transmitter and the fiber channel can be estimated. The feed-forward equalizer T-FFE response is the inverse function of the estimated frequency responses of the transmitter and the fiber channel. When the T-FFE response is known, the tap values of the T-FFE can be calculated by performing an inverse discrete Fourier transform on the T / 2-FFE frequency response.

[0038] Since the BPAM signal is subject to the diode square-law detection effect during reception and its spectrum is broadened to twice the original, the receiver needs to sample at twice the symbol rate. Therefore, the receiver response compensation should also use an equalizer T / 2-FFE with a twice symbol sampling rate. For the response compensation of the transmitter-fiber, only an equalizer T-FFE with a single symbol sampling rate is required.

[0039] Example: Transmission of a 28Gbaud BPAM-16 signal over 1 km.

[0040] The implementation platform is as Figure 2 shown. The implementation platform consists of two parts: offline digital signal processing and an intensity modulation direct detection optical transmission system. Among them, the offline digital signal processing is completed on the computer side. The digital signal generated by the computer is transmitted to the digital-to-analog converter DAC, or the digital acquisition signal is obtained from the analog-to-digital converter ADC. The transmitting end of the optical transmission system includes a DAC, a radio frequency amplifier, a Mach-Zehnder modulator MZM, and a laser. The receiving end includes an optical attenuator, a photodiode, and an ADC.

[0041] The specific implementation steps are as follows:

[0042] 1. Receiver parameter measurement:

[0043] Use a spectrum analyzer or a vector analyzer to separately measure the ADC frequency response and find its inverse function for receiver response compensation in a multi-level detector.

[0044] 2. Signal generation

[0045] At the transmitting end, a random bit sequence is mapped to 28Gbaud BPAM-16 symbols. Among them, the first 3000 symbols can be used as the training sequence x. Subsequently, pre-emphasis, raised cosine pulse shaping with a roll-off factor of 0.3, and resampling are performed. Then the signal is loaded into the DAC to generate an analog waveform. The output waveform is amplified by a radio frequency amplifier and drives the MZM. At the same time, a continuous laser signal with a wavelength of 1550 nm and an optical power of 14.5 dBm is input into the MZM, and its maximum output optical power is about 7 dBm. The bias voltage of this MZM modulator is set at the optical power zero point to generate a bipolar optical signal (BPAM signal).

[0046] 3. Transmission and Reception:

[0047] After the modulated optical signal is transmitted through 1 km of single-mode optical fiber, the received optical power is adjusted by a variable optical attenuator. Subsequently, the optical signal is subjected to square-law detection by a photodiode with a bandwidth of 50 GHz and finally sampled by an ADC at a rate of ≥2 times the symbol rate and processed offline.

[0048] 4. Offline Signal Processing:

[0049] The sampled signal passes through a Bessel filter and undergoes clock recovery and downsampling to 2 samples per symbol. The downsampled samples are fed into a multi-stage detector for processing. The specific process and parameter settings of the multi-stage detector are as follows:

[0050] a) Receiver Response Compensation: For samples sampled at twice the symbol rate (representing sampling points at even and odd positions respectively), a feed-forward equalizer T / 2-FFE with a twice-symbol sampling rate is used to compensate for the receiver response.

[0051] b) Polarity Recovery: Using the interpolation relationship between even and odd samples, the polarity of the sample symbols is estimated by the maximum likelihood estimation algorithm. Here, only the polarity of even (or odd) sample points needs to be recovered to obtain signal samples at the single sampling rate.

[0052] c) Transmitter-Fiber Response Compensation: For the single-rate samples with recovered polarity, a feed-forward equalizer T-FFE with a single-symbol sampling rate is used to compensate for the transmitter-fiber response.

[0053] The multi-stage detector outputs the estimated values of BPAM-16 symbols, which are then demapped and the bit error rate is statistically calculated.

[0054] The embodiment only takes a 1 km transmission system of 28 Gbaud BPAM-16 signal as an example to illustrate the main technical concept and characteristics of the present invention, that is, how the multi-stage detection scheme and module are applied in practice. However, it is not a limitation on the implementation manner of the present invention. Any modification, equivalent transformation, and improvement made on this application shall be included within the protection scope of the claims of the present invention.

Claims

1. A multi-level detection method for a bipolar pulse amplitude modulated signal, characterized in that: The following steps are involved: Receiver response compensation step: inputting a signal sampled at twice the symbol rate from the receiver response output into a feedforward equalizer T / 2-FFE using twice the symbol sampling rate, and the feedforward equalizer T / 2-FFE compensates the receiver response and outputs power estimation values ​​of odd and even sampling symbols; Polarity recovery step: using the interpolation relationship between the power estimation values ​​of the odd and even sampling symbols, the polarity of the sampling symbols of the single sampling rate is recovered through maximum likelihood estimation to obtain samples sampled at the single sampling rate; Transmitter-fiber response compensation step: input the samples sampled at the single sampling rate into the feedforward equalizer T-FFE at the single symbol sampling rate, and the feedforward equalizer T-FFE compensates the transmitter-fiber response to output a bipolar pulse amplitude modulated electrical signal that completes the distortion coupling compensation.

2. The method according to claim 1, characterized in that The receiver response compensation step is preceded by a measurement of the receiver frequency response and an estimation of the frequency response of the transmitter and the optical fiber channel.

3. The method according to claim 2, characterized in that The tap value of the feedforward equalizer T / 2-FFE is obtained according to the T / 2-FFE response; the T / 2-FFE response is an inverse function of the receiver frequency response; The tap values ​​of the feed-forward equalizer T-FFE are obtained according to the T-FFE response; the T-FFE response is an inverse function of the frequency response of the transmitter and the optical fiber channel.

4. The method according to claim 1, characterized in that The specific method of restoring the polarity of the sampling symbol of the single sampling rate by maximum likelihood estimation using the interpolation relationship between the power estimation values ​​of the odd and even sampling symbols is as follows: Even-numbered sampling symbols as sampling symbols of single sampling rate Polarity The polarity variable sequence c′ is used to maximize the objective function in the following formula: Among them, the constant σ 2 represents the variance of noise n, the kth error metric c′ m ∈{-1,1} represents a polar variable; is the power estimate of the even sampling symbol, is the power estimate of the odd sampling symbol; k and m are the sequence variables in the symbol sequence; the polarity variable sequence c′=[c1′,…,c′ m ,…].

5. A multi-level detector for bipolar pulse amplitude modulated signals, characterized in that It includes a receiver response compensation unit, a polarity recovery unit and a transmitter-fiber response compensation unit; The receiver response compensation unit is used to input the signal sampled at twice the symbol rate from the receiver response output into the feedforward equalizer T / 2-FFE using twice the symbol sampling rate, and the feedforward equalizer T / 2-FFE compensates the receiver response and outputs the power estimation value of the odd and even sampling symbols; The polarity recovery unit is used to recover the polarity of the sampling symbols of the single sampling rate by using the interpolation relationship between the power estimation values ​​of the odd and even sampling symbols through maximum likelihood estimation, so as to obtain samples sampled at the single sampling rate; The transmitter-fiber response compensation unit is used to input samples sampled at a single sampling rate into a feedforward equalizer T-FFE at a single symbol sampling rate. The feedforward equalizer T-FFE compensates the transmitter-fiber response to output a bipolar pulse amplitude modulated electrical signal that completes distortion coupling compensation.

6. The multi-stage detector according to claim 5, characterized in that: The receiver response compensation unit receives a known receiver frequency response obtained by measurement; the transmitter-fiber response compensation unit receives a known transmitter and fiber channel frequency response obtained by estimation.

7. The multi-stage detector according to claim 6, characterized in that: The receiver response compensation unit sets the tap value of the feedforward equalizer T / 2-FFE according to the T / 2-FFE response; the T / 2-FFE response is an inverse function of the receiver frequency response; The transmitter-fiber response compensation unit sets the tap value of the feedforward equalizer T-FFE according to the T-FFE response; the T-FFE response is an inverse function of the frequency response of the transmitter and the fiber channel.

Images