The large-scale deployment of wireless sensor networks is indispensable to the success of Internet of Things. Considering dynamic spectrum access and the limited spectrum resources in cognitive radio sensor networks, sub-Nyquist spectrum sensing based on the modulated wideband converter is introduced. Since the transmission signals are usually modulated by different carrier frequencies, the interested spectrum can be modeled as the multiband signal. Modulated wideband converter (MWC) is an attractive alternative among several sub-Nyquist sampling systems because it has been implemented in practice and the frequency support reconstruction algorithm is the most important part in MWC. However, most existing reconstruction methods require the sparse information, which is difficult to acquire in practical scenarios. In this paper, we propose a blind multiband signal reconstruction method, referred to as the statistics multiple measurement vectors (MMV) iterative algorithm to bypasses the above problem. By exploiting the jointly sparse property of MMV model, the supports can be obtained by statistical analysis for the reconstruction results. Simulation results show that, without the sparse prior, the statistics MMV iterative algorithm can accurately determine the support of the multiband signal in a wide range of signal-to-noise ratio by using various numbers of sampling channels.

Sub-Nyquist Spectrum Sensing Based on Modulated Wideband Converter in Cognitive Radio Sensor Networks

Frequency hopping (FH) communications are widely used in Satcom systems. The latest FH systems have exceeded 3 GHz with hop rates up to 100 000 hops/s. Traditional spectrum sensing systems hardly monitor such signals. Modulated wideband converter (MWC) is an emerging compressive sampling structure applied to sample multiband signals. This article analyses the effects of the nonlinearity of analog devices and proposes an optimization method based on a greedy algorithm for mixing sequences to make reconstruction performance uniform across sub-bands. To correct the errors caused by the nonideality of analog front end circuits between the theoretical sensing matrix and the practical one, we also develop a calibrating method that obtains all the sensing matrix coefficients through a single measurement. In conventional MWC, out-band noise will be blended during reconstruction. Instead of matrix factorization, we reserve signal characters and designed sliding filters to improve narrowband signals reconstruction sensitivities. We produced a four-branch MWC principle prototype for FH signal detection. The sensing bandwidth is 3 GHz, and the sampling rate is 400 MHz, whereas the reconstruction sensitivity is as low as 13-dB in-band signal-to-noise.

Implementation of Mixing Sequence Optimized Modulated Wideband Converter for Ultra-Wideband Frequency Hopping Signals Detection

A Double Screening Orthogonal-Matching-Pursuit Algorithm for Compressed Sensing Receiver with High Column Correlation Sensing Matrix

The modulated wideband converter (MWC) is a recently proposed compressive sampling system to acquire the blind multiband signals, which we can only know the number of bands and their widths. The process of quantization is inevitable in engineer realization. In this paper, we consider the limiting case of 1-bit quantization, which preserves only the sign information of measurements, and a group binary iterative hard thresholding lp method solving multiple measurement vector problems (M-GBIHTlp) was proposed to recover the wideband signals under MWC system. The proposed algorithm utilizes the group sparsity of recovered signal, of which the nonzero locations are piecewise together. Experiments show that the proposed algorithm achieve higher reconstruction probability than the existed simultaneous binary iterative hard thresholding I2 norm (SBIHTI2) algorithm, particularly in low signal to noise ratio (SNR).

Chang Chen

Papers

Recovery of multiband signals using group binary compressive sensing