Array processing techniques for interference suppression in mobile communications systems

Array processing techniques for interference suppression in mobile communication systems are investigated. First, a multiuser detection scheme for code-division multiple access communication systems is presented. The algorithm, referred to as a partitioned linear interference canceller (PLIC) is based on constrained optimization and bears resemblance to the so-called anchored multiuser detector. However, by deriving the PLIC within a constrained optimization framework, additional benefits beyond those of the anchored detector are realized. For example, multiple constraints, when they can be specified, are shown to be an effective means of incorporating knowledge of additional users' codes into the detector, thus improving performance.; While the PLIC is not an adaptive array, well known array processing techniques are used to develop the algorithm. Specifically, the detector implementation is based on an array processing structure known as the generalized sidelobe canceller (GSC). This structure yields advantages over conventional projection gradient implementations by providing a degree of robustness to finite precision effects. Moreover, the GSC-based structure lends itself to partially adaptive implementations, which, because of their potential for significantly reducing computational complexity, are also investigated in this thesis.; A multisensor implementation of the detector is also proposed. Relying on diversity reception with maximal ratio combining, the multisensor PLIC is suitable for interference suppression in multipath fading environments. Both predetection and postdetection combining are investigated. The PLIC and its various proposed extensions are all tested via simulation in a number of realistic interference scenarios.; Finally, a maximum-likelihood co-channel separation technique for constant modulus signals is presented. This technique addresses directly the ability of an adaptive array to increase the spectral efficiency of mobile communication systems by decreasing the reuse distance between co-channel cells. Given the constraint that the signals have a constant modulus, the algorithm is capable of estimating parameters for approximately twice the number of signals compared to conventional techniques. The algorithm is tested via simulation for the additive white Gaussian noise and multipath fading channels.