In wireless environments, signals bounce off many obstacles such as mountains, buildings, trees, etc. as they propagate between transmitters and receivers. The resultant signal at the receive antenna is, therefore, often the sum of the attenuated transmitted signal and one or more delayed versions of the transmitted signal. The received signal also suffers from intersymbol interference which degrades the quality of signal to a certain extent.
However, MIMO-OFDM systems are designed to take advantage of the multi-path properties in wireless communications and are capable of improving transmission rate, range and reliability simultaneously. MIMO-OFDM attracts a good deal of research and commercial interest because of the perceived benefits, and has been adopted in many wireless standards such as IEEE 802.1 In, IEEE 802.16e. Such systems are also potential candidates for fourth-generation (4G) systems. However, practical problems still exist in implementing MIMO-OFDM, for example, in the estimation of channel state information (CS1). This thesis studies the issues of MIMO, OFDM and the relevant techniques of MIMO-OFDM, and focuses on proposing a practical, low complexity and accurate channel estimation method for such systems.
In a MIMO-OFDM system, CSI is required at the receiver to perform space-time decoding or diversity combining. In many practical wireless applications, the propagation environment is both complex and time-variant, leading to CSI estimation errors and overall system performance degradation. A variety of channel estimation approaches have been proposed in the literature to address this problem. One of the most important parameters of CSI is the number of significant or dominant propagation paths, also referred to as the number of channel taps. However, in most existing estimation schemes for MIMO-OFDM, there is an assumption that the number of channel taps is known at the receiver. In reality, in order to perform space-time decoding, the receiver needs to estimate the number of channel taps from the received signal with this estimation process sometimes aided by the insertion of pilot tones into the transmitted signal.
In this thesis, a pilot-assisted, conditional model-order estimation (CME) based channel estimation algorithm is presented. The approach can be utilised to detect both the number of channel resolvable paths and channel gains for MIMO-OFDM systems. The performance of the proposed algorithm is compared with the commonly used minimum description length (MDL) algorithm by mean of simulation in the context of a 2x2 MIMO-OFDM system. Results indicate that the new algorithm is superior to the MDL algorithm in channel order estimation over an unknown, noisy, multipah fading channel with limited pilot assistance. Furthermore, the proposed scheme is tested in both fixed and mobile broadband MIMO-OFDM systems based on WiMAX techniques in Matlab simulation, and its capacity is verified again for those near practical broadband MIMO- OFDM systems in the absence of prior knowledge of model parameters.
Finally, with the purpose to “make the thing work in practice”, a 2x2 MIMO baseband platform is built in order to demonstrate the proposed scheme. The platform consists of two DSP based, real-time development boards called SignalWAVe, produced by Lyrtech. Given the existing hardware components, the whole platform is built based on a fixed MIMO-OFDM system according to WiMAX standard, and the results demonstrate that the proposed algorithm is a valid approach in practice.