Compressed Training Based Massive MIMO

Massive multiple-input-multiple-output (MIMO) scheme promises high spectral efficiency through the employment of large scale antenna arrays in base stations. In time division duplexed implementations, co-channel mobile terminals transmit training information such that base stations can estimate and exploit channel state information to spatially multiplex these users. In the conventional approach, the optimal choice for training length was shown to be equal to the number of users, $K$ . In this paper, we propose a new semiblind framework, named as “MIMO Compressed Training,” which utilizes information symbols in addition to training symbols for adaptive spatial multiplexing. We show that this framework enables us to reduce (compress) the training length down to a value close to $\log _2(K)$ , i.e., the logarithm of the number of users, without any sparsity assumptions on the channel matrix. We also derive a prescription for the required packet length for proper training. The framework is built upon some convex optimization settings that enable efficient and reliable algorithm implementations. The numerical experiments demonstrate the strong potential of the proposed approach in terms of increasing the number of users per cell and improving the link quality.