Logo image
Successive MMSE group decoding and max-min power control for uplink multiceli NOMA systems under pilot contamination
Conference paper

Successive MMSE group decoding and max-min power control for uplink multiceli NOMA systems under pilot contamination

Y.-W. Peter Hong, An-An Lee and Yu-An Chen
Proceedings - 9th Asia-Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2017, Vol.2018-February, pp.823-831
02/2018

Abstract

Massive MIMO Max-min fair power control Non-orthogonal multiple access Pilot contamination Artificial Intelligence Human-Computer Interaction Information Systems Signal Processing
This work proposes a successive minimum mean square error (MMSE) group decoding scheme and a joint pilot and data power control scheme based on the max-min weighted rate criterion. Successive decoding is a key enabling technology in non-orthogonal multiple access (NOMA) systems, but its effectiveness may be limited by intercell interference in the data transmission phase and also by pilot contamination in the training phase. The proposed successive MMSE group decoding scheme limits these effects, and also avoids the complexity of maximum likelihood decoding by utilizing a linear MMSE filtering followed by parallel single-user decoders in each stage of the successive group decoding process. A closed-form approximation of the achievable ergodic rate is derived and utilized to determine the power allocation among pilot and data symbols. The coordinated power control problem is formulated as a max-min weighted rate problem subject to a total power constraint, and is solved using an outer iteration that alternates between the optimization of pilot and data powers, and an inner iteration that borrows techniques from nonlinear Perron-Frobenius theory to solve the data and pilot powers in their respective subproblems. Moreover, a coordinated greedy user grouping scheme is also proposed to further maximize the worst weighted rate. The effectiveness of the proposed schemes are demonstrated through computer simulations.

Metrics

1 Record Views

Details

Logo image