A Digital Signal Processing Solution for Multichannel Base Stations

Radio communication systems of the future will require a large increase in user capacity. To achieve this, cell sizes will reduce and the number of base stations will increase. Current base station architectures use analogue combining techniques which are expensive, voluminous and inflexible. This thesis investigates a Digital Signal Processing (DSP) solution which produces a cheaper, smaller and more flexible multichannel base station transmitter design. The main design challenges of the new DSP low power combining architecture are the multichannel combining algorithm, the frequency translation of the multichannel signal to radio frequency, the Digital to Analogue Conversion (DAC) interface and the wideband ultra-linear power amplifier. This thesis considers the pre-power amplifier stages. Combining the channels in a digital signal processing environment provides considerable flexibility, but the computational requirements are very high. The minimisation of computational load is achieved by combining the channels at baseband with efficient algorithms. Four upconversion techniques, currentiy used in single channel applications, are investigated for use in a wideband multichannel environment. Comparisons reveal that analogue direct upconversion currently provides the most attractive solution because a lower performance DAC is required and less computation is needed. However, amplitude and phase mismatch between the In Phase and Quadrature circuits cause undesired sideband responses that exceed radio system specifications. A novel adaptive compensation method leads to an improved performance and a lower computational overhead compared to previous techniques. Multichannel radio systems can only tolerate very low harmonic and intermodulation products and the DAC is a major source of these spurious responses. A new implementation of bandlimited dithering is shown to improve the performance of the DAC by 2 to 4 dB, with only a small increase in hardware. Preface This thesis describes the development of a low power combining digital signal processing architecture for future radio base station transmitters. The goal of the project is to produce a base station that is small, cheap and greatly more flexible than the conventional base station design. The idea for the low power combining architecture was described in the 1980's and its advantages over the conventional design have since been well recognised. However, four technological challenges have impeded the immediate development of the digital signal processing low power combining technique: • the large amount of computation required to combine multiple channels in a digital signal processor, • the high degree of linearity needed to convert the multichannel signal from baseband to RF, • the high speed and precision of the DAC required to convert the digital multichannel signal to an analogue signal, and • the ultra-linear power amplifier needed to boost the power level of the multichannel signal for radio frequency transmission. The ultra-linear power amplifier is seen as the most difficult design obstacle to overcome and most researchers and manufacturers of radio base stations have, and still are, focusing much research and deyelopment on this area. Through the use of advanced linearisation techniques, feedforward, feedback and predistortion, a sufficiently linear and wideband power amplifier has been developed. These developments haye only occurred since the thesis began in 1991 although the power amplifier is still limited in total power output. The objective of the thesis has been to concentrate on the design challenges of the prepower amplifier stages. A detailed examination of the combining of multiple channels and the upconversion of the multichannel signal to RF provide clarification of the pertinent issues regarding a digital signal processing base station. These discussions lead to the choice of the analogue direct upconversion technique. A new compensation algorithm required to ensure that this upconversion technique meets the radio system specifications