Circularly symmetric algorithm for UWB RF signal receiving channel based on noise cancellation

2.1 Noise cancellation structure for input stage of receiving channel

The noise cancellation mechanism consists of three stages, namely, the input stage, the intermediate stage and the output stage. The input stage adopts the complementary cascade structure [7], because the cascade structure has good input matching in the broadband range, and the input impedance is about 1 / g m . Intermediate stage amplifier has a very high gain and cascade intermediate stage amplifier, so that the gain reaches a certain value, which is suitable for the receiving system. The output stage uses a low gain original follower and has a good output matching [8,9,10,11]. Therefore, the noise cancellation structure of the input stage of the receiving channel is constructed, which is described in Figure 1. Two cascade configurations are used in the input stage to construct the noise cancellation technology, which ensures channel noise cancellation in the channel, reduces noise and improves gain.

Figure 1

Signal noise diagram of input stage amplifier.

As shown in Figure 1, because the L S 3 and L S 4 are small, the computation is ignored, and the transconductance of the structure is as follows:

According to the formula, g m 1 is current and R D 4 is voltage.

The channel noise current I ¯ n 1 2 = 4 k T γ g m 1 generated by transistor M 1 will generate noise voltage at the source of transistor M 1 . It can be transmitted to the gate of M 3 and M 4 through inductance L in and C block . M 1 and M 2 share a common source, i.e., point A , so that the input voltages of M 1 and M 4 , M 2 and M 3 are the same. Therefore, the noise voltage of M 1 is used to control transistor M 2 and M 3 , which makes M 2 and M 3 produce size controlled current I ¯ n 1 → n 2 2 = I ¯ n 1 2 G m 1 2 ⋅ G m 2 2 . After amplification, V ¯ n 1 → n 2 2 is output at point B , the expression is shown in equation (3):

M 1 produces a noise voltage at the drain, as shown in equation (4):

Similarly, the channel noise of M 2 is amplified by M 1 and M 4 to output V ¯ n 1 → n 2 2 V ¯ n d 1 2 B as shown in equation (5) at point, while the noise voltage V ¯ n d 1 2 produced by M 2 at the drain is shown in equation (6):

In order to make the channel noise cancellation, it should satisfy:

V ¯ n d 2 2 = V ¯ n 2 → n 1 2 V ¯ n d 1 2 = V ¯ n 1 → n 2 2 .

The optimal transconductance of the receiving channel can be obtained as follows:

2.2 Main parameters of receiving channel

UWB RF signal receiving channel with high quality should not only complete the frequency conversion work under the RF signal but also ensure that the main parameters of the signal in the receiving process have better performance [12,13,14]. The main parameters of the channel are as follows: input signal frequency: 2 , 100 – 2 , 350 MHz ; input signal level range: − 90 to − 30 dBm ; noise figure: ≤6 dB; gain: 50 dB ± 1.5 dB ; mirror suppression: > 80 dBc ; output signal frequency: 215 – 265 MHz ; input third-order intermodulation value: > − 35 dBm ; intermediate frequency output spurious suppression: >65 dBc ; and output signal level value: ≤ 10 dBm .

2.3 UWB RF signal receiving method

UWB RF signal is received reasonably by selecting the RF devices of the UWB RF signal receiving channel through a five-way power divider [15,16,17]. The UWB RF signal receiving method is amplified by the first filter. After three layers of bi-power divider, one RF signal is divided into five branches for down conversion and amplification filtering, to obtain five branches of intermediate frequency signal with 240 MHz as the central frequency and output.

According to the design index of UWB RF signal receiving channel, each receiving channel requires a gain of 50 dB and a noise figure of ≤ 6 dB . Noise figure is an important index to determine the ability of receiving channel to receive small signal. The following is the noise figure formula of the whole system under cascade condition:

According to the formula, F is the noise figure. In order to obtain a lower noise figure, it is necessary to minimize the loss and gain of the front-end circuit in the receiving channel.

From the principle block diagram of the receiving channel, we can see that since each receiving channel has only one RF signal input but produces five independent IF signal outputs, power divider will be added to the design of the front-end circuit for signal shunting, which will introduce attenuation in the front-end circuit, resulting in the noise figure of the receiving channel. Therefore, how to divide the signal and choose the front-end amplifier reasonably is the key to achieve the noise figure index requirement of the algorithm in this study [18,19,20]. Considering many factors, the HMC753LP4E amplifier (Hittite Company) is selected, with gain 16.5 dB , noise figure 1.5 dB , output 1 dB and compression point 18 dB ; and power divider is BW491SM4. It is shown that the noise figure of the channel is about 4.33 dB , which meets the requirements of the index.

The size of the third-order intermodulation signal of the receiving channel reflects the anti-interference ability of the receiving channel when multiple signals are input at the same time. Combined with module gain, the index is converted to output [21], that is, “output third-order intermodulation value I IP3 > 0 dBm .” According to the receiving channel, the devices affecting the index are the amplifier, attenuator and mixer on the link. The requirements of device indicators and the actual performance of device indicators are shown in Table 1.

From the output I IP3 of Table 1 to the situation analysis table, it can be seen that in order to meet the requirement of “output third-order intermodulation I IP3 > 0 dBm ,” the output third-order intermodulation value of all levels of devices must be satisfied, and the gain of the subsequent link of the device is more than 15 dBm . Receiving channel I IP3 value is mainly determined by the last stage amplifier [22], which requires the highest performance of the last stage amplifier. This index can be improved by choosing a high linearity amplifier. In the receiving channel scheme, the final amplifier PMA-5451+ is determined.

Combined with the receiving channel scheme, there are low-pass filter, limiting attenuation and impedance matching circuits after the last stage amplifier PMA-5451+. The losses brought by these circuits are 7 dB , so the receiving channel O IP3 is 20.9 dB . Through the following calculation process, the input I IP3 value is obtained: I IP3 = O IP3 − G = 20.9 − 50 = − 29.1 dBm , which meets the requirements of the third-order cutoff point > − 20.5 dBm in the input band of the receiving channel.

2.4 UWB RF signal receiving channel based on circularly symmetric Gabor transform (CSGT)

CSGT is applied to the UWB RF signal receiving channel to reduce the redundancy in the channel and ensure the strict rotation invariance of the channel.

3.1 Wavelet denoising for communication signals

The experiment uses MATLAB simulation software to denoise a TH-PPM-UWB communication signal with an SNR of −10 dB and Gauss white noise, using the interference estimation algorithm, the ESPRIT algorithm and the algorithm in this study. Figure 2 shows a UWB TH-PPM-UWB communication signal transmitted at the transmitter, that is, a noiseless signal.

Figure 2

Non-noise TH-PPM-UWB signal.

Figure 3 shows a UWB TH-PPM-UWB communication signal with an SNR of −10 dB received by the receiver after the transmission of additive white Gaussian noise channel. It can be seen that the signal is submerged in the noise due to the influence of noise.

Figure 3

TH-PPM-UWB signal with an SNR of −10 dB.

As can be seen from Figure 3, in the UWB TH-PPM-UWB communication signal processing, the unprocessed signal is disorderly, and the signal is submerged in noise.

Figure 4 shows the simulation result of the denoised signal with an SNR of −10 dB after the wavelet denoising treatment by the interference estimation algorithm.

Figure 4

Denoising results of the interference estimation algorithm.

Figure 4 shows that the signal quality has been improved after denoising by the interference estimation algorithm, but because of the constant deviation, the information of the signal is lost more and there is a certain distortion.

Figure 5 shows the simulation result of the denoised signal with an SNR of −10 dB after the wavelet denoising treatment by the ESPRIT algorithm; Figure 6 shows the simulation results of the denoised signal with an SNR of −10 dB after the wavelet denoising treatment by the proposed algorithm.

Figure 5

Result of the ESPRIT algorithm de-noising.

Figure 6

Result of the de-noising algorithm in this study.

According to Figure 5, the effect of ESPRIT algorithm is not ideal, which is caused by its own discontinuity, and the noisy signal oscillates in some areas.

According to Figure 6, the algorithm guarantees thesignal quality in the process of processing communication signals, and the signal information is transmitted steadily, and the denoising effect is obviously improved.

In the experiment, SNR and root mean square error (RMSE) are used to compare the denoising effects of different denoising methods. The SNR and RMSE are defined as follows:

where s ( i ) is the original signal and s ˆ ( i ) is the signal processed by wavelet. Table 2 presents the SNR and RMSE of UWB TH-PPM-UWB communication signal after denoising by three algorithms.

According to Table 2, after denoising by the three algorithms, the SNR of the proposed is 3.8672, the RMSE is 0.4078, which is 2.6754 and 1.531 higher than that of the interference estimation algorithm and the ESPRIT algorithm, respectively. The RMSE of the proposed algorithm is 0.1246 and 0.0848 lower than that of the interference estimation algorithm and the ESPRIT algorithm. It shows that the proposed algorithm can effectively suppress noise, improve the detection SNR and create a good condition for UWB communication signal detection.

3.2 Performance analysis of receiving channel

In order to test the performance of UWB RF signal receiving channel based on the proposed algorithm, link budget simulation, third-order intermodulation, high-order harmonic and mirror frequency suppression simulation are carried out for the channel, aiming at the gain, noise figure, linearity of the channel and the channel’s suppression range for spurious, harmonic and mirror frequency interference signals, respectively. Next, the proposed algorithm is used to analyze one way of the five receiving branches.

3.2.1 Link budget analysis

The link budget simulation schematic diagram of the receiving channel is shown in Figure 7. In ADS software, the proposed algorithm is used to simulate the channel budget controller through the link budget to analyze the parameters of the channel. In Figure 7, B is the broadband; N C is the center frequency; N F is the noise figure, the unit is dB . The frequency of the input RF signal is 2,125 MHz and the power is − 90 to ∼ − 30 dBm . The simulation results show that the channel gain, noise figure, cutoff point of output third-order intermodulation and 1 dB compression point change with input power. Table 3 presents the simulation result of link budget for small signal, and Table 4 presents the simulation result of link budget for large signal.

Figure 7

Budget simulation diagram of receipt channel link.

Table 3

Comparison of SNR and RMSE of three algorithms

	Disturbance estimation method	ESPRIT algorithm	This study’s algorithm
Improved SNR	1.1918	2.3362	3.8672
RMSE	0.5324	0.4926	0.4078

Table 4

Results of link budget simulation for small signals

Meas_Name	BPF1	BPF2	BPF3	BPF8	ATTEN1
Device noise	2	2	2.999	3	7
Total noise	2	3.541	4.09	4.242	4.242
Output power	−52.002	−37.505	−30.508	−13.51	1.287
Output gain	−2.002	12.495	19.492	36.49	51.287
Output P 1 dB	1,000	15.964	14.711	13.773	6.961
Export O IP3	1,000	27.998	26.357	224.582	17.871

The simulation results show that the channel gain, noise figure, cutoff point of output third-order intermodulation and 1 dB compression point change with input power. Table 4 presents the simulation result of link budget for small signal, and Table 5 presents the simulation result of link budget for large signal.

1. When the input signal amplitude is − 60 dBm , Table 3 presents the result of calculating the input signal amplitude through the proposed algorithm. The total gain of the channel is about 51.3 dB , the noise is 4.24 dB , the values of O IP3 and P 1 dB are 17.9 dBm and 6.96 dBm , respectively, and the output power is 1.28 dBm .

2. When the input signal amplitude is − 30 dBm , Table 4 presents the result of calculating the input signal amplitude through the proposed algorithm. The total gain of the channel is about 39.7 dB , the noise is 4.24 dB , the values of O IP3 and P 1 dB are 17.9 dBm and 6.96 dBm , respectively, and the output signal value is 9.73 dBm .

Table 5

Results of link budget simulation for large signal time

Meas_name	BPF1	BPF3	MIXER	BPF8	ATTEN1
Device noise	2	2.999	1	3	7
Total noise	2	4.09	4.193	4.242	4.242
Output power	−32.002	−10.51	−12.518	6.355	9.731
Output gain	−2.002	19.49	17.482	36.335	39.731
Export P 1 dB	1,000	14.711	5.617	13.773	6.961
Export O IP3	1,000	26.357	17.096	24.582	17.871

These results all meet the requirements of parameters of the UWB signal receiving channel, which shows that the link budget of receiving channel under the proposed algorithm meets the requirements of both small signal and large signal.

3.2.2 Third-order intermodulation analysis

Dual-tone signals with a power of − 60 dBm and frequencies of 2,125 MHz and 2,130 MHz are input to the receiving channel. In ADS, the proposed algorithm is used to simulate and analyze the non-linear characteristics of the channel through the “HARMONIC BALANCE” harmonic simulator. Schematic diagram of the third-order intermodulation simulation is shown in Figure 8, and the third-order intermodulation result diagram is described in Figure 9.

Figure 8

Three-level intermodulation simulation diagram.

Figure 9

Level 3 crossover result map.

From the results in Figure 9, it can be seen that when the input dual tone signal is − 60 dBm , the suppression degree of the IF output signal to its third-order intermodulation signal is about 53 dBc , which is substituted into the formula:

(20) I IP3 = − 60 dBm + Δ / 2 .

The value of the input third-order intermodulation coefficient I IP3 is − 33.6 dBm , which is larger than the required value − 35 dBm . It shows that the third-order intermodulation coefficient of the signal receiving channel under the control of the algorithm in this study meets the requirements of the indicators.

3.2.3 Analysis of high-order harmonic and frequency suppression

The combined frequencies generated by the non-linear characteristics of mixers and amplifiers are amplified by harmonic analysis to observe the impact of these interferences on the system. The simulation principle is shown in Figure 10. In the simulation results of high-order harmonics (Table 5), Mixer (1) denotes the number of harmonics of RF signal 2 , 125 MHz and Mixer (2) denotes the number of harmonics of local oscillator 2,365 MHz . The results show that the high-order combination components are very small except for the frequency combination required for the design of the algorithm in this study.

Figure 10

Signal output results of different mirror frequencies.

The results show that the high-order combination components are very small except for the frequency combination required for the design of the algorithm in this study (Table 6).

Table 6

High-order harmonic simulation results

Frequency	Mix (1)	Mix (2)	P out (dBm)
240 MHz	1	−1	−8.852
480 MHz	2	−2	−423
1.855 GHz	−1	2	−501
2.125 GHz	0	1	−510
2.365 GHz	1	0	−492
2.605 GHz	2	−1	−509
4.010 GHz	−1	3	−529
4.250 GHz	0	2	−532
4.490 GHz	1	1	−538
4.730 GHz	2	0	−534
4.970 GHz	3	−1	−536
6.375 GHz	0	3	−553
6.615 GHz	1	2	−555

Figure 10 shows the intermediate frequency output signal corresponding to 2 , 125 MHz , and the intermediate frequency output signal corresponding to 2 , 605 MHz . It can be seen that the output of the mirror frequency signal 2 , 605 MHz under the control of the proposed algorithm is very small when it is input into the receiving channel system, which meets the requirements of the mirror frequency rejection index.