Equalization and pre-equalization techniques and implementation in software defined radios

Abstract

The huge expansion of wireless communication the recent decades and the increased demand for high-speed wireless connections points out the importance of efficient use of frequency bandwidth which is generally expensive and limited. Communication links between multiple transmit and receive antennas, the so-called Multiple Input and Multiple Output (MIMO) channels, have been under extensive theoretical studies for their potential to provide extra dimensions for communication, without sacrificing bandwidth resources, by multiplexing various independent data streams over the wireless medium. Optimal detection and separation of the multiplexed messages is provided by the Maximum-Likelihood method, at the cost of high complexity. The sub-optimal linear Zero-Forcing Equalization and Pre-equalization methods can drastically decrease the detection complexity at the cost of performance degradation. Both ML and linear Equalization exploit Channel-State Information (CSI) at the receiver for retrieving the various interfering symbols. Linear pre-equalization in contrast, precancels interference at the transmitter which requires feedback in order to obtain the required CSI. The need for feedback can be avoided in Time-Division-Duplex (TDD) systems, where the transmitter can exploit the reverse-channel estimate, based on the reciprocal property of wireless propagation. Utilizing the reciprocal property in practical transceivers requires some form of calibration in order to compensate for the non-reciprocal relationship of their transmit and receive chains. A suitable testbed for implementing and evaluating the above methods by means of Software Defined Radio is provided by the USRP hardware platform and the GNU radio software toolkit