24-06-2013, 04:27 PM
Smart Antennas for Wireless Communications
Smart Antennas.pdf (Size: 184.94 KB / Downloads: 39)
Introduction
Over the last decade, there has been a growing interest on improving system
capacity and increasing coverage area in wireless communications using smart
antennas. Smart antenna technology is being considered for mobile platforms
such as automobiles, cellular phones (mobile unit), and laptops. This paper
presents the design of a smart antenna along with the associated signal processing
algorithms for use in the next generation 20 GHz wireless communications
systems. The results presented here are part of a broader project that considers
this antenna system in the context of reconfigurable broadband (high-speed)
networks. Although the design of such a system entails several other tasks, such
as the design of network protocols, feed network, and physical layer
communications algorithms, this paper concentrates only on the antenna design
and on the development of efficient adaptive algorithms for beamforming and
direction-of-arrival (DOA). The objective is to design an adaptive antenna that
directs the maximum radiation of the antenna pattern toward the signal-of-interest
(SOI), and places nulls toward the signal-not-of-interest (SNOI). The proposed
antenna design and associated beamforming algorithms take into account mutual
coupling for greater accuracy.
Smart Antenna System Design
The design is subdivided into three major areas: antenna design, adaptive
beamforming algorithm, and adaptive DOA estimation. In this paper, we
concentrate mostly on the antenna design and the beamforming algorithm.
Antenna Design
Because microstrip patch antennas are inexpensive, lightweight, conformal, easy
to manufacture and versatile, they are the most suitable type for a portable device.
The substrate material used in this design is Silicon with a dielectric constant of
11.7, a loss tangent of 0.04, and a thickness of 550 mic. The operating frequency,
fr, is 20 GHz. The square patch dimensions, L, and the probe location, yo, depend
on the input impedance (e.g., 50 SL) of the patch, and they are computed
iteratively using the estimate function of Ensemble@ [I] together with (1).
Results and Discussions
Fig. 2 shows the detrimental effects if the weights of the LMS are not adjusted to
take into account mutual coupling. In fact, the thin solid line in Fig. 2 shows that
the nulls are not only not as deep but also slightly shifted with respect to the
theoretical pattern (the pattern computed directly from the weights of the LMS,
dotted line). On the other hand, the adjusted weights (thick solid line) follow the
theoretical pattern more accurately.