Abstract | This paper presents a method that deploys nonuniform sampling and appropriates to it a processing algorithm to monitor the activity of a number of non-overlapping spectral bands. The proposed approach facilitates the use of sampling rates well below the ones demanded by uniform-sampling-based DSP. Randomized sampling scheme, namely random sampling on grid, in conjunction with a periodogram-type spectral analysis tool is utilized to accomplish the task. The statistical characteristics of the endorsed analysis tool are examined for a finite set of nonuniformly distributed signal samples contaminated with noise. General guidelines are provided to ensure the reliability of the adopted sensing technique where it is affirmed that the sampling rates can be arbitrarily low. The additional requirements on such recommendations imposed by the presence of noise are given. It is demonstrated that in certain scenarios the proposed technique can considerably reduce the complexity of the spectrum sensing procedure. The presented analytical results are illustrated by numerical examples. This paper establishes a new framework for efficient spectrum sensing methods that exploit randomized sampling schemes. Unlike a number of similar approaches in the literature, it offers solutions that are well suited for practical implementation in hardware. |
---|