Interference suppression algorithm for wireless communication network based on joint estimation

Abstract

When the wireless communication network is interfered, the communication effect will be affected. In order to improve the interference signal processing effect and the identification accuracy of the interference signal, a wireless communication network interference suppression algorithm based on joint estimation is proposed. Using the deep learning method to identify the interference signal, obtain the effective interference signal of wireless communication network, improve the accuracy of interference signal identification, and track and parameter modulation the identified signal; The node model of wireless communication network is established, and the joint estimation method is used to suppress the interference signal for the nodes in the model. The interference suppression of wireless communication network is realized through the state estimation of single tone interference and narrowband interference. The experimental results show that the proposed algorithm has a high accuracy of interference signal recognition, the highest value reaches 98%, and the wireless communication data packet loss rate is low, the highest value is only 0.37, which verifies its interference suppression effect.

Keywords

Joint estimation wireless communication network deep learning parameter modulation signal tracking

1. Introduction

With the rapid development of communication technology, wireless communication, as an important means of communication, has become an indispensable part of people’s life [1]. However, the rapid development of wireless communication, the rapid expansion of communication scale, the surge in the number of users, the increasing complexity of communication network structure and the large occupation of frequency resources have led to a large amount of interference in the information transmission between and within wireless communication systems, which greatly reduces the accuracy of transmission signals [2]. Therefore, taking effective interference suppression measures is one of the important measures to ensure the reliability of wireless communication [3]. Traditional interference suppression methods always have different degrees of damage to communication signals. At the same time, due to the complexity of the algorithm and the problem of slow convergence, interference suppression has great limitations [4]. In order to improve the problems of traditional interference suppression technology, relevant scholars and experts have conducted a lot of research.

Liu and Xu [5] propose an interference suppression method combining power control and dynamic spectrum allocation algorithm. Based on the traditional soft frequency multiplexing technology, combined with dynamic spectrum allocation technology and power control parameter adjustment, it realizes the authorized call of frequency band resources between adjacent cells. The coverage of user transmission power in the cell is controlled in real time to realize interference suppression. The simulation results show that this method improves the performance of the communication system and has more scientific coverage control. However, when the interference signal is suppressed, there is a problem that the signal is unstable. Dou [6] proposed a carrier communication interference coupling suppression method based on sparse sampling, analyzed the basic principle of sparse sampling and the composition of interference coupling in carrier communication, constructed the measurement matrix of interference signal, and took the interference coupling features extracted from it as the feature vector of communication interference coupling suppression. Based on the multi-target detection interference, the optimization matrix is designed and transformed into column vector to obtain the interference coupling signal to be suppressed. Using the correlation coefficient relationship and the interference coupling edge characteristics in the vector, the edge position of the interference coupling signal component is calculated to complete the interference coupling suppression of carrier communication. The experimental results show that this method not only greatly suppresses the interference signal, but also has no significant impact on the useful signal in the communication system, and the suppression time is short, but this method has the problem of low accuracy in the interference signal recognition. Ao and Zheng [7] designed an interference suppression algorithm for short-range short wave communication system. The noisy part of the signal is processed by ionization processing to determine the signal source. On this basis, the interference signal of short-range short wave system is classified and the interference filtering algorithm is used to suppress the interference of short wave communication system. Experimental results show that the algorithm can effectively control the number of blind spots, shorten the calculation time and improve the interference suppression efficiency, but the algorithm has the problem of large packet loss rate in wireless communication.

Because the interference signal recognition effect plays a very important role in interference suppression, and the above traditional methods have the problem of low accuracy of interference signal recognition in interference suppression, incomplete signal processing and high packet loss rate of wireless communication data. Therefore, this paper proposes an interference suppression algorithm for wireless communication network based on joint estimation. The main technical route of the algorithm is as follows:

(1)
The deep learning method is used to identify the interference signal and improve the accuracy of interference signal identification;
(2)
Tracking and parameter modulation processing are carried out on the identified signal to further improve the signal processing effect;
(3)
The joint estimation method is used to suppress the interference signal, and the joint state estimation method of single tone interference and narrowband interference is used to suppress the interference of wireless communication network.

2. Interfering signal preprocessing

2.1 Interfering signal identification

Interference signal identification is an important part of interference suppression. In the process of communication, if the communication party can effectively identify the type of interference signal, it can take corresponding anti-interference measures to avoid or suppress interference to the greatest extent and reduce the damage of interference to communication quality. Traditional interference identification methods generally extract features manually, and then classify them through threshold comparison or machine learning algorithms. These methods mainly rely on manual extraction of features, and there may be problems such as incomplete or redundant extraction of features, and the complexity of the algorithm is often high, which affects the efficiency of interference signal identification. In view of the above problems, this paper uses the convolutional neural network in the deep learning method to identify the interference signal [8]. This method can speed up the network learning speed and avoid signal loss, which can not only improve the efficiency of signal recognition, but also improve the accuracy of signal recognition.

In order to avoid the influence of interference signals of different strengths on the interference suppression effect and speed up the network learning speed, the power normalization operation is performed on the complex baseband signal $f\left(n\right)$ :

$\displaystyle f\left(n\right)=\frac{f\left(n\right)}{\sum\limits_{i=1}^{N}{f_{% i}\left(n\right)}}$ (1)

Among them, $N$ represents the number of signal sampling points; $f_{i}\left(n\right)$ represents the signal channel.

In addition, in order to reduce the spectral energy leakage caused by time-domain truncation, window processing is performed on the received signal. In this paper, a Hanning window with large sidelobe attenuation is selected, and the received interference signal sequence is set as $x=\left[{x_{1},x_{2},\ldots,x_{n}}\right]^{k}$ , and its standardized mathematical model is:

$\displaystyle x=h+n\in D^{M}$ (2)

Among them, $h$ represents the interference signal; $n$ represents the additive Gaussian noise; $D^{M}$ represents the number of sampling points of the signal sequence.

Preprocess the received interference signal sequence $x$ , represented by $x^{\prime}$ . Two data formats are used as the input of real and complex interference identification networks. The first is time domain input data: the $R$ and $I$ channels of interference signal is $\left[{Re\left({x^{\prime}}\right),\textit{In}\left({x^{\prime}}\right)}\right]$ , and its dimension is $K\times 2$ ; The second is the frequency domain input data: the real part of the interference signal are $\textit{Re}\left({\textit{FFT}\left({x^{\prime}}\right)}\right)$ , and its dimension is $K\times 2$ .

The jammer recognition network is based on the basic elements of the convolutional neural network CNN [9], a common and relatively simple network structure that is often used as the basis for building network models. As shown in Fig. 1, the actual CNN and the interference detection network built on top of the complex CNN have the same network structure [10].

Figure 1.

Schematic diagram of the interference signal identification network.

In the interference signal identification network shown in Fig. 1, the size of the input data is set to $l\times 2$ , where $l$ represents the number of sampling points and 2 represents the number of characteristic channels. The input data is converted into a three-dimensional size $l\times 1\times 2$ suitable for processing by the interference identification network through matrix transformation. Through the convolution network composed of real/complex convolution units and pooling layers, the over fitting of the network is avoided on the premise of ensuring the accurate expression of data characteristics. The pool is a layer average pool, and the window size is 3 $\times$ 1, the step size is 2. There are two convolution layer elements. The convolution kernel step size is 1 by default, and the edge compensation is 0 by default, then through the full connection layer, and finally classified by softmax function. The rectified linear unit (ReLU) is selected as the activation function in real CNN, which converges quickly and can alleviate the problem of gradient disappearance. The concatenated rectified linear units (CReLU) is selected as the activation function in complex CNN, which is identified based on phase information and can retain more information, thus, more effective interference signals of wireless communication network can be obtained to avoid signal loss, so as to improve the accuracy of interference signal recognition.

2.2 Interfering signal tracking

After identifying the interference signal, the tracking calculation is performed on the interference signal. The process is as follows:

Set the signal source parameter of the interference signal to 2, namely $H=2$ , and calculate the acquisition matrix of the frequency band space for the secondary channel frequency band:

$\displaystyle Z\left({h_{k}}\right)=\frac{1}{l}\sqrt{z_{i}\left({h_{k}}\right)% +z_{j}\left({h_{k}}\right)}$ (3)

Among them, $z_{i}\left({h_{k}}\right)$ represents the network coverage; $z_{j}\left({h_{k}}\right)$ represents the spectral efficiency.

The relationship between $z_{i}\left({h_{k}}\right)$ and $z_{j}\left({h_{k}}\right)$ is decomposed, and the locking matrix $Z^{\prime}\left({h_{k}}\right)$ can be obtained according to the decomposed singular values. After obtaining $Z^{\prime}\left({h_{k}}\right)$ , the locked signal co-party tracking amount is obtained according to Eq. (4):

$\displaystyle G=\sum\limits_{k=1}^{N}{Z^{\prime}\left({h_{k}}\right)}\times x_% {ij}^{2}+Z\left({h_{k}}\right)$ (4)

Among them, $x_{ij}^{2}$ represents the strong autocorrelation of the spreading sequence in the interference signal.

The eigenvalue variance calculation is performed on Eq. (4), and the space vector tracking matrix $W_{H}$ of the interference signal can be obtained.

2.3 Modulation of interference signal parameters

After completing the identification and tracking of the interference signal, it is necessary to modulate the parameters of the wireless communication network signal under strong interference to reduce the rate of the pseudo-random code. The modulation and modulation process of the interference signal parameters is shown in Fig. 2.

Figure 2.

Modulation process of interference signal parameter modulation.

It can be seen from Fig. 2 that the autocorrelation function within a chip range is obvious and there is almost no peak, indicating that the wireless communication network signal has good correlation. When the pseudo code phase in Fig. 2 is fully aligned, the peak of the correlation function will be very obvious. If the pseudo code phase deviation exceeds the time length of one chip, the correlation value is 0. When the autocorrelation peak of pseudo-random code and wireless communication network signal is affected by signal-to-noise ratio, the receiver can be used to automatically capture and modulate the wireless communication network signal parameters according to the good autocorrelation of pseudo-random code.

In order to ensure the accuracy of the modulation parameters, the instantaneous phase of the wireless communication network signal needs to be modulated first. The most critical step is to obtain the accurate carrier frequency. The nonlinear phase part of the wireless communication network signal can be estimated as:

$\displaystyle\mu_{k}=\frac{\left|{p_{i}\left(k\right)}\right|-\left|{p_{j}% \left(k\right)}\right|}{\left|{p\left(k\right)}\right|^{2}}\times\left({f_{a}% \times\phi_{b}}\right)$ (5)

Among them, $p_{i}\left(k\right)$ and $p_{j}\left(k\right)$ represent the nonlinear component; $\left|{p\left(k\right)}\right|^{2}$ represents the standard deviation of the absolute value of the nonlinear component of the instantaneous phase of the wireless communication network signal; $f_{a}$ represents the carrier frequency; $\phi_{b}$ represents the unformatted phase sequence. The carrier frequency is calculated by Eq. (5) to ensure the accuracy of the modulation parameters of the interference signal of the wireless communication network.

3. Wireless communication network interference suppression

After the interference signal identification, tracking and parameter modulation, the preprocessing of the wireless communication network interference signal is completed, which improves the signal processing efficiency and accuracy to a certain extent, and is conducive to improving the network interference suppression efficiency. In order to achieve effective suppression of wireless communication network interference, the research on wireless communication network interference suppression will be realized by building a wireless communication network model and joint interference estimation.

3.1 Node model of wireless communication network

The wireless communication network node model constructed in this paper is mainly composed of three parts: data processing unit, processor unit and memory unit. The specific structure of the model is shown in Fig. 3.

Figure 3.

Node model of wireless communication network.

It can be seen from Fig. 3 that the processor unit is the core of the wireless communication network node. It selects the control chip of the ATMegaL128 model as the main component, and completes data collection, processing and transmission and reception together with other units. The processor not only meets the needs of the entire data processing, but also can reduce power consumption and improve the utilization rate of communication network nodes. The data processing unit is mainly responsible for collecting real-time signal data, and the storage unit stores the data collected by the data processing unit.

In the wireless communication network node model shown in Fig. 3, assuming that the distance between communication nodes is $d$ , the signal transmission model is:

$\displaystyle S_{d}=L_{\theta}\sqrt{F_{s}\sum\limits_{i=1}^{N}{\left({g_{i}g_{% j}}\right)^{2}+h_{1}h_{2}}}$ (6)

Among them, $F_{s}$ represents the transmit power of the node; $g_{i}$ and $g_{j}$ represent the antenna gain of the sending node and the receiving node respectively; $h_{1}$ and $h_{2}$ represent the antenna height of the sending node and the receiving node respectively; $L_{\theta}$ represents the interference loss.

Ensure that the variables of node power $F$ , node antenna gain $G$ , and communication node antenna height $H$ remain unchanged, and change the distance $d$ to obtain the change of the packet loss rate, that is, the communication loss at different distances, using Eq. (7) to represent:

$\displaystyle P_{r}=\left({u_{l}}\right)^{s}\left[{\left({a-s}\right)}\right]p% ^{s}$ (7)

Among them, $a$ is a constant; $p^{s}$ represents the distance between the communication node and the main control terminal; $P_{r}$ represents the interference loss, that is, the packet loss rate; $\left({u_{l}}\right)^{s}$ represents the interference loss value, and the interference coefficient value is obtained according to the distance and the packet loss rate.

3.2 Joint estimation of interfering signals

The joint estimation method is used to suppress the interference signal, and the interference problem of the wireless communication network node is further analyzed [11]. When the parameters of the wireless communication network node model are known, the state equation of the received signal is as follows:

$\displaystyle R_{s_{i}}=\frac{c\left(n\right)}{(f_{s_{i}}+n_{s_{i}})}$ (8)

Among them, $c\left(n\right)$ represents the process noise; $f_{s_{i}}$ represents the known interference power; $n_{s_{i}}$ represents the known model parameters.

When the model parameters are unknown, the unknown parameters are represented by Eq. (9):

$\displaystyle m_{s_{i}}=(1-n_{s_{i}})\ast N_{e}$ (9)

Among them, $N_{e}$ represents the state estimation value of node interference.

Assuming that $N_{e}$ is a fixed value, it is used for state estimation of single-tone interference, and will be further extended to narrow-band interference for joint estimation of the state of the interference signal and model parameters. According to the least squares criterion, the extended state vector is expressed as:

$\displaystyle N_{e}=\left[{\beta_{1}\left(n\right),\beta_{2}\left(n\right),% \ldots,\beta_{m}\left(n\right)}\right]$ (10)

Among them, $\beta_{1}\left(n\right),\beta_{2}\left(n\right),\ldots,\beta_{m}\left(n\right)$ all are state vectors of wireless communication network nodes.

When it is narrowband interference, $\beta_{1}\left(n\right),\beta_{2}\left(n\right),\ldots,\beta_{m}\left(n\right)$ are the parameter to be estimated [12]; when it is single-tone interference, the state vector value depends on the unknown interference frequency. Therefore, to jointly estimate the state vector, that is, to extend the state vector in Eq. (10), the corresponding matrix is:

$\displaystyle N_{e}=\left[{{\begin{array}[]{cccc}{\beta_{1}\left({n-1}\right)}% &{\beta_{1}\left(n\right)}&{\beta_{1}\left({n+1}\right)}&0\\ 0&0&1&0\\ 0&1&0&1\\ 0&0&0&1\\ \end{array}}}\right]$ (11)

The state vector time update formula is as follows:

$\displaystyle\beta_{\left({n+1}\right)}=\log_{10}\frac{F\beta_{n}}{P(x)P(y)}$ (12)

Among them, $P(x)$ represents the predicted state vector; $P(y)$ represents the covariance matrix.

The state vector estimation update formula is as follows:

$\displaystyle I_{n}=\frac{\text{In}\lambda\left(n\right)}{\left|{\text{In}% \lambda\left(n\right)+M\left(n\right)}\right|^{2}}$ (13)

Among them, $\lambda\left(n\right)$ represents the filter estimate; $M\left(n\right)$ represents the conditional covariance matrix of $\lambda\left(n\right)$ [13]. If $u_{n}$ and $U_{n}$ are nonlinear terms due to non-Gaussian distributed observation noise, the expressions are:

$\displaystyle u_{n}=\frac{1}{\pi}F\left({\frac{1}{x}}\right)F\left({f\left(x% \right)}\right)$ (14) $\displaystyle U_{n}=\frac{1}{\pi}\int_{-\infty}^{\infty}{\frac{f\left(\mu% \right)}{x\left(\mu\right)}}$ (15)

Among them, $f\left(\mu\right)$ and $x\left(\mu\right)$ both represent the signal-to-noise variance of the Gaussian distribution [14].

Then the signal-to-noise ratio of the signal output is:

$\displaystyle B\left(x\right)^{2}=\sqrt{u^{2}\left(x\right)+u_{i}^{2}\left(x% \right)}$ (16)

Among them, $u^{2}\left(x\right)$ represents the transition matrix; $u_{i}^{2}\left(x\right)$ represents the covariance matrix of the interference estimation vector.

According to the above analysis, in order to reduce the influence of the interference signal on the wireless communication network, from the perspective of joint estimation, the joint estimation of the interference signal is carried out to achieve the effect of improving the interference suppression of the wireless communication network.

3.3 Implementation of wireless communication network interference suppression

In the wireless communication network node model, through blind signal separation, a random number between 0 and 1 is generated at the interfering node [15]. If the random number is less than the threshold $\varphi$ , then this node becomes a candidate cluster head. There are $v$ subcarriers in the Peller domain, and the number of all pilot symbols in a frame is set to $X$ , and the wireless transmission channel contains $c$ paths, then the time domain impulse response of the signal blind separation anti-interference channel is:

$\displaystyle K_{f}=\frac{1}{\sqrt{\left|X\right|}}\int\limits_{x}{f\left(x% \right)}f\left(c\right)dx$ (17)

Among them, $f\left(x\right)$ represents the training length of the training sequence; $f\left(c\right)$ represents the electromagnetic wave interference of wireless communication. Based on the sparse feature of the delay-Doppler domain, the improved single-cycle channel feature registration result of the wireless communication network is obtained as follows:

$\displaystyle\Delta G_{y}=G_{0}+\alpha^{2}G_{i}$ (18)

Among them, $G_{0}$ represents the time width of the communication signal; $\alpha^{2}$ represents the modulation parameter of the time width of the communication signal component; $G_{i}$ represents the time domain center of the communication signal component.

At this time, the characteristic quantity of the communication channel is determined, and the communication signal output after the interference filtering of the wireless communication network node is obtained, and the expression is:

$\displaystyle\beta=\frac{1}{1+\sqrt{\frac{2\ln\partial^{2}}{A\left({\varpi_{k}% +\rho_{k}}\right)}}}$ (19)

Among them, $\partial^{2}$ represents the observed communication signal; $A$ represents the number of communication source signal components; $\varpi_{k}$ represents each communication source signal component; $\rho_{k}$ represents the initial communication signal.

The node density is defined to represent the sampling interval of each round time of the wireless communication network, and the dynamic anti-jamming suppression is performed [16], and the filtering and modulation results are obtained as follows:

$\displaystyle G_{r}=\sum\limits_{t=1}^{N}{\left({P_{i}(t)\Delta T+\sum\limits_% {i=1}^{n}{P_{j}(t)\Delta T}}\right)}$ (20)

Among them, $P_{i}(t)$ represents the initial communication signal transmission amount; $P_{j}(t)$ represents the communication signal state vector; $\Delta T$ represents the communication signal weight vector.

Based on the above analysis, the design of the wireless communication network interference suppression algorithm is realized, and the flow chart of the algorithm design is shown in Fig. 4.

Figure 4.

Flow chart of interference suppression in wireless communication network.

4. Simulation experiment analysis

In order to verify the effectiveness of the proposed wireless communication network interference suppression algorithm based on joint estimation, simulation experiments are carried out.

4.1 Simulation experiment environment and parameter settings

In the method proposed in this paper, the deep learning model is divided into 3 layers, the number of neurons is 45, the learning rate is 0.0001 during the training process, the bath_size is 32, and the optimizer is Adam. Set up different node communication interference models in a space of 300 km $\times$ 300 km $\times$ 300 km, conduct node communication interference experiments in free space and under signal shielding, count data packet reception, and obtain the performance of the entire network. The wireless signal frequency of the node is 2.4 GHz, the data packet sent is 16 Byte, and the signal transmission rate is 250 kb/s. In order to verify the superiority of the wireless communication network interference suppression algorithm based on joint estimation, the interference suppression method of carrier communication based on sparse sampling and the interference suppression algorithm of short-range short-wave communication system are used as comparison methods to analyze the interference suppression effect of different methods. The parameters of the simulation experiment are shown in Table 1.

Table 1
Simulation experimental parameters

Parameter	Parameter value
Transmission bandwidth/kHz	495.05
Subcarrier spacing/Hz	8653.27
Number of effective subcarriers/piece	60
Frequency modulation factor Hz/s	0.125–0.900
Pulse duration/s	0.54–12.5 s
Bandwidth factor	0.1–0.7
Interference carrier offset/kHz	$-$ 500–500
Number of array elements/piece	16
Channel Rice Factor/dB	10

In the above experimental environment, the comparative experimental analysis will be carried out from the three aspects of signal processing stability, signal recognition accuracy and data packet loss rate.

4.2 Experimental results

4.2.1 Signal processing stability

The received interference signals are processed by the proposed algorithm, the carrier communication interference coupling suppression method based on sparse sampling and the interference suppression algorithm of short-range short wave communication system respectively. The comparison results are shown in Fig. 5.

Figure 5.

Signal processing stability comparison results. (a) Original signal; (b) Sparse sampling method; (c) Interference suppression algorithm of short-range short-wave communication system; (d) The proposed algorithm.

It can be seen from Fig. 5 that after the signal is processed by the proposed algorithm, the signal amplitude is relatively stable, with only slight fluctuations, while the signal processed by the traditional sparse sampling method and the short-range short-wave communication system interference suppression algorithm, although compared with the original signal, the signal fluctuation has been suppressed to a certain extent, but there is still a large fluctuation. It shows that the proposed algorithm can effectively process the communication signal and suppress the signal disturbance caused by the interference signal. This is because the proposed algorithm uses the joint estimation method to suppress the interference signal, and the joint state estimation method of single-tone interference and narrowband interference can reduce the influence of the interference signal on the wireless communication network. Therefore, the signal processing effect of the proposed algorithm is better.

4.2.2 Packet loss rate

When the receiver receives a correct data packet, it will send a confirmation packet to the sender. After receiving the confirmation packet, the sender will verify the serial number of the confirmation packet and the cached serial number. If the serial number is the same, change the identification in the code to true, indicating correct reception. If the identification is always false, indicating that the confirmation packet of the receiver has not been received, then the packet loss rate will be generated, need to resend. Because the wireless communication network will produce a large amount of data during operation, when disturbed, it will lead to a certain packet loss rate in data transmission. Therefore, take the packet loss rate as the experimental index to compare the interference suppression effects of different methods. The comparison results are shown in Fig. 6.

Figure 6.

Comparison results of packet loss rate.

According to the experimental statistical results shown in Fig. 6, with the increase of communication distance, the data packet loss rate of different methods basically shows a continuous downward trend. Among them, the data packet loss rate of sparse sampling method and interference suppression algorithm of short-range HF communication system is higher than that of the proposed algorithm, and the maximum data packet loss rate of the proposed algorithm is only 0.37, while the maximum data packet loss rate of sparse sampling method and interference suppression algorithm of short-range HF communication system is 0.75. Through comparison, it can be seen that the proposed algorithm not only solves the communication interference. It reduces the packet loss rate and is easy to implement, which can improve the communication performance of wireless communication network nodes.

4.2.3 Interfering signal identification accuracy

The frequency $F_{z}$ of the wireless communication network interference signal has a linear relationship with the modulation noise $U_{z}$ , and its time domain expression is:

$\displaystyle F_{z}=\sum\limits_{i,j=1}^{N}{P_{ij}\sqrt{a_{ij}}}\left[{1-\frac% {1}{\sqrt{U_{z}}}}\right]^{-1}+\varphi_{k}\eta_{z}$ (21)

Among them, $P_{ij}$ represents the amplitude of the interference; $a_{ij}$ represents the center frequency of the interference; $\varphi_{k}$ represents the frequency modulation coefficient. $U_{z}$ is Gaussian noise with non-zero mean and variance $\sigma^{2}$ . The interfering signal bandwidth is jointly determined by $F_{z}$ and $U_{z}$ . The time-domain real waveform and power spectrum of the noise FM interference signal are shown in Fig. 7, where the sampling frequency is 10 MHz, the number of sampling points is 1024, and the positions of the interference frequency points are random.

Figure 7.

Time-frequency waveform and power spectrum under noise interference. (a) Time-frequency real part waveform; (b) Power spectrum.

Under the conditions shown in Fig. 7, the accuracy rates of the three methods for identifying interference signals are compared, and the comparison results are shown in Fig. 8.

Figure 8.

Comparison results of interference signal identification accuracy.

It can be seen from Fig. 8 that although the sparse sampling method and the short-range short-wave communication system interference suppression algorithm reach 2 iterations, the interference signal identification accuracy of the two remains at a relatively stable level. However, compared with the proposed algorithm, the interference signal recognition effect still has a certain room for improvement. The proposed algorithm has a higher recognition accuracy of interference signals than the sparse sampling method and the short-range short-wave communication system interference suppression algorithm, and the highest value reaches 94%, the overall performance is better. It can be concluded that the proposed algorithm has better identification effect of interference signals, and can provide a reliable reference for interference suppression. This is because the proposed algorithm uses the modified linear unit ReLU as the activation function in the training of the deep learning model. The function converges quickly, which can alleviate the problem of gradient disappearance, and the signal identification based on the phase information can retain more information. Obtain more effective wireless communication network interference signals to avoid signal loss, thereby improving the accuracy of interference signal identification.

5. Conclusion

In order to improve the effect of interference signal processing and improve the accuracy of interference signal recognition, an interference suppression algorithm for wireless communication network based on joint estimation is proposed. The following are the main innovations of this study:

(1)

The deep learning method is used to identify the interference signal, which improves the accuracy of the interference signal identification;

(2)

Tracking and parameter modulation processing are performed on the identified signals to realize preprocessing of interference signals and improve signal processing efficiency;

(3)

On the basis of the wireless communication network node model, the joint estimation method is used to suppress the interference signal, so as to realize the interference suppression of the wireless communication network.

(4)

The experimental results show that the interference signal recognition accuracy of the proposed algorithm is high, the highest accuracy reaches 94%, and the packet loss rate of wireless communication data is low. Compared with the traditional methods, the proposed algorithm has obvious advantages, and its interference suppression effect is verified.

Although the proposed algorithm effectively improves the interference suppression effect of wireless communication network, the signal stability should be further strengthened, which will be the focus of future research.

References

Ullah

Ahmad

Khan

Tahir

Flint

. An hp-shape hexa-band antenna for multi-standard wireless communication systems. Wirel Netw. 2019; 25(3): 1361-1369.

Mardani

Deng

Linnartz

Khalid

. Compensating dynamic nonlinearities in LED photon emission to enhance optical wireless communication. IEEE Trans Veh Technol. 2021; 70(2): 1317-1331.

Vogt

Enzner

Sezgin

. State-space adaptive nonlinear self-interference cancellation for full-duplex communication. IEEE Trans Signal Process. 2019; 67(11): 2810-2825.

Hassan

Ghassemlooy

Zvanovec

Biagi

Huang

. Interference cancellation in MIMO NLOS optical-camera-communication-based intelligent transport systems. Appl Opt. 2019; 58(34): 9384-9391.

Liu

. Radio communication interference suppression scheme in FAO vehicle system. Comput Eng Des. 2019; 40(7): 1829-1834+1845.

Dou

. Simulation of Interference coupling suppression in carrier communication based on sparse sampling. Comput Simul. 2020; 37(6): 142-146.

Zheng

. Research on interference suppression algorithms for short distance HF communication systems. Ship Sci Technol. 2019; 41(22): 151-153.

Brik

Ksentini

. Toward optimal MEC resource dimensioning for a vehicle collision avoidance system: a deep learning approach. IEEE Network. 2021; 35(3): 74-80.

Kim

Lee

Jeong

Ahn

. Sound-based remote real-time multi-device operational monitoring system using a convolutional neural network (CNN). J Manuf Syst. 2021; 58(2): 431-441.

10.

Tripathi

Singh

Vishwakarma

. Convolutional neural networks for crowd behaviour analysis: a survey. Vis Comput. 2019; 35(5): 753-776.

11.

Tan

Chen

. Joint channel estimation and power allocation for the CRS-NOMA. Chinese J Electron. 2020; 29(1): 177-182.

12.

Chandrasekhar

Kumaraswamy

. Performance evaluation of pisarenko harmonic decomposition and music-like algorithms for narrowband spectrum sensing in cognitive radio. J Sci Ind Res. 2019; 78(7): 473-477.

13.

Zheng

Chen

Cui

. Hypothesis testing on linear structures of high-dimensional covariance matrix. Ann Stat. 2019; 47(6): 3300-3334.

14.

Montalvo

. Noise variance estimation through joint analysis of intrinsic dimension and differential entropy. IEEE signal process lett. 2019; 26(9): 1330-1333.

15.

Dabiri

Sadough

Khalighi

. Blind signal detection under synchronization errors for FSO links with high mobility. IEEE Trans Commun. 2019; 67(99): 7006-7015.

16.

Sanguinetti

Björnson

Hoydis

. Towards massive MIMO 20: understanding spatial correlation, interference suppression, and pilot contamination. IEEE Trans Commun. 2020; 68(1): 232-257.