On the impact of transmit waveforms on channel estimation inaccuracies in distributed MIMO ad hoc networks

Abstract

In this paper, we present a general analysis to evaluate the performance of correlator-based channel estimation in distributed MIMO ad hoc networks. This analysis is required to characterize the behavior of networks employing advanced physical-level (PHY) techniques, which call for precise assessment of channel effects in order to be operated correctly. To this end, we assume that several non-orthogonal signals are simultaneously received in a coordinated (i.e., frame-synchronous) but symbol-asynchronous fashion, and undergo flat, independent Rayleigh fading. We show that, in this scenario, channel estimates are affected not only by noise, but also by a displacement depending on the current channel realization. Such additional component can be modeled as a Gaussian random variable, of which we calculate the mean and variance, highlighting its dependence on the number of training symbols and the shape of transmit impulse. In particular, we show that an interesting interplay exists between MAC parameters (e.g., throughput) and the type of impulse chosen in an ad hoc networking protocol simulator.