Wireless communication systems require high spectral efficiency and throughput in order to be cost-effective. Link adaptation schemes are known to be a good solution to achieve this goal. However, the necessity of additional information or increased complexity prevents these schemes from being implemented. In this context, research on resource allocation based on different constraints, such as complexity or feedback, is important. The major contribution of this dissertation is the development of three novel techniques to enhance performance in practical implementations of the adaptive OFDM systems. This dissertation first introduces a new multiuser OFDM system to enhance performance in the low SNR regime. In this scheme, multiuser diversity can be efficiently amplified from random power allocation and opportunistic scheduling. Higher spectral efficiency can be achieved without an increase of complexity or feedback amount compared to conventional multiuser OFDM systems using equal power allocation. This dissertation also presents a modified multi-mode power loading scheme. A modified multi-mode power loading scheme can circumvent the limit of current multi-mode power loading schemes by significantly reducing search amount from 2N - 1 to N, where N is the number of subcarriers. Finally, this dissertation has introduced adaptive OFDM systems using channel gain order information in limited feedback environments. Adaptive OFDM systems using the order mapping technique achieve comparable performance to conventional adaptive OFDM systems in terms of bit error rate and average spectral efficiency, while the amount of feedback is significantly reduced. Furthermore, by simply exploiting order mapping and interpolation, the analyzing technique circumvents the practical shortcomings of previous limited feedback techniques for OFDM systems. / text