In this thesis, novel schemes for transceiver design in orthogonal frequency-division multiplexing (OFDM) systems to support heterogeneous traffic are proposed by jointly considering system properties at both the physical and the link layers. Especially, a robust slow fading channel estimation algorithm based on H∞ approach is proposed to provide accurate and low complexity channel fading estimates without the requirements of a priori knowledge of channel fading and disturbance statistics. An Expectation and Maximization (EM) channel estimation algorithm is proposed to provide estimates for fast fading channels by considering the existence of intercarrier interference (ICI) and the time-selectivity of the channel fading. Based on the estimates of channel impulse response, multicarrier detection (MCD) schemes are proposed to effectively eliminate ICI based on the equality between the OFDM system and the synchronous CDMA (SCDMA) model and a modified successive interference cancellation (MSIC) scheme is introduced to improve the detection delay performance by taking into account the strong correlation between neighboring subcarriers. At the link layer, an optimal resource management scheme with adaptive modulation, subcarrier allocation, bit loading, power allocation, and scheduling is proposed to achieve maximum system throughput under the constraints on total transmission power and the heterogeneous QoS requirements, and providing fairness to all the users admitted into the system. A truncated generalized processor sharing (TGPS) and a dynamic GPS (DGPS) scheduling schemes are also proposed to achieve the practical implementation of the ideal GPS scheduling and improve the resource utilization by considering the delay tolerance of non-real-time traffic. Simulation results demonstrate that the proposed transceiver designing schemes can achieve both effective quality of service (QoS) guarantee and efficient resource utilization.