Neural network based smart antennas for cellular and mobile communications systems

New wireless systems such as cellular, personal communication systems (PCS) and personal communication networks (PCN) must satisfy an increasing demand for coverage, capacity, and service quality. For this purpose the need for more powerful tools to improve different aspects of modern communications systems has become increasingly important. In recent years, it has become clear that the area of smart antennas will provide a key technological boom for the wireless communications industry. This dissertation discusses the development of a neural network based smart antenna capable of performing detection, direction finding, adaptive beamforming and interference cancellation in various modern cellular communication systems. The developed algorithms are able to estimate mobile users' locations, track these mobiles as they move within or between cells, then allocate narrow beams in the directions of the desired users while simultaneously nulling unwanted sources of interference. This Space Division Multiple Access (SDMA) will improve the coverage as well as increase the system capacity of existing cellular and mobile communications systems.; Some of the attractive features of this novel neural network-based approach is that it can yield results in real-time, hence outperforming other conventional techniques. This leads to the accurate estimation of the mobile location and the computation of the weights of the adaptive array antennas in a few characteristic time constants of the circuit, normally, in the order of 100s of nanoseconds. Moreover, conventional beamformers require highly calibrated antennas with identical element properties. Performance degradation often occurs due to the fact that these algorithms poorly adapt to element failure or other sources of errors. Neural network-based array antennas do not suffer from this shortcoming. The antenna behavior (uniform, non-uniform spacing, or non-uniform elements, etc.) can be incorporated in the training of the neural network under different circumstances and scenarios. The network being able to generalize, can then be used to predict the aperture behavior at all points.; Other applications for smart antennas include multiple sources tracking for early warning systems, performance monitoring and failure correction of phased array antennas.