Interference alignment (IA) is a promising joint-transmission technology that essentially enables the maximum achievable degrees-of-freedom (DoF) in K-user interference channels. Fundamentally, wireless networks are interference-limited since the spectral efficiency of each user in the network is degraded with the increase of users. IA breaks through this barrier, that is caused by the traditional interference management techniques, and promises large gains in spectral efficiency and DoF, notably in interference limited environments. This dissertation concentrates on overcoming the challenges as well as exploiting the opportunities of IA in K-user multiple-input multiple-output (MIMO) interference channels. In particular, we consider IA in K-user MIMO interference channels in three novel aspects. In the first aspect, we develop a new IA solution by designing transmit precoding and interference suppression matrices through a novel iterative algorithm based on Min-Maxing strategy. Min-Maxing IA optimization problem is formulated such that each receiver maximizes the power of the desired signal, whereas it preserves the minimum leakage interference as a constraint. This optimization problem is solved by relaxing it into a standard semidefinite programming form, and additionally its convergence is proved. Furthermore, we propose a simplified Min-Maxing IA algorithm for rank-deficient interference channels to achieve the targeted performance with less complexity. Our numerical results show that Min-Maxing IA algorithm proffers significant sum-rate improvement in K-user MIMO interference channels compared to the existing algorithms in the literature at high signal-to-noise ratio (SNR) regime. Moreover, the simplified algorithm matches the optimal performance in the systems of rank-deficient channels. In the second aspect, we deal with the practical challenges of IA under realistic channels, where IA is highly affected by the spatial correlation. Data sum-rate and symbol error-rate of IA are dramatically degraded in real-world scenarios since the correlation between channels decreases the SNR of the received signal after alignment. For this reason, an acceptable sum-rate of IA in MIMO orthogonal frequency-division-multiplexing (MIMO-OFDM) interference channels was obtained in the literature by modifying the locations of network nodes and the separation between the antennas within each node in order to minimize the correlation between channels. In this regard, we apply transmit antenna selection to MIMO-OFDM IA systems either through bulk or per-subcarrier selection aiming at improving the sum-rate and/or error-rate performance under real-world channel circumstances while keeping the minimum spatial antenna separation of half-wavelengths. A constrained per-subcarrier antenna selection is performed to avoid subcarrier imbalance across the antennas of each user that is caused by per-subcarrier selection. Furthermore, we propose a sub-optimal antenna selection algorithm to reduce the computational complexity of the exhaustive search. An experimental testbed of MIMO-OFDM IA with antenna selection in indoor wireless network scenarios is implemented to collect measured channels. The performance of antenna selection in MIMO IA systems is evaluated using measured and deterministic channels, where antenna selection achieves considerable improvements in sum-rate and error-rate under real-world channels. Third aspect of this work is exploiting the opportunity of IA in resource management problem in OFDM based MIMO cognitive radio systems that coexist with primary systems. We propose to perform IA based resource allocation to improve the spectral efficiency of cognitive systems without affecting the quality of service (QoS) of the primary system. IA plays a vital role in the proposed algorithm enabling the secondary users (SUs) to cooperate and share the available spectrum aiming at increasing the DoF of the cognitive system. Nevertheless, the number of SUs that can share a given subcarrier is restricted to the IA feasibility conditions, where this limitation is considered in problem formulation. As the optimal solution for resource allocation problem is mixed-integer, we propose a two-phases efficient sub-optimal algorithm to handle this problem. In the first phase, frequency-clustering with throughput fairness consideration among SUs is performed to tackle the IA feasibility conditions, where each subcarrier is assigned to a feasible number of SUs. In the second phase, the power is allocated among subcarriers and SUs without violating the interference constraint to the primary system. Simulation results show that IA with frequency-clustering achieves a significant sum-rate increase compared to cognitive radio systems with orthogonal multiple access transmission techniques. The considered aspects with the corresponding achievements bring IA to have a powerful role in the future wireless communication systems. The contributions lead to significant improvements in the spectral efficiency of IA based wireless systems and the reliability of IA under real-world channels.