[Paper Review] A Survey of Wideband Spectrum Sensing Algorithms for Cognitive Radio Networks and Sub-Nyquist Approaches

Bashar I. Ahmad|arXiv (Cornell University)|Jan 2, 2020

Cognitive Radio Networks and Spectrum Sensing84 references6 citations

TL;DR

This survey presents a comprehensive analysis of wideband spectrum sensing algorithms for cognitive radio networks, emphasizing sub-Nyquist sampling techniques to overcome high sampling rate and hardware complexity constraints. It compares compressed sensing and alias-free sampling methods, demonstrating that sub-Nyquist approaches enable low-complexity, energy-efficient wideband sensing with competitive detection performance, especially at low SNR and reduced sampling rates.

ABSTRACT

Cognitive Radio (CR) networks presents a paradigm shift aiming to alleviate the spectrum scarcity problem exasperated by the increasing demand on this limited resource. It promotes dynamic spectrum access, cooperation among heterogeneous devices, and spectrum sharing. Spectrum sensing is a key cognitive radio functionality, which entails scanning the RF spectrum to unveil underutilised spectral bands for opportunistic use. To achieve higher data rates while maintaining high quality of service QoS, effective wideband spectrum sensing routines are crucial due to their capability of achieving spectral awareness over wide frequency range(s)\ and efficiently harnessing the available opportunities. However, implementing wideband sensing under stringent size, weight, power and cost requirements (e.g., for portable devices) brings formidable design challenges such as addressing potential prohibitively high complexity and data acquisition rates. This article gives a survey of various wideband spectrum sensing approaches outlining their advantages and limitations; special attention is paid to approaches that utilise sub-Nyquist sampling techniques. Other aspects of CR such as cooperative sensing and performance requirements are briefly addressed. Comparison between sub-Nyquist sensing approaches is also presented.

Motivation & Objective

• Address the challenge of high sampling rates and hardware complexity in wideband spectrum sensing for cognitive radio networks.
• Survey both Nyquist-rate and sub-Nyquist sampling-based wideband sensing algorithms, focusing on practical implementation constraints.
• Evaluate the performance and trade-offs of compressed sensing and alias-free sampling (DASP) for multiband spectrum sensing.
• Identify open research challenges in adaptive sensing, unknown sparsity, and collaborative sub-Nyquist detection for dynamic spectrum access.
• Provide a comparative analysis of sub-Nyquist techniques to guide low-SWAP-C (size, weight, power, and cost) system design.

Proposed method

• Formulate wideband spectrum sensing as a multiband detection problem using binary hypothesis testing over L spectral subbands.
• Classify wideband sensing methods into Nyquist-rate (parallel sensing) and sub-Nyquist approaches, including compressed sensing (CS) and non-uniform sampling (DASP).
• Apply compressed sensing using random projections and sparsity in frequency domain, enabling reconstruction via convex optimization (e.g., L1-minimization).
• Implement alias-free sampling via modulated wideband converter (MWC) architecture, which uses bandpass filtering and sub-Nyquist sampling to preserve signal structure.
• Use performance metrics such as detection probability (P_D), false alarm probability (P_FA), and sampling rate efficiency to compare algorithms.
• Conduct numerical simulations with varying SNR, sampling rate ratios (α/f_Nyq), and sensing time (T_ST) to evaluate detection performance under realistic conditions.

Experimental results

Research questions

• RQ1 How do sub-Nyquist sampling techniques such as compressed sensing and alias-free sampling reduce the data acquisition rate in wideband spectrum sensing?
• RQ2 What are the trade-offs between computational complexity, detection performance, and sampling rate in CS-based versus DASP-based sub-Nyquist sensing?
• RQ3 How does the sparsity level of wideband signals affect the required sampling rate in sub-Nyquist sensing, and what are the implications for dynamic spectrum access?
• RQ4 What are the limitations of current sub-Nyquist methods when the sparsifying basis is unknown or time-varying?
• RQ5 How can collaborative sensing be integrated with sub-Nyquist techniques to improve reliability in fading and hidden terminal environments?

Key findings

• Sub-Nyquist sampling techniques such as MWC and DASP can achieve low false alarm and acceptable detection probability at sampling rates as low as 1.5 times the theoretical minimum, even at low SNR.
• DASP-based methods achieve lower computational complexity and simpler implementation compared to compressed sensing, especially in low-SNR regimes.
• Compressed sensing provides a more robust framework for downstream cognitive radio functions such as PU characterization and signal decoding due to its structured signal reconstruction capability.
• At very low sampling rates (e.g., α/f_Nyq = 0.15), DASP-based sensing outperforms CS in detection quality due to effective aliasing suppression via random time-domain sampling.
• The performance of sub-Nyquist methods is highly sensitive to signal sparsity; assuming maximum occupancy leads to over-conservative sampling, wasting system resources.
• Future systems require adaptive, real-time selection of sensing parameters (e.g., sampling rate, data acquisition scheme) without prior knowledge of sparsity, especially under time-varying channel conditions.