Abstract:

In uplink grant-free massive multiple-input multiple-output (mMIMO) systems, the performance of available methods for joint user activity and signal detection deteriorates when the correlation of receiving antennas or the number of active devices increases.Moreover, the available methods require the knowledge of noise power, which is often practically unknown.To address the above issues, combining approximate message passing with unitary transformation and expectation maximization algorithm to jointly implement user activity and signal detection was proposed.Different from the conventional approximate message passing algorithm, the proposed one assumes that the noise power was unknown.Firstly, by exploiting the approximate message passing algorithm with unitary transform, the distribution of transmitted symbols together with the distribution of noise power was obtained.Secondly, expectation maximization algorithm was applied to estimate the user activity.Finally, the signal detection was implemented by deriving the posterior distribution of the decoupled signal belongs.Simulation results show that the proposed method is better than the traditional method in joint user activity and signal detection.

Key words: massive multiple-input multiple-output, user activity detection, signal detection, approximate message passing with unitary transformation