Communications and radar signal processing for multiple base stations
This dissertation primarily focuses on design techniques for systems where multiple towers cover a given geographical area. In the first portion of this dissertation, these are communication towers and the goal is the reliable broadcast of information from these towers to mobile users. In the second portion of this dissertation, these towers are radars, and the goal is to recover the refractivity of the intervening atmosphere using phase measurements of the returned signal from various ground targets. Finally, in the third section, we depart from this general scenario and study transmission schemes on the wireless exchange channel.
In macroscopic space-time coding, the codewords are spread across antennas at wide geographical separations instead of a single antenna array, as in standard space-time coding. Since the mobile users to whom the signals are transmitted can be spread across a wide geographical region, the times of arrival of the signal from different antennas are not identical across users. In contrast to standard space-time coding, codes under such a scenario must be robust to such relative differences. This demands more rigorous code designs. The performance criterion, the optimal receiver, and a family of code designs are discussed. In particular, a constructive code design for block codes and two trellis codes are presented. A matched filter bound analysis with associated numerical and simulation results demonstrates the robustness of the proposed code designs to the relative arrival times of the signals from different transmit antennas. Then, the matched filter bound analysis is extended to demonstrate significant improvements in coverage for such a system over currently employed systems and standard space-time coding approaches applied across the same set of transmit antennas.
Retrieval of surface-layer refractivity via the method of Fabry is considered. A mathematical framework is constructed and signal processing algorithms derived that facilitate refractivity retrieval from the returns from multiple radars viewing a common geographical area. In particular, an approximate discrete model is derived to relate the measured phases to the surface layer refractivity fields, and a modified least-squares (LS) estimation algorithm is then proposed for the resulting, often ill-conditioned, inversion problem. Because the measurement technique is subject to modulo 2π uncertainties which impact retrievals, a novel algorithm, which jointly estimates the unwrapped phases and refractive index (RI) field, is also provided. Numerical results indicate the effectiveness of the derived algorithms in both the single and multiple radar case, as well as clearly establishing that having multiple views of the same geographical area from separate radars provides for significant improvement of the RI field estimates.
Finally, the achievable rates of various transmission schemes for the exchange channel are studied. Network coding, where relay nodes combine the information received from multiple links rather than simply replicating and forwarding the received packets, has shown the promise of significantly improving system performance. In very recent works, multiple researchers have presented methods for increasing system throughput by employing network coding inspired methods to mix packets at the physical layer: physical-layer network coding (PNC). A common example used to validate much of this work is that of two sources exchanging information through a single intervening relay - a situation that we denote the "exchange channel''. In this dissertation, achievable rates of various schemes on the exchange channel are considered. Achievable rates for traditional multi-hop routing approaches, network coding approaches, and various PNC approaches are considered. A new method of PNC inspired by Tomlinson-Harashima precoding (THP), where a modulo peration is used to control the power at the relay, is introduced, and shown to have a slight advantage over analogous schemes at high SNRs.