The main objective of this thesis is to utilize antennas in novel ways so as to achieve performance benefits at the system level through spatial interference management. To this end, we study CDMA multi-antenna wireless access networks (where the transmission and reception are through multiple antennas distributed in the service area), and compare their performance with that of the conventional networks where only one central antenna (CA) exists.

We first demonstrate that by using CDMA distributed antenna (DA), the power control dynamic range can be reduced significantly;​ this yields a notable reduction in the outage.

The DA system simulates a hypothetical optimal CA system;​ therefore, even though the capacity of a DA system may be considerably higher than that of a CA type, the capacity is still low per antenna element (AE). In order to overcome this shortcoming, we present the sectorized distributed antenna (SDA) system, where each AE is connected to a separate feeder. We further suggest a novel nonlinear power control algorithm which balances the SIR in SDA systems.

We demonstrate analytically and through simulations that in a CDMA SDA system the reverse link capacity increases approximately linearly with increasing numbers of AE's. This increase is still valid, despite the overlapping antenna patterns and non-uniform traffic, due to the joint decoding (enhanced macrodiversity) capability of the SDA system, and also to the effective power control algorithm used. Therefore, the reverse link capacity of an SDA system with L AE's is higher than that of an L-cell duster serving the same region.

In order to achieve maximum performance from the SDA architecture, the interference picked up by an AE for a user should be uncorrelated with those picked up by other AE's. To this end, the conditions, under which the correlated interference occurs, are evaluated, and the effects of the system parameters on the correlation are analyzed.

Finally, AE interconnection strategies are studied in order to determine cost-efficient as well as robust and flexible interconnection architectures, by using results from the theory of minimal networks, especially those on Steiner trees.