Sparse NOMA: An Information Theoretic Perspective

Non-orthogonal multiple access (NOMA) is a key enabler in the design of future overloaded beyond-5G communication systems with many more designated users than available physical resources, precluding the conventional orthogonal multiple access (OMA) paradigm. NOMA technologies generally comprise two main manifestations, power-domain NOMA and code-domain NOMA. Power-domain NOMA essentially relies on direct superposition of the transmitted signals, successive interference cancellation at the receivers, and appropriate power allocation to different users in order to achieve desired performance objectives. Under the code-domain NOMA paradigm, the users’ signals are distinguished by different spreading signatures chosen to facilitate efficient multiuser detection at the receivers. In particular, sparse NOMA has gained considerable interest in recent years due to its appealing attributes: enhanced spectral efficiency and effective implementation exploiting iterativemessage passing algorithms.

Notwithstanding their great practical promise and potential, sparse NOMA techniques often pose serious analytic challenges, and their information-theoretic performance limits are not easily tractable even in the simplest settings. We review recent progress in the analytic assessment of the information theoretic capacity, focusing on regular sparse NOMA, where a fixed (and finite) number of orthogonal channel resources is allocated to any designated user, and vice versa. Considering a single-cell system, additive white Gaussian noise (AWGN) channels and the large system limit, three settings of interest are explored. Firstly, we examine a system comprising a single class of users, employing regular sparse NOMA with fully coordinated grant-based access and equal received powers. For this baseline setting, the corresponding optimum spectral efficiency is shown to admit a closed-form expression, while relying on a fundamental limiting spectral distribution result. Furthermore, regular sparse NOMA exhibits superior performance with respect to random dense code-domain NOMA, as well as certain irregular sparse NOMA alternatives. Next, a mixed setting comprising two classes of users is considered, where one class employs regular sparse NOMA, and a second class employs dense code domain NOMA. This setting represents practical use cases comprising a combination of users with fully coordinated access and users for which signature coordination is hard to enforce. Here, an exact characterization of the achievable ergodic class-throughput region is derived and, in particular, the maximum achievable sum rate is specified via a free additive convolution of probability measures. Finally, a sparse two user-class setting is considered, where both user classes employ regular sparse NOMA, and the two classes are distinguished by their received powers. This setting represents, e.g user cases, comprising users with different power constraints, or combinations of cell-edge and cell-center users. It can also be alternatively viewed as a combination of power-domain and code-domain NOMA. Introducing random coordinate transformations via Haar matrices to facilitate the analytic treatment, the achievable class-throughput region in this setting is derived also via a free additive convolution of probability measures. For both two user-class settings, the superiority over random fully dense code-domain NOMA is clearly demonstrated. These observations broaden the information-theoretic perspective on code-domain NOMA, establish key tools for investigating more diverse use-cases of interest, and provide insights into the development of efficient NOMA
schemes for future wireless networks.
-----------------------------------------------------------------------
This talk is based on joint research with B. Zaidel, O. Shental, C. Eger, supported by the German-Israeli Project
Cooperation (DIP), under Project SH 1937/1-1, and by the Israel Scientific Foundation (ISF), under grant No. 1897/19.