High-order contrasts for independent component analysis
Neural Computation
Adaptive ICA with Order Statistics in Multidimensional Scenarios
IWANN '01 Proceedings of the 6th International Work-Conference on Artificial and Natural Neural Networks: Bio-inspired Applications of Connectionism-Part II
Space-Time Coding
| Hi-index | 0.00 |
Spatial multiplexing is currently one of the most promising techniques exploiting the spatial dimension to increase data rates. Most of existing methods are based on coherent detection techniques that imply multichannel estimation. This procedure, especially for time-varying channels, increases the overhead rate due to the periodical training requirement. A suitable approach dealing with this scenario proposes the use of Blind Source Separation (BSS) principles to minimize the mentioned overhead still offering the increased data rate. The authors have developed in previous publications a new BSS technique based on Order Statistics (OS) labeled as ICA-OS with very satisfactory performance in static scenarios. In these studies it was already realized that the amount of data required for convergence was significantly less than other well known methods. Therefore, in this current contribution we present some results showing the capability of our procedure to deal with time-varying channels typical of mobile applications without training requirement.