Sampling with internal filtering in digital radios.

摘要

All contemporary communication receivers are digital. However, they still contain large analog and mixed-signal front-ends. These front-ends limit the dynamic range, attainable bandwidth, flexibility, and scale of integration of the receivers. The major causes of these limitations are conventional sampling and antialiasing filtering, which are based on the traditional interpretation of the sampling theorem.;An alternative interpretation of this theorem developed with the author's participation enables the design of new sampling circuits with flexible internal antialiasing filtering that accumulates signal energy during several sampling intervals for each sample, whereas conventional sampling utilizes only small fraction of this energy. The properties of the new sampling circuits are determined by internally generated weight functions that can be dynamically changed. Implementation of these sampling circuits allows the development of a high dynamic range wideband receiver on a chip, which performs digitization close to the antenna and has high adaptivity and reconfigurability.;The purpose of this dissertation is a detailed theoretical analysis of the properties of the new sampling circuits with internal antialiasing filtering, as well as the development of approaches to their practical implementation in digital radios. To this end, the following three major problems are addressed here.;The first and most important problem is the optimization of the weight functions used in the sampling circuits. Two approaches to this optimization are proposed: (1) the formal approach based on a certain optimality criterion like least-squares or Chebyshev, and (2) heuristic approach that leads to easily realizable weight functions with suboptimal filtering properties. Examples of weight functions resulting from both approaches are analyzed and compared. As a part of this study, the theoretical constraints that must be imposed on the weight functions are determined. Besides that, a promising class of suboptimal weight functions based on B-splines is identified and examined.;The second problem is the optimization of the sampling circuit architecture for simplicity of practical implementation. Several simplification methods are proposed and investigated. All these methods reduce (in some cases, to zero) the number of analog multipliers in the sampling circuits. This is important because the multipliers can potentially cause higher nonlinear distortions than other parts of the sampling circuits (although these distortions are still lower than those caused by conventional sampling). In particular, it is shown that sampling circuits with B-spline-based weight functions can be implemented without analog multipliers.;The third problem is the analysis of the influence of non-ideal hardware implementation on the performance of the sampling circuits, as well as the development of the ways to minimize or eliminate this influence. The attention here is focused on the analysis of the influence of non-ideal integration and on the methods of channel mismatch mitigation in the sampling circuits. It is shown that it is relatively easy to achieve sufficient quality of integration, and effective methods of the channel mismatch mitigation are proposed and investigated.;The methodology of comparative simulations of the new and conventional sampling circuits, as well as corresponding Matlab models, have been developed. The results of these simulations presented in the dissertation show the advantages of the new sampling with internal filtering over the conventional one. They also demonstrate that the influence of non-ideal hardware implementation on the performance of the new sampling circuits can be successfully eliminated.;The work performed in the course of this research has created a theoretical basis for practical implementation of sampling with internal antialiasing filtering.