A Low-complexity Receiver for Massively Concurrent Non-orthogonal Multiple Access

We consider the multi-user joint detection problem of overloaded massively-concurrent NOMA (MC-NOMA) systems by designing a receiver capable of resolving effectively and efficiently the multi-user interference associated with the nonorthogonality inherent in such systems. The proposed receiver is effective in the sense that it is found to achieve the same performance of the brute-force maximum likelihood detector, and is efficient in the sense it significantly outperforms the latter in terms of computational complexity (i.e. run time). The design of the new receiver is based on a two-fold contribution. The first is a new tree optimization (or reduction) technique obtained via successive row- and column-permuted QR decomposition of the MC-NOMA-precoded channel, while the second is a new probabilistic thresholding scheme with a threshold constructed from estimated residual noise distribution and magnitude. We validate the proposed receiver in the context of a downlink MC-NOMA system, with simulation results supporting and quantifying the gains in overloading capability, as well as, complexity reduction.