Iterative precoding vs decoding in virtual mimo systems with channel mismatching errors

This contribution addresses a mathematical formulation and a comparison in performance of two realistic approaches of cooperation in wireless sensor networks. Tomlinson - Harashima precoding has been used as representative of transmitter cooperation while V-BLAST equalization technique was used as representative for receiver cooperation. The key point of this contribution is that non ideal conditions of carrier synchronization are considered. Recent theory suggests that cooperating at the transmitter results more beneficial than doing it at the receiver. In our hand we will show that, when carriers are not accurate synchronized, receiver cooperation appears to be a more robust selection.

1. Introduction

¹

This work has been partly supported by National Spanish Projects PCT-350100-2004-1, TEC2004-06915-C03-02/TCM, TEC2006-13067-C03-01/TCM and TIC2003-09061-C03-01

Sensor networks and ad-hoc networks are receiving more and more attention from the research community. It has been shown that, under certain assumptions, WSN can be viewed as virtual Multiple-Input Multiple-Output (MIMO) systems [1], [2]. This point allows to relate developed theory and experience in MIMO systems with WSN. One important topic of research in WSN is cooperative communications, that is, how the elements of a WSN arrange the distribution of some common resource to improve the system performance. Basically, there are three possible configurations for cooperation depending on the available information: a) at the transmitter side a group of nodes interchange their codewords and also the channel coefficients, b) at the receiver side nodes share their received signals and channel coefficients and c) cooperation is performed jointly at both sides of the system. It is important to say that if cooperation were ideal, i.e. no restrictions on power or bandwidth and ideal cooperation channel, performance of a WSN can achieve that of a MIMO system. However this assumption can not be applied as cooperation is always by some means penalized. In [5] a very suitable model is provided that addresses penalizations in terms of power, bandwidth and even network topology and establishes a trade-off between them in order to achieve cooperation benefits. Conclusions in [5] reveal that cooperating at the transmitter is more suitable than doing it at the receiver side and even better than doing it at both sides because extra gain is negligible. However when considering some non ideal conditions in the communication channel this assessment may not hold any more.