CN110768750A

CN110768750A - Low-complexity sliding decoding method based on bit reversal in OVFDM (orthogonal frequency division multiplexing) system

Info

Publication number: CN110768750A
Application number: CN201910907521.8A
Authority: CN
Inventors: 张鸿涛; 陈莹; 王亚峰; 李道本
Original assignee: Beijing University of Posts and Telecommunications
Current assignee: Beijing University of Posts and Telecommunications
Priority date: 2019-09-24
Filing date: 2019-09-24
Publication date: 2020-02-07

Abstract

The invention provides a low complexity sliding decoding method based on bit reversal in an OVFDM system. In the method, the convolutional code is decomposed into the block code by using the sliding window class, and the decoding complexity is reduced because the length of the sliding window is far less than that of the frame. In addition, in order to obtain lower decoding complexity, the method proposes a bit reversal algorithm and is applied to the block code decoding of each sliding window. Where the maximum absolute value criterion is applied to determine the bits to flip, this criterion can result in a near-optimal bit-flip vector in each iteration, while having a low complexity. Simulation results show that under the condition of not reducing the performance of the bit error rate, the bit flipping algorithm has fewer iteration times compared with the steepest descent algorithm, the complexity of the algorithm is approximately linearly increased along with the increase of the spectral efficiency instead of exponentially increased, and the decoding complexity in the high spectral efficiency process is greatly reduced.

Description

Low-complexity sliding decoding method based on bit reversal in OVFDM (orthogonal frequency division multiplexing) system

Technical Field

The present invention relates to the field of wireless communication technologies, and in particular, to a low-complexity decoding algorithm in an Overlapped frequency domain multiplexing (OVFDM) system with high frequency spectrum coding.

Background

Radio spectrum resources are limited, non-renewable resources, and in recent years, the problem of scarcity of spectrum resources is becoming more serious due to the explosive increase of mobile data traffic. The problem of low utilization efficiency of spectrum resources exists while the contradiction of scarcity of spectrum resources is prominent, so research on extremely high spectrum efficiency coding is urgently needed.

The overlap multiplexing principle states that: the mutual overlapping of data symbols inside and among users of the system is not interference at all, but a beneficial coding constraint relation, and only destructive factors outside the system are interference. Based on the principle of overlapping multiplexing, lie dao teach establish a brand-new waveform coding theory suitable for ultrahigh frequency spectral efficiency, and through the shift overlapping of the data weighting multiplexing waveform frequency domain, a novel coding mode with high spectral efficiency, no coding residue and high coding gain is formed, namely, the OVFDM waveform coding theory.

Currently, technologies for improving the spectrum efficiency of a communication system mainly include ftn (fast triple nyquist), MIMO (massive multiple-input multiple-output) systems, mqam (high amplitude quadrature amplitude modulation), and other key technologies. FTN is an oversampling system, and obtains a smaller spectral efficiency gain by considering only independent samples in the time domain and ignoring independent samples in the frequency domain, whereas OVFDM is an undersampling system, and obtains a spectral efficiency gain of K times (where K is a superposition multiple) by considering independent samples in two time and frequency domains. Large-scale MIMO systems use antenna arrays containing hundreds of antennas, occupying a large amount of space resources, whereas OVFDM can be applied to single antenna systems; although 2048QAM modulation was implemented in the experimental system, since QAM is amplitude-phase modulation, the performance on fading channel is poor, whereas OVFDM has hidden diversity gain and can resist fading.

The main challenge for OVFDM implementation is to reduce the coding complexity. OVFDM can employ a fast sequence decoding algorithm for convolutional codes, however, when the constraint length is large, the difference in the maximum likelihood metrics of the two paths is small, which poses a challenge to find the optimal path. Therefore, OVFDM signals can replace convolutional code decoding methods with block code decoding methods, but the complexity of the block code decoding methods depends largely on the size L of the frame, and therefore signal detection based on reception of the entire frame should be avoided. In practical communication systems, the data should be decoded based on sliding window reception to reduce complexity.

Through gradient or iterative algorithm, one effective search direction is obtained, which can be used to decode the block code in each sliding window in OVFDM system. To further reduce the number of iterations, a sliding window based decoding bit reversal algorithm is proposed and effectively selects the bits to be flipped in each iteration.

Disclosure of Invention

The low complexity sliding decoding method based on bit reversal in the OVFDM system of the present invention is mainly divided into two parts, as shown in fig. 1. First, a sliding decoding method in the OVFDM system is proposed to reduce the decoding complexity. The data length of each sliding window is much smaller than the frame length. Thus, as the frame length increases, the decoding complexity increases at a slower rate; second, a bit reversal algorithm is proposed and applied to block code decoding for each sliding window, where the maximum absolute value criterion is applied to determine the bits to flip. The criterion can achieve a near-optimal bit flip vector in each iteration, while having a low complexity. The bit flipping algorithm has a smaller number of iterations than the steepest descent algorithm without degrading the bit error rate performance.

The low complexity sliding decoding based on bit reversal is as follows:

as shown in FIG. 2, the sliding window is denoted as χ ═ SL-1/K, (SL + X-1)/K]Where SL is the index of the first symbol of the sliding window. The data symbol estimates (denoted as the data symbol estimates) that are not truncated by the multiplexed waveform in the sliding window χ

) Is considered to be relatively reliable. However, a high degree of reliability cannot still be guaranteed because of the estimated data symbols obtained from other incompletely multiplexed waveforms

May affect their reliability.

W_sl(f) Representing the information in the sliding window χ. Can obtain W₁(f)＝V(f)G_χ(f) Wherein the gate function is

χ＝[(sl-1)/K,(sl+X-1)/K]|_sl＝1＝[0,X/K]。W_sl(f) The information in (a) is decoded by a bit inversion algorithm to be described later. Then, the sliding window slides forward to obtain a new sliding window region χ ═ [ (sl + X-K)/K, (sl +2X-K)/K]. Information in the new sliding window is updated to

In a similar manner to that described above,

according to W_sl+X-K+1(f) The inner bit-by-bit flipping algorithm.

In addition, the invention provides a bit reversal algorithm, and the sliding window decoding based on bit reversal is described as follows:

step 100, initializing W₁(f)＝V(f)G_χ(f)。

Step 110, each time according to

And updating the sliding window information.

Step 120, within each sliding window, W_sl(f) Is bit-reversed decoded by steps 140 and 150.

Step 130, in a sliding window χ, the data symbol estimates (denoted as data symbol estimates) that are not truncated portions of the multiplexed waveform

) Is considered to be relatively reliable as the decoded output for each sliding window.

Step 140, the gradient of each bit is calculated. The data bits corresponding to the relatively largest ones of the absolute values of the gradients are inverted and it is checked whether the gradient of the new data vector approaches 0.

If the gradient of some bits after inversion does not fall or rise, the inversion is invalid, and other bits are tried.

Advantageous effects

The low complexity sliding decoding method based on bit reversal in the OVFDM system can effectively reduce the decoding complexity. A bit flipping algorithm is used to decode the block code for each sliding window. By decomposing the convolutional code into block codes using sliding window classes, the decoding complexity is reduced since the sliding window length is much smaller than the frame length. In order to obtain lower decoding complexity, the maximum absolute value criterion of a bit flipping algorithm is applied to obtain an optimal bit flipping vector with the complexity of the same order as the size of the sliding window. Simulation results show that when the Viterbi algorithm is very time consuming (when the overlap factor is greater than 20), OVFDM sliding window block decoding can be performed because the complexity of the sliding window bit reversal algorithm increases approximately linearly rather than exponentially with increasing spectral efficiency.

Drawings

FIG. 1 is a schematic diagram of a low complexity sliding decoding method based on bit reversal in OVFDM system;

FIG. 2 is a flow chart of a sliding window decoding algorithm;

fig. 3 is the effect of OVFDM sliding window size on error performance.

Detailed Description

The implementation steps of the low complexity sliding decoding based on bit reversal are as follows:

200. known pilot data may be transmitted during the initial sliding interval, the pilot data having no previous data and no interference from signals outside the initial processing interval. The sliding interval is smaller than the overlapping weight, namely the front processing section and the rear processing section are partially overlapped.

210. The reliable decision data of the start interval should also be removed in the following interval when sliding towards the following interval. But several or even all of the multiplexed signals within the sliding processing interval may be truncated. When sliding to a subsequent interval, the "reliable" decision data for the sliding interval should also be removed in the subsequent interval, and so on until the end of the data frame.

220. Within each sliding window, the gradient of each bit is calculated. The block code decoding is rapidly carried out by a bit reversal method.

230 invert the data bits corresponding to the relatively largest ones of the absolute values of the gradients and check whether the gradient of the new data vector approaches 0.

240 maintain the inversion if the gradient decreases after some bit inversion.

250 if the gradient after some bit inversion does not rise or fall, the inversion is invalid and other bits (with relatively smaller absolute values) are tried.

The effect of OVFDM sliding window size on error performance is shown in fig. 3. As can be seen from the figure, the error rate performance of the bit reversal decoding algorithm is close to that of the steepest descent algorithm. Furthermore, the bit reversal algorithm has the advantage that it has a smaller number of iterations, since the complexity of using a bit reversal algorithm based on sliding block decoding varies linearly rather than exponentially with the increase in the number of overlapping multiples. The result shows that when the overlapping multiple is high and the Viterbi algorithm is not feasible due to high complexity, OVFDM sliding block decoding can be carried out.

Claims

1. A low complexity sliding decoding algorithm based on bit reversal in OVFDM system is characterized in that:

in the OVFDM system, a sliding decoding method is applied to reduce the decoding complexity;

a bit reversal algorithm is proposed and applied to block code decoding for each sliding window.

2. The sliding coding method according to claim 1, wherein:

the data length of each sliding window is much smaller than the frame length, so as the frame length increases, the decoding complexity increases at a slower rate;

known pilot frequency data can be sent in the initial sliding interval, the pilot frequency data has no data before, signals outside the initial processing interval do not interfere with the pilot frequency data, the sliding interval is smaller than the overlapping weight, namely, the front processing interval and the rear processing interval are partially overlapped;

during the sliding to the following interval, the reliable decision data of the starting interval should be removed in the following interval, but several or even all multiplexed signals within the sliding processing interval will be truncated, and during the sliding to the following interval, the "reliable" decision data of the sliding interval should be removed in the following interval, and so on until the end of the data frame.

3. The bit inversion algorithm of claim 1, wherein:

the bit to be inverted is determined by applying a maximum absolute value criterion, the criterion can obtain a bit inversion vector close to the optimal value in each iteration, and meanwhile, the complexity is low, and the bit inversion algorithm has fewer iteration times compared with the steepest descent algorithm under the condition of not reducing the bit error rate performance;

calculating the gradient of each bit, inverting a plurality of data bits corresponding to the maximum gradient absolute value, and checking whether the gradient of the new data vector approaches 0;

if the gradient after some bit inversion does not fall or rise reversely, the inversion is invalid, and other bits are tried.