Abstract
Spectrum sensing is one of the important functions in the context of cognitive radio systems. It determines the presence or absence of free channels in the spectrum and makes them available for the secondary users. Cyclostationary spectrum sensing is one of the spectrum sensing techniques which involves the detection of signals based on their features such as cyclic frequencies, symbol rates, carrier frequencies and modulation types. It detects signals at very low signal-to-noise ratios. Cyclostationary spectrum sensing involves the use of large number of samples for detection resulting in high complexity, cost and low efficiency. In addition there are performance degrading constraints such as cyclic and sampling clock offsets that can occur at the receiver end. These offsets are caused by local oscillator frequency offsets, Doppler effects and jitter.In order to address some of these issues in the absence of the constraints, the thesis proposes an efficient low complexity multi-slot cyclostationary spectrum sensing technique that uses small number of samples to detect small spectral components made possible by the use of fast Fourier transform and slots of small lengths. Statistical and simulation tests are performed to verify the functionality of the model to offer low complexity and consequently low cost and efficiency.
The thesis also proposes another multi-slot cyclostationary spectrum sensing model that included the receiver constraints such as cyclic frequency offset and sampling clock offset in the test statistic. This model is analysed statistically and uses small lengths of fast Fourier transform and slots to effect significant reduction of these constraints which is also verified with Matlab simulation i ii results.
In order to have a non ad hoc systematic way of detecting and optimizing the sizes of fast Fourier transform and slots, the thesis also proposes step by step algorithms that can be applied to any set of total number of samples representing a wideband channel. This will result in getting the appropriate sizes of slots and fast Fourier transform that will produce low complexity, cost and efficient detection. Matlab simulations are also used to verify this.
Finally, the proposed models are able to address the issues previously mentioned which are associated with the cyclostationary spectrum sensing.
Date of Award | Oct 2018 |
---|---|
Original language | English |
Awarding Institution |
|
Supervisor | John Chiverton (Supervisor), Misha Filip (Supervisor) & Abdulkarim Tawfik (Supervisor) |