Wideband wireless transmission often suffers from performance degradation caused by the varying channel characteristic within the spectrum. The fading channel creates Inter Symbol Interference (ISI) and leads to symbols overlapping with each other. Also, a boost in transmission power is required when the channel experiences severe fading. This would shorten the battery life and create extra interference to others using the same channel....[ Read more ]

Wideband wireless transmission often suffers from performance degradation caused by the varying channel characteristic within the spectrum. The fading channel creates Inter Symbol Interference (ISI) and leads to symbols overlapping with each other. Also, a boost in transmission power is required when the channel experiences severe fading. This would shorten the battery life and create extra interference to others using the same channel.

A promising solution to these problems is Orthogonal Frequency Division Multiplexing (OFDM) with adaptive modulation. OFDM, by itself, could effectively alleviate ISI. The addition of adaptive modulation allows OFDM systems to adjust the transmit power and the amount of information in each sub-carrier according to the channel characteristic.

In this thesis, we consider a multiuser OFDM system with adaptive sub-carrier allocation and adaptive modulation. Assuming knowledge of the instantaneous channel gains for all users, we propose an optimal multiuser OFDM sub-carrier,bit, and power allocation algorithm to minimize the total transmit power. This is done by assigning, to each user, a set of sub-carriers, and by determining, for each sub-carrier, the number of bits and the transmit power level. We obtain the performance of our proposed algorithms in various multiuser frequency selective fading environments. The results show that our proposed algorithm outperforms other sub-carrier allocation schemes. To cope with the time varying channel, we also propose a suboptimal real-time sub-carrier allocation algorithm that requires much less computation time. Finally, we discuss the practical implementation issue of the algorithm.