As an emerging system architecture, heterogeneous cloud radio access networks (H-CRANs) can improve system capacity, enlarge coverage, and enhance energy/spectral efficiency. Meanwhile, this newborn architecture also brings many open problems for traditional topics, including synchronization, channel estimation, and data detection. In this article, we present a comprehensive analysis on obtaining CSI in H-CRANs.
Specifically, we recognize seven challenges in channel estimation that are caused by a large number of channel parameters, heterogeneity of access nodes in H-CRANs, and the time delays among different nodes. Several research directions for handling these challenges are also proposed, for example, array signal processing and channel compression can eliminate the number of channel estimates, while channel prediction and modification for high-speed railway communications and adaptive downlink array from uplink measurements excel in overcoming the non-reciprocity in channel parameters.