Reciprocity calibration of Distributed Massive MIMO Access Points for Coherent Operation

Novel network architectures for 6G distributed massive MIMO systems rely on coherent signaling by distributed antenna panels which are coordinated by a central controller. This type of network architecture is based on reciprocity operation where antenna panels rely on uplink channel estimates for coherent downlink precoding. This paper proposes a calibration method for distributed massive MIMO systems, which overcomes hardware non-reciprocities in order to enable reciprocity-based operation. Measurements for system calibration are collected via a beam-sweep between all pairs of antenna panels. We lay out the system model for this new setup, and propose a maximum likelihood-based procedure to compute calibration coefficients based on the collected measurement set. The procedure is computationally efficient and stable, since 1) each iteration has a closed-form, and 2) the procedure is guaranteed to converge to at least a local optimum (or saddle point). Simulations indicate significant calibration improvements compared to re-using state of the art calibration schemes for the problem at hand.