Grant-Free Uplink Transmission With Multi-Codebook-Based Sparse Code Multiple Access (MC-SCMA)

Abstract

In sparse code multiple access (SCMA)-based grant-free (GF) access, both user's channel and the selected SCMA codebook (CB) are considered as a transmission signature for multi-user detection (MUD). In conventional SCMA-based grant-free (GF) access where a single codebook (CB) is employed per a user's random access transmission, however, random selection of CBs and random fading channel reduce the minimum Euclidean distance among overloaded layers (signatures). This significantly degrades the performance. Furthermore, it is also challenging to reduce the collision rate in the course of selecting random access resources while maintaining low preamble overhead. In this paper, we consider a new approach that maps user's consecutive symbols to codewords from the different SCMA CBs in a round-robin manner. It is shown that a combinatorial nature of multiple CBs and the inherent code diversity therein allow to simultaneously reduce collision probability while providing sufficient separation among transmissions, consequently decreasing users' activity misdetection rate. It is also shown that the proposed multi-codebook SCMA (MC-SCMA)-based GF access scheme outperforms the conventional single CB-based SCMA version by overloading of more users (almost up to 200%) as it allows a large number of highly non-orthogonal preambles to be employed while keeping the same preamble length (overhead). Finally, we conclude that the proposed joint design of non-orthogonal preambles and data transmission based on multiple CBs provides a significant performance improvement for the grant-free access system. It can also readily be extended to other code-based non-orthogonal multiple access (NOMA) schemes.

Keywords

• Sparse code multiple access
• grant-free access
• random access
• compressed sensing
• preamble transmission
• channel estimation