Bandpass Sampling Methodology for Uniformly Spaced Multiband Energy-Sparse Spectrum
In this dissertation, a novel methodology for direct down-conversion of radio frequency (RF) is proposed, which in return is useful to minimize sampling frequency for an evenly spaced spectrum comprising multiband RF signals. The proposed methodology describes a set of rules to achieve the lowest possible sampling frequency without any compromise of spectrum folding or overlap of aliases in baseband after down conversion. It is shown that the minimum sampling rate has a unique relation with the layout of the spectrum of interest (SOI). For instance, if there are N number of information bands of equal bandwidth B in the SOI, then it is possible to down-convert the complete SOI using the sampling rate, 2NB, which is twice of the total information bandwidth only if all bands are evenly spaced in the SOI. Another factor introduced to the achieve the minimum sampling rate is the sparseness-nature in the SOI, which is the ratio of null bandwidth to information bandwidth. The proposed methodology is general in nature and is flexible to the number of input signals or bands as well as to their positions in the desired spectrum. In the proposed research work, simulations are carried out that verify that by using the recommended minimum sampling rates, the desired signal is extractable without any additional computational complexity due to spectrum folding or aliasing-overlap. Our proposed methodology has a vast scope in the design of general-purpose receivers, global navigational satellite system (GNSS) receiver and cognitive radios (CR) because of the use of a low speed ADC. Moreover, the same can be efficiently used to monitor a wide band spectrum in military communications especially for electronic warfare receivers, where reduction of the complexity, size and cost has significant importance.
As a model application of our proposed work, we present a composite design for the multiband-multistandard GNSS receiver. The design efficacy is based on the proposed bandpass sampling methodology that transforms the sparse-spectrum electromagnetic environment into a quasi-uniformly spaced spectrum of compact bandwidth, which is more appropriate and useful for simultaneous digitization and down-conversion of analogue signals. In this method, only a part of SOI is transformed to an intermediate frequency. In this way, the desired frequency bands of information that are widely spread, are grouped to form a contiguous-spectrum which is quasi-uniformly spaced. There on, a sub- sampling is carried out for simultaneous digitization and translation of input signals to the first-Nyquist zone. The proposed composite architecture is also helpful to circumvent the higher-order intermodulation components. The proposed design is validated for conventional Global Positioning System L1 and L2 bands and also for new L5 band used in GNSS (GLONASS, Galileo and Beidou) receivers. The presented results show considerable reduction in the sampling rates, and improvement in signal-to-noise and distortion ratio, which can be easily managed by a low sampling analogue to digital conversion.