JOINT CHANNEL AND SYMBOL ESTIMATION FOR MIMO CHANNELS USING OBLIQUE PROJECTIONS

Abstract

This dissertation: deals with blind estimation of spatially correlated MIMO (Multiple input Multiple output) channels. The procedure mainly comprises of two main parts. In the first part, ISI due to finite channel length is removed using the Oblique Projections. From the received signal, using oblique projection, an instantaneous mixture of the input symbol sequences is obtained by removing the ISI. In the second part, this instantaneous mixture is used to separate out individual sequences.. To achieve this by enumeration is computationally forbidden. For comparatively lesser computational complexity two block algorithms ILSP and ILSE can be used. However, they also impose high computational load. A new algorithm has been used in line with ILSP to reduce the computational complexity drastically. Computationally this new algorithm is much efficient due to use of a small received signal data length used in ILSP for main processing which is responsible for reducing computational load. A concept is also proposed to increase the capacity of the system for spatially correlated MIMO channels. Using this concept the degradation in capacity due to transmit antenna correlation is overcome making the system as efficient as uncorrelated transmit antenna system. The performance of the concepts has been tested by simulations. MATLAB 6.5 has been used for the simulation purpose