Multi-User Detection in Uplink Grant-Free NOMA with Dynamic Random Access Using Sinusoidal Sequences

Abstract

In the paradigm shift toward machine-type communications (MTC), the Third-Generation Partnership Project (3GPP) has identified massive machine-type communications (mMTC) as one of the three main use cases of 5G, which will facilitate wireless connectivity among the increasingly large number of devices in IoT. mMTC features uplink-dominated sporadic transmissions of small payloads. The excessive overhead signaling associated with complex grant-acknowledgement-based uplink legacy networks cannot be directly adopted for mMTC while maintaining the stringent latency requirements of IoT. Uplink grant-free non-orthogonal multiple access (NOMA) offers an array of features that effectively address these issues. Due to the inherent sparsity in the user activity pattern in mMTC systems, variations of compressive sensing-based multi-user detection (CS-MUD) along with either Zadoff–Chu or random sequences are employed for activity detection, channel estimation and data detection. This work assumes that substantial statistics about devices’ channels and noise are available at the base station (BS) and that the devices transmit data symbols across the entire random-access opportunity. Since these assumptions may not wholly reflect a realistic scenario, we provide ways of relaxing them, for example, by supposing that the devices do not have to transmit data throughout the random-access opportunity. We allow devices to enter and exit the random-access opportunity dynamically and detect their activity using statistical hypothesis tests. By using sinusoidal sequences with these relaxed assumptions, multi-user detection is possible. We also carry out numerical investigations to validate the potential of the proposed system in a realistic mMTC environment. The simulation results demonstrate the utility of sinusoidal spreading sequences in facilitating fast activity detection with dynamic random access.