基于贝叶斯压缩感知的协作频谱感知算法

80 浏览量更新于2024-08-28 收藏 1.44MB PDF 举报

本文主要探讨了一种基于贝叶斯压缩感知的协作频谱感知算法（Cooperative Bayesian Compressed Spectrum Sensing, C-BCSS）。在认知无线电（CR）网络中，随着多节点的参与，接收信号通常存在高度相关性，这可能导致传统宽带谱 sensing 的效率受限。针对这一问题，研究者们提出了一种创新的解决方案。 C-BCSS 算法的核心在于利用压缩感知理论，其目的是提高在 CR 网络中复杂环境下信号的高效重构能力。首先，算法强调了节点间的本地采样，每个节点根据自身接收到的信号独立建立一个共同的稀疏信号模型。这种模型考虑了信号之间的相关性，旨在捕捉到信号的本质特征，即使在高相关性情况下也能保持良好的压缩效果。在实际操作中，C-BCSS 算法将每个节点的局部信息融合到一个中央融合中心，通过协作方式进行信号的重建。在这个过程中，融合中心利用所有节点提供的数据，通过概率模型进行综合分析和处理，从而提高了整体的准确性和鲁棒性。与传统的多任务贝叶斯压缩感知（Multi-task BCS）方法相比较，C-BCSS 在处理严重欠采样条件时表现出更好的性能，尤其是在资源有限、频谱效率要求高的环境中。作者 Rikang Zhou、Yuyan Zhang、Xuekang Sun 和 Caili Guo 来自北京邮电大学通信教育学院，他们通过实验验证了 C-BCSS 算法的有效性，并且展示了在实际应用场景中，如无线通信、频谱共享等，该算法能够显著提升频谱利用率，降低误报和漏报率，这对于构建高效、可靠的 CR 网络具有重要意义。总结来说，C-BCSS 是一种结合了压缩感知和协作学习的智能频谱感知策略，它通过优化信号处理流程和利用网络中节点间的协同作用，为 CR 网络提供了一种有效的解决方案，有望推动未来无线通信技术的发展。

A Cooperative Spectrum Sensing Algorithm Based On Bayesian Compressed

Sensing

Rikang Zhou, Yuyan Zhang, Xuekang Sun, Caili Guo

School of Telecommunication Education

Beijing University of Posts and Telecommunications,

Beijing, 100876

e-mail: zhourikang@gmail.com

Abstract—Compressed sensing (CS) has been applicably used

in cognitive radio (CR) to sense wide-band spectrum, solving

the wide-band sensing rate restriction problem. For a CR

network which includes multiple nodes, it's considered that the

received signals is high correlated. So in this paper, we focus

on efficient compressed spectrum sensing reconstructions in

CR network, proposing a probabilistic model called

Cooperative Bayesian Compressed Spectrum Sensing (C-BCSS)

algorithm. In the algorithm, node samples locally and a

common sparse signal model is developed to the received

signals where correlation exists. And then C-BCSS

reconstruction is performed cooperatively at fusion center

utilizing the information from all nodes. We compared C-

BCSS with Multitask BCS and simulation results show that C-

BCSS outperforms in condition of severe under-sampling and

heavy noise.

Keywords-spectrum sensing, cognitive radio, Bayesian

compressed sensing, distributed network

I. INTRODUCTION

Cognitive radio [1] technology as a novel approach has

been proposed to improve the status of spectrum under-

utilization. Indeed, for the current static spectrum allocation

policy, some frequency bands have been utilized in an

inefficient way. Investigations show that some frequency

bands are fully occupied, while some others turn out to be

idle at most of the time [2]. The idle spectrum which has

been called spectrum hole generally is a band of frequencies

which is assigned to a licensed primary user (PU), but not

occupied at a particular time and specific geographic location.

CR network allows secondary users (SUs), who actually are

not authorized to utilize the frequency band,

opportunistically to access spectrum to use spectrum source

with principle of no harm to PUs. This sharing strategy

allows SUs to exist heterogeneously with PUs to utilize

spectrum efficiently, and nowadays it attracts great interests

of applying it in fifth generation of cellular wireless (5G)

technology [3].

Primarily, before being free to access spectrum holes for

SUs, spectrum sensing should be performed intelligently and

instantaneously to detect whether the spectrum is occupied

by PUs. However spectrum sensing is a challenge in the case

of sensing a wide-band spectrum. Conventional methods are

inefficient and impractical in terms of sensing rate and

accuracy. A general way like using analog-to-digital

converter (ADC) will suffer a severe hardware restriction

because of high sampling rate. To this problem, CS [4] is

adopted as an effective solution utilizing its merit that a

small collection of measurements of a sparse signal contains

enough information to reconstruct the underlying signal

perfectly, which means wide-band spectrum can be sensed at

a low rate. Some wide-band spectrum sensing schemes based

on CS methodology are introduced in a non-cooperation

scenario [5], using orthogonal matching pursuit (OMP)

algorithm and basis pursuit (BP) algorithm. Bayesian

compressed sensing (BCS) [6] is another efficient

reconstruction algorithm derived from the technology in

Relevance Vector Machine (RVM) [7]. In BCS model, prior

information is adopted to the signal in the Bayesian

perspective. It turns out that BCS has sparser solution and

outperformance of robustness to noise.

In fact, considering in a network, all SUs receive signals

and it is rich of interrelated information in space and time or

redundant structures about the received signals. The signals

can be reconstructed jointly and cooperatively. In the

literature of CS, this has been previously studied in the field

of distributed compressed sensing (DCS) [8] and

simultaneous sparse approximation [9]. Joint sparsity model

(JSM) has been proposed to model multi-signal ensembles

and applied in joint spares recovery algorithms. In paper [10],

Multitask BCS is proposed and shows a good performance,

where all node use a shared prior to help every node's

construction.

In this paper, to utilize the interrelated information and

promote the performance further, we propose an algorithm

called Cooperative Bayesian Compressed Spectrum Sensing

to offer an efficient cooperative reconstruction method over

the network. In the algorithm, we develop a common sparse

signal model to the underlying spectrum signals. And then a

joint recovery proceeds via the regression of joint BCS we

obtained. Rather than a shared prior, a shared sparse signal

model leads to promotions of C-BCSS compared with

Multitask BCS.

The reminder of this paper is organized as follow. In

section II, we describe the signal model and basic principle

briefly and lead to the problem statement. In section III, we

go into details about C-BCSS. Then we talk about simulation

results in section IV. Conclusion is drawn in section V.

II. SYSTEM MODEL AND PROBLEM FORMULATION

Assume that a wide-band spectrum is shared by SUs and

PUs in a CR network, and the spectrum is sensed by L SUs.

We adopt a slotted frequency segmentation structure to the

wide-band spectrum and divide the spectrum into N non-

overlapping sub-channels equably. We assume that K (K <<

N) stochastic sub-channels is occupied in a SU's detection

658

2017 3rd IEEE International Conference on Computer and Communications

下载后可阅读完整内容，剩余4页未读，立即下载

weixin_38738005

粉丝: 5
资源: 895

基于贝叶斯压缩感知的协作频谱感知算法

"数据融合的协作频谱感知方法研究：认知无线电新技术探索"。

协作频谱感知新算法：贝叶斯压缩感知应用

低开销协作压缩频谱感知：1-bit量化新方案

A Feasibly Efficient Cooperative Spectrum Sensing Scheme Based on Cholesky Decomposition of The Correlation Matrix

An Energy-Efficient Cooperative Spectrum Sensing Scheme based on D-S Theory in Cognitive Radio Sensor Networks

A Novel CoMAC-based Cooperative Spectrum Sensing Scheme in Cognitive Radio Networks

A Robust Cooperative Spectrum Sensing Method in Cognitive Radio Networks

RBM Based Cooperative Bayesian Compressive Spectrum Sensing with Adaptive Threshold

Cooperative Data Processing Algorithm Based on Mobile Agent in Wireless Sensor Networks

A novel node selection scheme for energy-efficient cooperative spectrum sensing using D–S theory

最新资源