A low resource overhead ultra-high speed modulation and demodulation method

Abstract

The application discloses a low-resource-overhead super-high-speed modulation and demodulation method. In view of the signal processing demand faced by high-frequency bands and large bandwidths in millimeter wave and terahertz communication, the application adopts a parallel architecture to meet the high-speed signal processing demand and correct non-ideal characteristics of a channel; and adopts a fusion processing method to fuse interpolation, anti-aliasing, timing interpolation, matched filtering, time domain equalization and decimation into a multi-phase filter, so that the signal overhead caused by high-speed demodulation is greatly reduced. The application can be applied to a millimeter wave / terahertz super-high-speed transmission system, and provides low-complexity super-high-speed modulation and demodulation to meet the performance and resource demand of super-high-speed transmission.

Description

Technical Field

[0001] This invention relates to the fields of millimeter wave, terahertz high-frequency band, and large bandwidth point-to-point transmission, and in particular to an ultra-high-speed, low-complexity modulation and demodulation method that can significantly improve the wireless communication transmission rate. Background Technology

[0002] With the rapid development of the communications field, the low end of the radio spectrum is nearing saturation. To meet the demands of emerging information technologies such as big data and cloud computing for higher-speed, wider-bandwidth wireless communication capabilities, developing and utilizing higher-frequency spectrum resources is an inevitable trend. The millimeter-wave band offers abundant available frequency resources, providing several GHz of available spectrum, enabling transmission rates of tens to hundreds of Gbps, comparable to terrestrial fiber optic transmission rates. By mounting millimeter-wave communication equipment on near-space vehicles, aircraft, and other airborne platforms, and constructing air-to-air backbone links and air-to-air / air-to-ground access links, an air-to-ground information system can be built, achieving seamless information network coverage and high-speed access.

[0003] The main problems faced in millimeter-wave wireless communication include two aspects:

[0004] 1) The ultra-large bandwidth of the millimeter wave high-frequency band brings about channel non-ideal problems, thus requiring equalization compensation for the non-ideal channel characteristics;

[0005] 2) For ultra-high-speed signal processing, a parallel algorithm architecture is required, which leads to a significant increase in the demand for hardware signal processing resources and logic resources. Summary of the Invention

[0006] The technical problem this invention aims to solve is to propose a low-resource-overhead ultra-high-speed modulation and demodulation method for millimeter-wave ultra-high-speed signal processing. This invention can adapt to the requirements of ultra-high-speed signal processing, perform equalization correction on channel characteristics, and significantly reduce the overhead on signal processing and logic resources.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0008] A low-resource-overhead ultra-high-speed modulation method includes the following steps:

[0009] Step 1: Convert high-speed information into multi-path parallel data, perform constellation mapping, and map it into parallel multi-path constellation points. Each constellation point consists of I and Q path data.

[0010] Step 2: Use a multiphase filter to perform parallel shaping filtering on the I and Q data respectively to obtain multiple parallel modulation signals, and then convert them into intermediate frequency signals through high-speed digital-to-analog conversion to complete ultra-high-speed modulation.

[0011] Furthermore, the number of parallel paths for constellation mapping is equal to the symbol rate divided by the processing clock value, and the number of parallel paths for high-speed digital-to-analog conversion is equal to the sampling rate divided by the processing clock value.

[0012] Furthermore, the multiphase filter in step (2) completes four operations—interpolation, anti-aliasing filtering, shaping filtering, and decimation—in a single process. The design method of the multiphase filter is as follows:

[0013] (201) Design interpolation, anti-aliasing filtering, shaping filtering and decimation operations. The anti-aliasing filtering and shaping filtering operations are in the form of transfer functions. The interpolation and decimation operations are zero-padding and decimation of time-domain data.

[0014] (202) The transfer functions of anti-aliasing filtering and shaping filtering are transformed into frequency domain transfer functions by FFT transformation with the same number of points, and then the final frequency domain transfer function is obtained by multiplying them in the frequency domain.

[0015] (203) After the final frequency domain transfer function is transformed into a time domain filter function, it is further transformed into a polyphase filter structure. Interpolation and decimation are applied to the input and output of the polyphase filter structure to obtain a polyphase shaping filter.

[0016] A low-resource-consumption ultra-high-speed demodulation method includes the following steps:

[0017] (1) Parallel multi-channel signals are obtained through high-speed analog-to-digital sampling, and then a receiving multiphase filter is constructed based on timing error information and channel equalization coefficient;

[0018] (2) The data is reduced from the sampling rate to the symbol rate by a multiphase filter to obtain parallel multi-channel data;

[0019] (3) Extract timing error information from the parallel multi-channel data output by the multiphase filter;

[0020] (4) Perform parallel frequency offset estimation and correction on the parallel multi-channel data output from the multiphase filter to achieve carrier synchronization;

[0021] (5) After carrier synchronization, the preamble information is extracted, the channel is estimated, and the channel state information is obtained;

[0022] (6) Extract UW information from the carrier-synchronized data, use UW to achieve phase tracking, and correct phase noise and residual frequency offset;

[0023] (7) After phase tracking, the signal demodulation is achieved by using constellation diagram mapping.

[0024] Furthermore, the number of parallel sampling paths for high-speed analog-to-digital sampling is equal to the sampling rate divided by the processing clock value, and the number of parallel paths for carrier synchronization, phase tracking, and constellation mapping is equal to the symbol rate divided by the processing clock value.

[0025] Furthermore, the specific method for constructing the multiphase filter is as follows:

[0026] (101) Design interpolation, anti-aliasing filtering, matched filtering, channel equalization, and timing decimation operations. Among them, the anti-aliasing filtering, matched filtering, and channel equalization operations are time-domain filtering functions, and the interpolation and timing decimation operations are zero-padding and decimation of time-domain data.

[0027] (102) By performing FFT transformation with the same number of points, the transfer functions of anti-aliasing filtering, matched filtering, and channel equalization are transformed into frequency domain transfer functions, and then the final frequency domain transfer function is obtained by multiplying them in the frequency domain.

[0028] (103) The final frequency domain transfer function is transformed into a time domain filter function, and further transformed into a polyphase filter structure. Interpolation and timing decimation are applied to the input and output of the polyphase filter structure to obtain a polyphase shaping filter.

[0029] Furthermore, the specific method of step (3) is as follows:

[0030] (301) For the output of the receiving multiphase filter, the timing error of each data frame is estimated by square law detection to obtain timing error information in the form of angle phase.

[0031] (302) The timing error information is fed back to the receiving multiphase filter. The receiving multiphase filter adjusts the filter coefficient according to the phase angle information, that is, the bit synchronization is achieved by adjusting the decimation position in the receiving multiphase filter.

[0032] Furthermore, the specific method of step (5) is as follows:

[0033] (501) After carrier synchronization, extract the preamble sequence of each frame for channel state estimation;

[0034] (502) Obtain the frequency domain channel equalization coefficients based on the channel state information;

[0035] (503) Feed the frequency domain channel equalization coefficient back to the receiving polyphase filter. The receiving polyphase filter adjusts the filtering coefficient according to the frequency domain channel equalization coefficient to correct the flatness of the radio frequency channel and the frequency selective fading of the propagation channel.

[0036] Compared with the prior art, the present invention has the following advantages:

[0037] 1. This invention adopts a parallel architecture, which has a large bandwidth parallel processing capability and achieves ultra-high-speed data transmission at low oversampling rates.

[0038] 2. The modulation and demodulation process of this invention can adapt to the high-speed signal processing requirements in millimeter-wave / terahertz communication and correct for undesirable characteristics such as channel radio frequency channel unevenness and phase noise difference.

[0039] 3. This invention simplifies the signal demodulation process, reducing the original process that required multiple filters to process the signal into a single polyphase filter, thus significantly reducing the signal processing resource requirements.

[0040] In summary, this invention can correct and compensate for non-ideal channel characteristics while significantly reducing computational complexity, system resource overhead, and the demand for hardware signal processing and logic resources. Attached Figure Description

[0041] Figure 1 This is a flowchart of the signal modulation process of the present invention.

[0042] Figure 2 This is a block diagram of the modulation multi-function filtering function of the present invention.

[0043] Figure 3 This is a flowchart of the normal signal demodulation process of the present invention.

[0044] Figure 4 This is a flowchart of the resource-optimized signal demodulation process of the present invention.

[0045] Figure 5 This is a block diagram of the demodulation multi-filter function of the present invention. Detailed Implementation

[0046] The present invention will now be described in detail with reference to the accompanying drawings.

[0047] A low-resource-overhead ultra-high-speed modulation method includes the following steps:

[0048] (1) High-speed information is converted into multiple parallel data and enters the constellation mapping, which maps it into a constellation point with multiple parallel paths. Each constellation point consists of two paths, I and Q. The number of parallel paths in the constellation mapping is equal to the symbol rate / processing clock.

[0049] (2) The IQ data are subjected to parallel shaping filtering using a multiphase filter to obtain multiple parallel modulated signals, which are then converted into intermediate frequency signals by a high-speed DAC. The number of parallel sampling channels of the DAC is equal to the sampling rate / processing clock.

[0050] This method reduces the processing clock speed by using a parallel approach, enabling the modulation signal processing to meet the hardware processing speed requirements.

[0051] In step (2), the design method of the multiphase filter is as follows:

[0052] (201) Construct four operations: interpolation, anti-aliasing filtering, shaping filtering, and extraction.

[0053] (202) The anti-aliasing filter and the shaping filter are converted into frequency domain transfer functions with the same number of points, and then the final frequency domain transfer function is obtained by multiplying them in the frequency domain.

[0054] (203) After the final frequency domain transfer function is transformed into a time domain filter function, it is further transformed into a polyphase filter structure to obtain a polyphase shaping filter.

[0055] A low-resource-consumption ultra-high-speed demodulation method includes the following steps:

[0056] (1) After obtaining parallel multi-channel signals by high-speed AD sampling, a multiphase filter is constructed based on timing error information and channel equalization coefficient.

[0057] (2) By using multiphase filtering, the data is reduced from the sampling rate to the symbol rate to obtain parallel multi-channel data.

[0058] (3) Extract timing error information from the received multiphase filtered data.

[0059] (4) The filtered data is subjected to parallel frequency offset estimation and correction to achieve carrier synchronization.

[0060] (5) Extract the preamble information after carrier synchronization, estimate the channel, and obtain the channel state information.

[0061] (6) After carrier synchronization, UW information is extracted from the data and phase tracking is achieved using UW to correct phase noise and residual frequency offset.

[0062] (7) After phase tracking, the signal demodulation is achieved by using constellation diagram mapping.

[0063] The demodulation process employs parallel processing. The number of parallel sampling paths equals the sampling rate divided by the processing clock, while the number of parallel paths for carrier synchronization, phase tracking, and constellation mapping equals the symbol rate divided by the processing clock. By using parallel processing, the processing clock speed is reduced, ensuring that the demodulated signal processing meets the hardware's speed requirements.

[0064] In step (1), the specific method for constructing the multiphase filter is as follows:

[0065] (101) Construct five operations: interpolation, anti-aliasing, matched filtering, channel equalization, and timed decimation.

[0066] (102) Transform anti-aliasing filtering, matched filtering, and channel equalization into frequency domain transfer functions with the same number of points, i.e., with the same frequency resolution, and then obtain the final frequency domain transfer function by multiplying them in the frequency domain.

[0067] (103) After the final frequency domain transfer function is transformed into a time domain filter function, it is further transformed into a polyphase filter structure to obtain a polyphase shaping filter.

[0068] Step (3) is as follows:

[0069] (301) After the data received by the AD passes through the multiphase filter, the timing error of each data frame is estimated by the square law detection method to obtain the timing error information in the form of angle phase.

[0070] (302) The timing error is fed back to the receiving multiphase filter. The receiving multiphase filter adjusts the filter coefficient according to the phase angle information, that is, the position synchronization is achieved by adjusting the extraction position in the receiving multiphase filter.

[0071] Step (5) is as follows:

[0072] (501) After carrier synchronization, extract the preamble sequence of each frame for channel state estimation;

[0073] (502) Obtain the frequency domain channel equalization coefficients based on the channel state information;

[0074] (503) Feed the frequency domain channel equalization coefficient to the receiving polyphase filter. The receiving polyphase filter adjusts the filtering coefficient according to the frequency domain equalization coefficient to correct the flatness of the radio frequency channel and the frequency selective fading of the propagation channel.

[0075] Here are more specific examples:

[0076] The entire modem has a symbol rate of 4.2 Gbaud, supports BPSK-64QAM modulation and demodulation, and has a maximum transmission rate of 25 Gbps. Figure 1 The diagram shows the signal modulation process. Figure 3 The diagram shows the normal signal demodulation process. To reduce resource overhead, it can be further optimized as follows: Figure 4 The demodulation process is shown below.

[0077] In the signal modulation process, shaping filtering and interpolation decimation are employed. Figure 2 The polyphase filter implementation shown; in the resource-optimized signal demodulation process, interpolation, anti-aliasing, timing synchronization, matched filtering, channel time-domain equalization, and decimation are employed. Figure 5 The polyphase filter shown is implemented.

[0078] A 25Gbps single-carrier ultra-high-speed modulation method includes the following steps:

[0079] High-speed information is converted into multi-path parallel data and enters constellation mapping, which maps it into multi-path parallel constellation points (BPSK, QPSK, 16QAM, 64QAM, etc.), with each constellation point consisting of two paths, I and Q.

[0080] The IQ data is subjected to parallel shaping filtering using multiple filters to obtain a 4.8Gsps 32-channel parallel modulation signal, which is then converted into an intermediate frequency signal by a high-speed DAC.

[0081] The number of parallel paths is related to the symbol rate and sampling rate. By using parallel processing, the processing clock is reduced, so that the signal processing meets the hardware processing speed requirements. The symbol rate is 4.2 Gbaud with 28 parallel paths, and the sampling rate is 4.8 Gsps with 32 parallel paths.

[0082] Due to the limitation of high-speed sampling rate, the high-speed sampling rate is not necessarily an integer multiple of the symbol rate. Interpolation and decimation are required to achieve the change between symbol rate and sampling rate. This requires 8 times interpolation and 7 times decimation to achieve the change from 4.2 Gaud symbol rate 28-way parallel to 4.8 Gsps symbol rate 32-way parallel.

[0083] To reduce resource consumption, this method combines the coefficients of shaping filter and anti-aliasing filter to generate a polyphase filter, while simultaneously implementing interpolation decimation and shaping filter functions.

[0084] For the modulated signal, demodulation can be performed using conventional methods, namely:

[0085] The 4.8Gsps sampling signal obtained by AD high-speed sampling is processed through interpolation, multiple matched filtering, timing error estimation, timing decimation, carrier synchronization, channel equalization, phase tracking, and constellation diagram mapping to obtain demodulated information data.

[0086] Due to the inherent sampling rate limitations of AD devices, high-level sampling cannot be achieved with large signal bandwidths, resulting in low oversampling rates that may not be integer multiples. To improve the estimation accuracy of timing errors, interpolation decimation is used to increase and adapt the sampling rate. Specifically, after 7x interpolation, multiple matching filtering and timing error estimation are performed, followed by 8x interpolation decimation.

[0087] The entire demodulation process adopts a parallel processing architecture, in which the matched filtering adopts a polyphase filtering structure to adapt to the high-speed processing requirements at the sampling rate; interpolation decimation and timing synchronization realize the conversion from sampling rate to symbol rate; carrier synchronization, channel equalization, and phase tracking are processed in parallel at the symbol rate.

[0088] Furthermore, the matching filtering, interpolation decimation, timing synchronization, and channel equalization processes consume significant logic and signal processing resources, placing high demands on the hardware signal processing platform. To reduce signal processing complexity and resource overhead, this method converts the anti-aliasing filter in interpolation decimation, the time-domain interpolation in timing synchronization, the matching filtering, and the time-domain channel equalization into a single polyphase filter structure for unified time-domain processing. Specifically:

[0089] 1) The anti-aliasing filter in interpolation decimation, the time-domain interpolation, matched filtering, and channel time-domain equalization in timing synchronization are all equivalent to frequency-domain transfer functions of the same length. Then, the final transfer function is obtained by multiplying them in the frequency domain.

[0090] 2) The frequency domain transfer function is transformed into a time domain filtering structure. In combination with the requirements of high-speed parallel signal processing, multiple filters are used to implement interpolation, anti-aliasing, matched filtering, channel equalization, and decimation functions.

[0091] In this way, a polyphase filter can be used to achieve anti-aliasing filtering, timing interpolation, matched filtering, and channel equalization. The post-processing rate after polyphase filtering is reduced from the sampling frequency to the symbol rate.

[0092] Based on the above analysis, the specific steps of the entire demodulation process can be described as follows:

[0093] (1) The data sampled by AD is first subjected to forward timing error estimation to obtain timing error, which is then transmitted to the polyphase filter in the form of phase angle;

[0094] (2) After carrier synchronization, the data needs to be channel estimated and then converted into equalization coefficients, which are then transmitted to the polyphase filter;

[0095] (3) Based on the timing error, equalization coefficient, matched filter coefficient, and interpolated anti-aliasing filter coefficient, the final filter expression is dynamically generated, and the filter coefficients of each branch are dynamically updated in combination with the multiphase filter structure.

[0096] The working principle of this invention is as follows: Information is transmitted in parallel and mapped to multiple constellation points. After multiphase filtering to achieve program and processing clock conversion, it is sent to the high-speed DA converter. The signal received by the DA converter undergoes multiphase filtering, carrier synchronization, phase tracking, and constellation demapping to obtain the received information. Specifically, the demodulation end performs timing error estimation at the sampling rate; after carrier synchronization, the demodulation end performs channel estimation to obtain equalization coefficients; the demodulation end's multiphase filter combines timing error, equalization coefficients, and matched filter coefficients, simultaneously performing real-time timing, matched filtering, and equalization functions. Therefore, the processing complexity and resource overhead are significantly reduced.

[0097] In summary, this invention employs a parallel architecture to meet the requirements of ultra-high-speed signal processing and performs equalization correction on channel characteristics. This invention combines matched filtering, timing synchronization, and channel equalization, significantly reducing the overhead on signal processing and logic resources. This invention can be applied to millimeter-wave continuous ultra-high-speed data transmission systems, significantly reducing hardware resource consumption.

Claims

1. A low resource overhead ultra-high speed demodulation method, characterized in that, Includes the following steps: (1) Parallel multi-channel signals are obtained through high-speed analog-to-digital sampling, and then a receiving multiphase filter is constructed based on timing error information and channel equalization coefficient; The number of parallel sampling paths for high-speed analog-to-digital sampling is equal to the sampling rate divided by the processing clock value; the number of parallel paths for carrier synchronization, phase tracking, and constellation mapping is equal to the symbol rate divided by the processing clock value. (2) The data is reduced from the sampling rate to the symbol rate by a receiving polyphase filter to obtain parallel multi-channel data; the specific method for constructing the receiving polyphase filter is as follows: (101) Design interpolation, anti-aliasing filtering, matched filtering, channel equalization, and timing decimation operations. Among them, the anti-aliasing filtering, matched filtering, and channel equalization operations are time-domain filtering functions, and the interpolation and timing decimation operations are zero-padding and decimation of time-domain data. (102) By performing FFT transformation with the same number of points, the transfer functions of anti-aliasing filtering, matched filtering, and channel equalization are transformed into frequency domain transfer functions, and then the final frequency domain transfer function is obtained by multiplying them in the frequency domain. (103) The final frequency domain transfer function is transformed into a time domain filter function, and further transformed into a polyphase filter structure. Interpolation and timing decimation are applied to the input and output of the polyphase filter structure to obtain the receiver polyphase shaping filter. The filter expression of the receiver polyphase shaping filter is dynamically generated based on the timing error, equalization coefficient, matched filter coefficient, and interpolation decimation anti-aliasing filter coefficient. The filter coefficients of each branch of the receiver polyphase shaping filter are dynamically updated according to the polyphase filter structure. (3) Extract timing error information from the parallel multi-channel data output by the multiphase filter; (4) Perform parallel frequency offset estimation and correction on the parallel multi-channel data output from the multiphase filter to achieve carrier synchronization; (5) After carrier synchronization, the preamble information is extracted, the channel is estimated, and the channel state information is obtained; (6) Extract UW information from the carrier-synchronized data, use UW to achieve phase tracking, and correct phase noise and residual frequency offset; (7) After phase tracking, the signal demodulation is achieved by using constellation diagram mapping.

2. The low-resource-overhead ultra-high-speed demodulation method according to claim 1, characterized in that, The specific method for step (3) is as follows: (301) For the output of the receiving multiphase filter, the timing error of each data frame is estimated by square law detection to obtain timing error information in the form of angle phase. (302) The timing error information is fed back to the receiving multiphase filter. The receiving multiphase filter adjusts the filter coefficient according to the phase angle information, that is, the bit synchronization is achieved by adjusting the decimation position in the receiving multiphase filter.

3. The low-resource-overhead ultra-high-speed demodulation method according to claim 1, characterized in that, The specific method for step (5) is as follows: (501) After carrier synchronization, extract the preamble sequence of each frame for channel state estimation; (502) Obtain the frequency domain channel equalization coefficients based on the channel state information; (503) Feed the frequency domain channel equalization coefficient to the receiving polyphase filter. The receiving polyphase filter adjusts the filtering coefficient according to the frequency domain channel equalization coefficient to correct the flatness of the radio frequency channel and the frequency selective fading of the propagation channel.

Images