Fairness Analysis of Wireless Beamforming Schedulers

This dissertation is devoted to the analysis of fairness at the physical layer in multi-antenna multi-user communications, which implies a new view on traditional techniques. However, the degree of equality/inequality of any resource distribution has been extensively studied in other fields such as Economics or Social Sciences. Indeed, engineers usually aim at optimizing the total performance, but when multiple users come into play, the overall optimization might not necessarily be the best thing to do. For instance in wireless systems, the user with a bad channel condition might suffer the consequences from the selective choice based on the instantaneous channel quality made by a centralized entity. In this sense, the problem has four different perspectives: antenna processing, power allocation, bit allocation, and combination of space diversity (SDMA) with multiple subcarriers (OFDM). The technical contribution of the author starts with the analysis of fairness conducted not only for transmit processing, but also for the upper bound that represents the cooperative strategy between the transmitter and the receiver. After that, power and bit allocation techniques are compared, and it is shown that the optimizing criterion shall be carefully chosen to avoid undesirable operating consequences. New algorithms are proposed to balance the traditional techniques on the extremes of the fairness axis, i.e. global performance and the individual needs. Finally, new approaches are given to realistic environments, so that practical algorithms can be easily deployed in future communication standards. The final conclusion is that the choices in a multi-antenna multi-user wireless systems are not straightforward, since there exist several trade-offs, among others: performance vs. complexity, global performance vs. individual needs, and performance vs. signalling. The ultimate behaviour of the system deeply depends on the design made by the manufacturers.