Direction Finding for Compressed Massive MIMO Systems Exploiting Mixed Circular and Non-Circular Sources

Massive MIMO systems can significantly enhance communication quality by exploiting the direction-of-arrival (DOA) information of users. However, the high system complexity and cost pose major challenges to their further development. Compressed arrays, which apply compressive sampling in spatial domain, can effectively reduce system complexity and cost. Nevertheless, this compression inevitably results in information loss, leading to degraded DOA estimation performance. Given that many communication signals exhibit non-circular properties, this paper considers a mixed circular and non-circular signal scenario, and proposes exploiting the conjugate information inherent in non-circular signals to mitigate the performance loss caused by compression. In this paper, the signal receiving model is first constructed for a compressed array considering a mixture of circular and non-circular sources. Subsequently, the Cramér-Rao Bound (CRB) for DOA estimation under the proposed scheme is derived, and the factors affecting the number of resolvable sources are discussed by analyzing the existence condition of the CRB. Two direction-finding algorithms are then developed for the proposed scheme, i.e., the non-circular compressed array-based multiple signal classification (NC-CA-MUSIC) algorithm and the non-circular compressed array-based compressive sensing (NC-CA-CS) algorithm. Both algorithms exploit the non-circular information contained in the pseudo-covariance matrix, thereby improving estimation performance in terms of degrees of freedom (DOFs) and accuracy. However, the two algorithms are designed for different scenarios. NC-CA-CS can exploit more degrees of freedom but exhibits lower angular resolution compared to NC-CA-MUSIC. Finally, numerical simulations are conducted to validate the effectiveness of the proposed scheme.