SPARSITYINDEPENDENT GREEDY COMPRESSIVE SENSING ALGORITHM FOR COGNITIVE RADIO

Abstract

With the developments in wireless communication and increased spectrum usage, the Cognitive newlineRadio has emerged as promising technology to ensure the efficient spectrum usage. The newlineDynamic Spectrum Access being one of the key functionalities of the Cognitive Radios, it is to newlinebe aided by the proficient Spectrum Sensing techniques. Compressive Sensing is an emerging newlinefield of Spectrum Sensing. newlineHaving spectral scarcity affecting the different fields of wireless communication, we need to newlineaddress the issues related to it. We need to manage the spectral resources in an efficient way, and newlinefor that, we require to implement effective spectrum sharing and flexible spectrum usage. The newlinetask demands from us an enhanced ability to sense the spectrum to achieve our goals. And this newlinehas motivated us to study and work in the area of spectrum sensing. newlineWe have presented the basics of Cognitive Radio, Software Defined Radio, Spectrum Sensing newlineand Compressive Sensing. Building the base of our work with these theoretical aspects we newlinepresent here a work based on Compressive Sensing algorithm study. The work is divided into newlinetwo phases: first, the two widely used CS algorithms are studied and compared, OMP and KLT, newlinewhich helped us to choose the OMP as the framework for our proposed scheme; and then, in the newlinesecond stage, we have proposed a greedy iterative algorithm, based on OMP: Sparsity- newlineIndependent-OMP (SI-OMP), for random frequency-sparse spectral conditions. newlineThe functionalities of the basic OMP, and KLT were realized through simulations in MATLAB newlineover a range of 0 MHz-60 MHz in our first phase of the work. During the research work, we newlinefound that the despite of the KLT being widely used and popular unitary transform for data newlinecompression, it is data-dependent and with dynamically varying sparse spectrum its performance newlinein terms of reconstruction ability deteriorates. The basic OMP on the other hand exhibited newlineconstant performance with different input sparsity