Today the most recent standard for mobile communications is known as LTE-Advanced (LTE-A), developed and organized by the 3rd Generation Partnership Project. As a 4th generation system for mobile communication, LTE-A offers high exibility in terms of deployment aspects and resource allocation as well as high spectral efficiency. To achieve these main features, sophisticated link level access schemes, such as orthogonal frequency division multiplexing and Multiple-Input Multiple-Output (MIMO) technologies, are employed. For such a wireless communication system, coherent detection at the receive side is essential to obtain intended overall performance. This requires knowledge of the wireless channel, making channel estimation not only necessary but a very important task within the signal processing chain. In LTE-A reference symbols, which are known to the receiver, are multiplexed with the transmitted data to enable channel estimation. In uplink MIMO transmissions, these pilot symbols are multiplexed on the same time-frequency positions for all spatial layers, making separation of estimated MIMO channels necessary. Many existing estimation methods aim for separation in time domain, exploiting the reference symbols' code-domain orthogonality. These algorithms suffer from overlapping Channel Impulse Responses (CIR) when estimating frequency selective channels, referred to as CIR leakage. In my work I present channel estimation algorithms that aim to separate all estimated MIMO channels in frequency domain. These methods do not suffer from CIR leakage and achieve a significantly lower estimation error when estimating frequency selective channels. Further, I introduce and compare several interpolation methods that enable to obtain channel coefficients in between pilot positions, making estimation of doubly selective MIMO channels possible. I compare and discuss performance of presented estimation methods in terms of mean square error and resulting bit error ratio as obtained by simulations, in the context of single-user and multi-user MIMO.