Joint estimation of CFOs and Doppler shifts in mmWave distributed MIMO systems

Distributed Multiple Input Multiple Output (MIMO) systems, where the base stations (BSs) have several Radio Units (RUs) that are spatially distributed around the cell, are considered in this study. With such an architecture, each wireless link between the RUs and the devices experiences a different frequency offset. This accumulation of frequency offsets degrades the reconstruction of the coherent downlink signals at the receiver side and therefore limits the system capacity. Such frequency offsets are unavoidable, due to both independent local oscillators inducing Carrier Frequency Offsets (CFOs) and moving devices inducing Doppler shifts. The problem becomes more critical in millimeter wave (mmWave) communications because of large Doppler shifts even for low device speed. Estimating the frequency offsets at the device side requires the simultaneous estimations of multiple frequency offsets which can be computationally intensive. In this study we propose to move the estimation problem at the BS side so that the frequency offsets can be compensated for the downlink which eases the recombination of the precoded signals at the device side. To properly compensate the frequency offsets at each wireless link, the different sources of frequency offset must be jointly estimated. A joint CFOs and Doppler shifts estimation is first described and asymptotically analysed. The paper also includes a performance evaluation of the proposed joint estimation technique. It appears that significant performance gains are achieved for realistic systems. It makes the proposed solution interesting for some beyond 5G MIMO scenarios like indoor reliable industry connectivity or vehicular communications.