Using dependent component analysis for blind channel estimation in distributed antenna systems
Janis Nötzel, Christian Arendt · 2016
This contribution is motivated by a key question that car manufacturers have to face in the context of vehicular connectivity: What is the benefit of a higher number of spatially separated antennas mounted on a vehicle resulting in a so-called `distributed antenna system' (DAS)? We study this question in a situation where demodulation and pre-decoding of the incoming signal is performed independently at each antenna. The resulting bit stream is fed to a central processing unit, where bit error rate estimation and error correction are performed. It turns out that an increase in number of antennas from 1 to 2-5 unleashes completely new features of the antenna system. In this contribution, we provide the conceptual study of an algorithm that exploits these new features. It has the advantage of learning the channel statistics blindly-no pilot sequences are necessary in an idealized scenario.