CN109618249B

CN109618249B - OFDM-PON receiving end fixed point FFT word length optimization method

Info

Publication number: CN109618249B
Application number: CN201811316980.0A
Authority: CN
Inventors: 何浩; 张俊杰; 唐晓东
Original assignee: Shanghai Liyun Electronics Co ltd
Current assignee: SHANGHAI YUWEI COMMUNICATION TECHNOLOGY CO.,LTD.
Priority date: 2018-11-07
Filing date: 2018-11-07
Publication date: 2021-12-07
Anticipated expiration: 2038-11-07
Also published as: CN109618249A

Abstract

The invention discloses a fixed-point FFT word length optimization method at an OFDM-PON receiving end. The proposed word length optimization method is very simple to obtain an FFT intermediate-level word length mapping table, and can be obtained by a mathematical formula method according to channel parameters obtained by an OFDM receiving end The word length of the intermediate stage is easy to implement in practical engineering; 2. The word length optimization method proposed by the present invention has nothing to do with the word length of the FFT and the modulation format adopted by the OFDM system, and the method is suitable for different received optical powers, and has better robustness. Robustness; 3. The word length optimization method proposed by the present invention has a very high accuracy. When it is set to 0.4dB, the performance of <0.4dB can be achieved by using the optimization strategy proposed by the present invention. 4. The word length proposed by the present invention The optimization method deeply optimizes the FFT intermediate-level word length, not only the precision of the intermediate-level word length (that is, the fractional part of the word length), but also the range of the intermediate-level word length (that is, the integer part of the word length). ), so that the word lengths of all FFT intermediate stages are almost the same.

Description

OFDM-PON receiving end fixed point FFT word length optimization method

Technical Field

The invention relates to the technical field of data processing, in particular to a fixed-point FFT word length optimization method for an OFDM-PON receiving end.

Background

An Optical Orthogonal Frequency Division Multiplexing (OOFDM) technique is a new optical transmission technique, in which a Fast Fourier Transform (FFT) is one of core techniques in the OOFDM system. The speed of optical OFDM system transceivers has reached several tens of Gbits/s, which requires very high throughput rates from the FFT processing unit in the system. For the OOFDM system, the logic resource occupied by the DSP algorithm, especially the calculation of FFT occupies about 80% of the existing FPGA resource. Therefore, the design of FFT also becomes one of the huge challenges for implementing real-time systems.

In order to meet the requirements of a high-speed real-time system and realize an OFDM system with high throughput, the FFT needs to adopt a parallel pipeline structure, and meanwhile, in order to reduce resources required by the implementation of an FFT algorithm, the existing FFT implementation adopts fixed-point FFT, and the resource optimization mainly focuses on word length optimization required by the fixed-point FFT. For the optimization of the word length of the FFT, scholars at home and abroad put forward different research strategies and search algorithms of the optimal word length. The literature [ Bouziane R, Koutsoyannis R, Miller P, et al, optimizing FFT Precision in Optical OFDM transmitters [ J ]. IEEE Photonics Technology Letters,2011,23(20):1550 and 1552 ] mainly analyzes the influence of the FFT input and output word length on the EVM performance of the system under different AD/DA resolutions, but does not analyze the intermediate word length. The document Zhang J, Yuan W Y, Giddings R P, et al, IFFT Stage-dependent Minimum Bit Resolution Maps for Real-time OFDM transmitters [ C ], Optical Fiber Communications reference and inhibition (OFC),2014:1-3 ] carries out the joint optimization design for the DIT-2 structure-based 32-point FFT/IFFT, but does not mention the word length optimization strategy. Document 3[ Wang C Y, Kuo CB, Jou J Y. hybrid Wordlength Optimization Methods of threaded FFT Processors [ J ]. IEEE Transactions on Computers,2007,56(8): 1105) proposes a scheme for searching the intermediate-level word length of FFT using statistical analysis, but the search complexity is high. Document 4[ Yuanwenyan et al, a statistical property analysis-based IMDD-OFDM-PON fixed-point FFT word length optimization method, Chinese patent application: CN107018109A ] only takes into account the word length of the fractional part of the FFT mid-stage when obtaining the word length of the FFT mid-stage, and does not take into account the word length optimization of the integer part of the FFT mid-stage.

Disclosure of Invention

The invention aims to provide a fixed-point FFT word length optimization method for an OFDM-PON receiving end, so as to solve the problems in the background technology.

In order to achieve the purpose, the invention provides the following technical scheme:

a fixed-point FFT word length optimization method for an OFDM-PON receiving end comprises the following steps:

s1, setting m in the data format of the receiving end signal as 1, and obtaining ideal inverse error vector magnitude IEVM of the OFDM-PON channel by using floating point FFT_{channel_dB}Obtaining the PAPR of the signal at the receiving end;

s2, presetting the reverse error vector magnitude IEVM of the fixed point FFT calculation acceptable by the system_totaldBThe IEVM_totaldB≥IEVM_{channel_dB}-β，

S3, setting the word length of the integer part of the output signal of the v-th stage of FFT

The ceil function is rounding up;

s4, setting the fractional part word length of the output signal of the v stage of FFT

The ceil function is rounded up, the

As a still further scheme of the invention: β in step S2 is a system performance degradation tolerance value allowed by the fixed-point FFT.

As a still further scheme of the invention: the number of points of FFT is set as N, a Radix-2 DIT structure is adopted, and the FFT has log₂And N levels, wherein the data format of each level of fixed point number adopts minf, wherein m is the word length of the integer part of the fixed point number, i represents an integer, N is the word length of the decimal part of the fixed point number, and f represents a decimal.

Compared with the prior art, the invention has the beneficial effects that: 1. the word length optimization method provided by the invention is very simple in obtaining the FFT intermediate word length mapping table, can obtain the intermediate word length by a mathematical formula method according to the channel parameters obtained by the OFDM receiving end, and is easy to realize in actual engineering; 2. the word length optimization method provided by the invention is irrelevant to the word length of FFT and the modulation format adopted by the OFDM system, and the method is suitable for different receiving optical powers and has better robustness; 3. the word length optimization method provided by the invention has very high precision, and when the word length optimization method is set to be 0.4dB, the optimization strategy provided by the invention can realize the performance of <0.4 dB; 4. the word length optimization method provided by the invention carries out deep optimization on the FFT intermediate-level word length, not only optimizes the precision of the intermediate-level word length (namely the decimal part of the word length), but also optimizes the range of the intermediate-level word length (namely the integer part of the word length), thus ensuring that the word lengths of all FFT intermediate levels are almost consistent.

Drawings

Fig. 1 is a block diagram of the OFDM-PON fixed-point FFT word length optimization method of the present invention.

FIG. 2 is a flow chart of an OFDM-PON experiment based on intensity modulation direct detection in the present invention.

Fig. 3 is a diagram of an OFDM-PON physical frame structure in an embodiment of the present invention.

Fig. 4 is a mapping table of the word length output of the intermediate stage of the 32/64/128 point FFT in the embodiment of the invention.

Fig. 5 is a graph of 32/64/128 point FFT inverse error vector magnitude verification in an embodiment of the invention.

Fig. 6 is an experimental graph of the inverse error vector magnitude difference under different received optical powers of the 32/64/128 point FFT in the embodiment of the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

Referring to fig. 1-6, in one embodiment:

a fixed-point FFT word length optimization method for an OFDM-PON receiving end is characterized in that the number of FFT points is set to be N, a Radix-2 DIT structure is adopted, and FFT has a log₂N levels, wherein the data format of each level of fixed point number adopts minf, m is the word length of the integer part of the fixed point number, i represents an integer, N is the word length of the decimal part of the fixed point number, and f represents a decimal, and the fixed point FFT word length optimization method comprises the following steps:

s1: setting m in data format of receiving end signal as 1, obtaining ideal inverse error vector magnitude IEVM of OFDM-PON channel by floating point FFT_{channel_dB}Obtaining the PAPR of the signal at the receiving end;

s2: presetting system acceptable fixed point FFT calculated inverse error vector magnitude IEVM_totaldBThe IEVM_totaldB≥IEVM_{channel_dB}- β, said β being a system performance degradation tolerance value allowed by the fixed point FFT;

s3: setting the word length of the integer part of the output signal of the v-th stage of FFT

The ceil function is rounding up;

s4: setting FFractional part word length of FT (field-programmable) output signal

The ceil function is rounded up, the

Example two:

fig. 2 is a strength modulation direct detection optical OFDM transceiver platform constructed in this embodiment, a digital signal modulation and demodulation module of an OFDM transceiver system is completed by offline analysis of MATLAB, and parameters thereof are shown in table 1. The FFT point number is set to 32/64/128 points, the PRBS sequence adopts 21-order pseudo-random code, and the modulation format of the OFDM system is 16-128 QAM.

The physical frame structure of OFDM is shown in fig. 3, each frame includes a frame header with 80 zero sample values and two Training Sequences (TSs), the length of each training sequence is the number of FFT points, and each frame of data includes a plurality of OFDM symbols (500 OFDM data symbols are used in this experiment); the frame header with 80 zeros is used for coarse synchronization of a receiving end, and two Training Sequences (TSs) are used for fine synchronization and channel estimation; the TS is generated using a PN code, and the OFDM data signal is generated using a PRBS sequence. The length of the Cyclic Prefix (CP) before the training sequence and OFDM symbol is set to 16.

In the built experiment platform, the precision of a digital-to-analog converter DAC is 12bits, the precision of an analog-to-digital converter ADC is 10bits, and the digital amplitude limiting rate (CR) is 12 dB. The parameter setting table of the OFDM system is as follows:

based on the hardware platform, the generated OFDM signals are transmitted to an RAM of a Xilinx XC6VLX240T FPGA board card in a User Datagram Protocol (UDP) mode through an Ethernet interface, the running clock of the ML605 FPGA board card is 125MHz, data in the RAM enter a 4GS/s @12-bit DAC after parallel-serial conversion, analog signals output by a DAC module are amplified to generate baseband electric OFDM signals, and finally the generated baseband electric OFDM signals are converted into an optical domain through a narrow-band distributed feedback laser (DFB-LD) and then transmitted on an optical fiber.

At the receiving end, after passing through a 25KM SSMF IMDD system, the receiving optical power is adjusted by a Variable Optical Attenuator (VOA) to evaluate the optical power budget and the system performance of the whole OFDM system. And after the optical signal is converted into an electrical domain through a 2.7GHz PIN tube, the electrical OFDM signal enters an ADC module after being amplified through an adjustable electrical amplifier, and the function of the adjustable electrical amplifier is to ensure that the amplitude of the analog signal can cover the whole dynamic range of the 4GS/s @10-bit ADC. In the experiment, 1M ADC sampling data is stored by using a built-in on-chip RAM of another ML605 FPGA board, and then the data acquired by the FPGA is transmitted to a computer for off-line analysis through a UDP protocol. The off-line demodulation step of the OFDM signal sequentially comprises symbol synchronization, cyclic prefix removal, FFT operation, channel estimation/equalization, and finally analysis is carried out to obtain IEVM and BER of the system. The receiver FFT offline analysis adopts a code with adjustable middle level and based on Cooley-Tukey radix-2 resolution-in-time (DIT), so that the word length of each level of the FFT middle level can be flexibly adjusted on an offline platform to evaluate the influence of the word length of the FFT middle level on the system performance.

In this embodiment, the 10-bit signed number quantized by the ADC is converted into a fixed-point number of 1i9 f. The ideal IEVM performance of the system measured at the different FFT points (32/64/128) was then experimentally obtained using floating point FFT at an optical power of-5 dBm, with a PAPR of the received signal of 12dB, as shown by the red line in fig. 5. In this embodiment, the system IEVM performance penalty β allowed with the fixed point FFT is set to 0.4 dB. According to the calculation formula provided by the invention, an 32/64/128-point middle-level word length mapping table can be obtained as shown in fig. 4. As can be seen from fig. 4, the FFT intermediate-stage integer part word length needs to increase to about 1 bit per two stages, while the fractional part word length decreases by about 1 bit per two stages, so that all FFT intermediate-stage word lengths remain substantially the same, i.e., about 11 bits. Substituting the word length mapping table into the offline analysis platform shown in fig. 2 results in system IEVM performance as shown by the blue line in fig. 5. As can be seen from the blue line of fig. 5, the system IEVM performance degradation obtained using the word length mapping table of fig. 4 is <0.4 dB. The correctness and the accuracy of the embodiment are proved. The IEVM performance degradation test results obtained by the method of the present embodiment with β set to 0.4dB at different received optical powers are shown in fig. 6, from which it can be seen that the method of the present invention can use different received optical powers.

The FPGA resources used by the fully-parallel 32/64/128-point FFT constructed based on the embodiment are shown in Table 2, and compared with the SPIRAL FPGA FFT engineering project based on an open source, the FPGA resources required by the embodiment are less. The number of multipliers required for a fully parallel 32/64/128 point FFT FPGA implementation is tabulated below:

the above embodiments are only for illustrating the technical solutions of the present invention, and are not intended to limit the scope of the present invention, and any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention are included in the protection scope of the present invention.

It will be evident to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, and that the present invention may be embodied in other specific forms without departing from the spirit or essential attributes thereof. The present embodiments are therefore to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference sign in a claim should not be construed as limiting the claim concerned.

Furthermore, it should be understood that although the present description refers to embodiments, not every embodiment may contain only a single embodiment, and such description is for clarity only, and those skilled in the art should integrate the description, and the embodiments may be combined as appropriate to form other embodiments understood by those skilled in the art.

Claims

1. An OFDM-PON receiving end fixed-point FFT word length optimization method, the number of FFT points is set to N, the Radix-2 DIT structure is adopted, the FFT has a total of log ₂ N levels, and the data format of the fixed-point number of each level adopts minf, and the m is the word length of the integer part of the fixed-point number, the i represents an integer, the n is the word length of the fractional part of the fixed-point number, and the f represents a decimal number, which is characterized in that it includes the following steps:

S1. Set m in the data format of the signal at the receiving end to 1, obtain the ideal inverse error vector magnitude IEVM _{channel_dB} of the OFDM-PON channel with floating-point FFT, and obtain the peak-to-average ratio PAPR of the signal at the receiving end;

S2. Preset the inverse error vector magnitude IEVM _totaldB calculated by the fixed-point FFT acceptable to the system, the IEVM _{totaldB ≥} IEVM _{channel_dB} -β, and the β is the system performance deterioration tolerance value allowed by the fixed-point FFT,

S3. Set the word length of the integer part of the output signal of the vth stage of the FFT

The ceil function is rounded up;

S4. Set the word length of the fractional part of the output signal of the vth stage of the FFT

The ceil function is rounded up, the

N is the number of FFT points.