Frequency-Domain NOMA With Two Sets of Orthogonal Signal Waveforms

We present a non-orthogonal multiple access (NOMA) scheme based on combining orthogonal frequency-division multiple access and multi-carrier code-division multiple access. Contrary to other NOMA architectures, the proposed approach does not require any power imbalance between user signals, making it possible to grant all users the same data rate and performance. Another major contribution concerns the receiver operation: Many papers in the recent NOMA literature refer to single-stage successive interference cancellation, where the strong user signal is first detected and subtracted from the received signal before detecting the weak user signal. While this approach may be optimal from information theoretic viewpoint, we show that this type of receiver cannot achieve performance results close to those of interference-free transmission and that, in our framework, an iterative receiver is actually required for this purpose. We investigate both hard-decision and soft-decision iterative receivers and show that a soft-decision receiver significantly improves the performance and increases the number of users.

I. Introduction

Non-orthogonal multiple access (NOMA) is currently a hot research topic for the physical layer of future 5G cellular networks. The interest in this multiple access technique originated from a well-established result in multi-user information theory, which indicates that orthogonal multiple access (OMA) is not optimal in general and that superposition coding coupled with successive interference cancellation (SIC) provides an optimal solution for multiple access [1]. Historically, orthogonality of different user signals was always perceived as a desirable property, and OMA schemes like time-division multiple access (TDMA), orthogonal code-division multiple access (OCDMA), and orthogonal frequency-division multiple access (OFDMA) have been used in different generations of cellular systems. But a closer look at the channel capacity, which indicates that OMA is not always optimal, has opened up new perspectives and research directions for future networks, and an impressive number of papers have been published on NOMA in the past few years (see, e.g., [2]–[10]). Notice that as pointed out in [11], the NOMA concept was actually introduced back in the year 2000 in the form of multiple access using two sets of orthogonal signal waveforms (see [12]–[14]).