Abstract
With the rapid development of wireless communication and the increase of various types of communication services, cognitive radio based on spectrum sensing is becoming a hot research topic. One of the key tasks of spectrum sensing is signal parameter identification, which is to quickly identify a series of parameters of signal based on signal detection, to provide basis for subsequent spectrum allocation and sharing. Bandwidth is one of the important parameters of communication signal. It is the premise of detecting "spectrum holes" to estimate bandwidth quickly and accurately. Due to the difficulty of Nyquist sampling in wideband spectrum sensing, this paper studies undersampling bandwidth estimation. We introduce the power spectrum estimation algorithm under the assumption of wide stability and put forward the corresponding bandwidth estimation strategy. The limitation of the strategy is analyzed. The performance of the algorithm is simulated under different SNR and compression ratios. Different random undersampling matrix is researched as well. The simulation shows that the algorithm discussed in this paper is feasible and reliable.
This work is supported by National Natural Science Foundation of China (NSFC) (Grant No. 61671176).
Access provided by Autonomous University of Puebla. Download conference paper PDF
Similar content being viewed by others
Keywords
1 Introduction
Recently, with the development of wireless communication technology, the spectrum resources are increasingly scarce. The concept of cognitive radio was proposed in 1999 [1]. In short, what cognitive radio needs to do is to effectively use "spectrum hole" to communicate [2].
One of the basic tasks is signal parameter identification. However, a wide band will cause great pressure on the ADC of the receiver. This paper focuses on algorithms based on undersampling. The basic point is to estimate the bandwidth based on the undersampling power spectrum estimation. From the point of view of cyclostationarity, Reference [3] systematically presents how to recover the cyclic spectrum of the signal in the framework of compressed sensing (CS).
In [4, 5], the problem of power spectrum estimation under the new sampling framework is studied. In [6, 7], a linear algorithm combining power spectrum estimation and wavelet edge detection is studied.
The rest of this paper is organized as follows: Sect. 2 introduces the algorithm of power spectrum estimation based on undersampling. After that, the corresponding bandwidth estimation strategy is given in Sect. 3. In Sect. 4, numerical simulations are conducted to evaluate the performance of the proposed algorithm. Finally, conclusions are drawn in Sect. 5.
2 Power Spectrum Recovery Algorithm of Undersampling Signal
We regard the modulated signal as a wide and stable random signal. According to the literature [8], under the framework of compressed sensing (CS), the sampling rate of CS receiver is \(f_{{{\text{s}},{\text{cs}}}} = \left( {M/N} \right)f_{{\text{s}}}\). fs is Nyquist sampling rate. \(M/N \in (0,1]\) is the compression rate. Such a linear compression sensing process can be described as:
For x (n), its autocorrelation function is independent of time:
According to Wiener-Khinchin law, we have
where \(s_{x} = \left[ {s_{x} \left( 0 \right), \ldots ,s_{x} \left( {N - 1} \right)} \right]^{T} ,r_{x} = \left[ {r_{x} \left( 0 \right), \ldots ,r_{x} \left( {N - 1} \right)} \right]^{T}\).
The mapping relationship between (1) and (2) can be written as follows:
Obviously, \({\text{vec}}\left\{ {R_{x} } \right\}\) can be linked directly with \(r_{x}\) by a linear mapping matrix \(P_{N} \in \left\{ {0,1} \right\}^{{N^{2} \times N}}\)
The specific form of \(P_{N}\) is:
The autocorrelation matrix \(R_{z} { = }E\left\{ {z*z^{H} } \right\} \in R^{M \times M}\) still has \(M\left( {M + 1} \right)/2\) degrees of freedom. Using the approach above, we can get
Due to \(R_{z} { = }E\left\{ {z*z^{H} } \right\} \in R^{M \times M}\), the relationship between \(R_{z}\) and \(R_{x}\) is:
Using the properties of matrix, \({\text{vec}}\left\{ {UXV} \right\} = \left( {V^{T} \otimes U} \right){\text{vec}}\left\{ X \right\}\) (\(\otimes\) stands for Kronecker-product) we can get:
By introducing (3), the linear relationship between power spectra is:
where \(\Psi { = }\Phi F^{ - 1}\). The size of \(\Psi\) is \(\left( {M\left( {M + 1} \right)} \right)/2 \times N\). The compression rate has a maximum:
If the received signal is sparse in the frequency domain, then L1 norm regularization can be introduced to obtain convex problems to guarantee sparsity:
These convex problems can be solved by existing convex optimization software kits (such as CVX toolkit based on MATLAB).
3 Bandwidth Estimation Strategy Based on Recovered Power Spectrum
In this section, we mainly consider how to estimate the bandwidth according to the power spectrum. The signal in Fig. 1 is BPSK signal, with signal-to-noise ratio of 5 dB, sampling rate of 40KHz, carrier frequency of 16KHz, bit rate of 1 kHz, and hyper parameter of 2. The strategy to directly estimate bandwidth is shown in Table 1.
4 Numerical Simulations
4.1 The Relationship Between Bit Rate and Bandwidth Estimation
The modulation signal simulated in this section is BPSK signal. Most of the simulation conditions are the same as Fig. 1. According to Fig. 2, the computational complexity of the least square method is much lower than that of the convex optimization method, and it can be more accurate.
4.2 The Performance of Bandwidth Estimation Against Noise
Most of the simulation conditions are the same as Fig. 1. Bit rate is 1.6 KHz.
According to Fig. 3, the computational complexity of convex optimization is not only higher (as shown in Table 2), but also has poor performance. When the SNR is low, as shown in Fig. 4, the power spectrum estimation has a large deviation with high probability. Therefore, in the later simulation, we use the least square method.
4.3 The Choice of Undersampling Matrix
We choose random 01 matrix, Bernoulli matrix, and Gaussian matrix to analyze. Most of the simulation conditions are the same as Fig. 2. According to Fig. 5, only the random 01 matrix can give reasonable bandwidth estimation results. In addition, the hardware implementation difficulty of random 01 matrix is the lowest.
5 Conclusion
In this paper, we first deduce the algorithm of undersampling power spectrum estimation under the assumption of wide stability and then give the bandwidth decision strategy. We analyze the performance of the algorithm from the aspects of bit rate, SNR, sampling matrix, and compression ratio.
The simulation results show that the least square method has great advantages in computation and estimation accuracy, which means for bandwidth estimation, sparse constraints are unnecessary. Meanwhile, random 01 matrix is a suitable undersampling matrix.
References
Liu X, Jia M, Zhang X, Lu W (2019) A novel multichannel internet of things based on dynamic spectrum sharing in 5G communication. IEEE Internet Things J 6(4):5962–5970
Liu X, Zhang X (2020) NOMA-based resource allocation for cluster-based cognitive industrial internet of things. IEEE Trans Indus Inf 16(8):5379–5388
Hong S (2011) Multi-resolution bayesian compressive sensing for cognitive radio primary user detection. In: IEEE global telecommunications conference
Ariananda DD, Leus G (2012) Compressive wideband power spectrum estimation. IEEE Trans Signal Process 60(9):4775–4789
Lexa MA, Davies M, Thompson J et al (2011) Compressive power spectral density estimation. In: IEEE international conference on acoustics. IEEE
Tian Z, Giannakis GB (2007) Compressed sensing for wideband cognitive radios. In: IEEE international conference on acoustics. IEEE
Ariananda DD, Leus G (2011) Wideband power spectrum sensing using sub-Nyquist sampling. In: IEEE international workshop on signal processing advances in wireless communications. IEEE
Tian Z, Tafesse Y, Sadler BM (2012) Cyclic feature detection with sub-Nyquist sampling for wideband spectrum sensing. IEEE J Sel Topics Signal Process 6(1):58–69
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2021 The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.
About this paper
Cite this paper
Gu, Y., Gao, Y., Wang, S., Li, B. (2021). Bandwidth Estimation Algorithm Based on Power Spectrum Recovery of Undersampling Signal. In: Liang, Q., Wang, W., Liu, X., Na, Z., Li, X., Zhang, B. (eds) Communications, Signal Processing, and Systems. CSPS 2020. Lecture Notes in Electrical Engineering, vol 654. Springer, Singapore. https://doi.org/10.1007/978-981-15-8411-4_82
Download citation
DOI: https://doi.org/10.1007/978-981-15-8411-4_82
Published:
Publisher Name: Springer, Singapore
Print ISBN: 978-981-15-8410-7
Online ISBN: 978-981-15-8411-4
eBook Packages: EngineeringEngineering (R0)