In cellular communication systems, Multiple Access Techniques define ways by which multiple users access a single communication channel.
- 2G uses TDMA (Time division multiple access) where multiple users are assigned a time-slot to access the channel.
- 3G uses CDMA (Code division multiple access) where multiple users are assigned a code to access the channel.
- LTE uses OFDMA (Orthogonal frequency division multiple access) where users are assigned orthogonal frequencies to access the channel.
- LTE-A also uses Multi User Superposition coding Transmission (MUST) where eNB serves two DL users at same OFDMA subcarrier. Users are separated by different power levels. Users with better channel conditions are assigned less power and users with bad channel conditions are assigned more power. eNB uses supposition coding and sends information for two users at the same time. This is a class of Non orthogonal Multiple Access (NoMA) scheme and only applicable in DL direction. [Ref: https://www.youtube.com/watch?v=JXkBSb-mhjQ]
NoMA Study Item for 5G-New Radio (NR) was approved and revised recently. Clear industry interest in NoMA has been shown, given the support of nearly 40 companies. It is expected that NoMA will benefit various use cases and deployments. This will be supported in both DL and UL directions.
NoMA is expected to serve many scenarios and use cases, for example, connection density, UE power consumption and signaling overhead reduction, Coverage extension, Reliability and resource utilization, Latency reduction and Number of Users and capacity enhancements.
One of the challenge in supporting NoMA is complex receiver design. For example,
“In orthogonal multiple access, the DMRS of different UEs are not overlapping and there is no pilot contamination. In fact, one has to estimate the channel only for the RE positions where a particular UE is active. Therefore, we only need to place pilots over the indices where each UE is active. In NOMA, though, multiple UEs are transmitting over the same set of resources hence we need to alleviate the impact of pilot contamination.” [Ref: R1-1805005]
“When multiple UEs are allocated on the same time/frequency resources then the gNB does not have a priori knowledge of the UEs’ ID; when the gNB detects a transmission over the shared resources, it does not know which of the assigned UE transmitted. Therefore, it is important to establish mechanisms for acquiring the UE ID, which is especially relevant for NoMA where resources are shared. Moreover, the identification mechanism should be at least as reliable as the payload transmission. The reason is that a passed CRC-check without a UE-ID, tied to the payload, will be lost effort since the gNB cannot acknowledge the packet reception; it will instead have to wait for a new transmission. In the opposite scenario, where the UE-ID is detected but the payload decoding fails, at least we can exploit the acquired UE-ID in order to take remedy actions, such as retransmissions on orthogonal resources.” [Ref: R1-1805005]
“One key issue with any multiuser detection is the power imbalance between users. More specifically, in most designs it is assumed that the received power from different users at the receiver is either the same or can be ideally controlled. However, in reality the power control process is not ideal and there can be a difference of ∆P between the target power and the realistic power at the receiver. It is important that the power control imperfection is considered in the evaluation of different NoMA schemes.” [Ref: R1-1805005]
“One of the most pronounced features of NR NOMA is the UE and gNB capabilities to operate in RRC inactive/idle states without a UL grant. The support for grant-free transmission can significantly reduce the power consumption and latency, which is a desirable property for NR mMTC and URLLC.” [Ref: R1-1804826]
“For mMTC use case, managing timing advance and maintaining synchronization across all NOMA UEs requires large power consumption and incur large latency. Therefore, the ability to operate asynchronously without any TA is an important feature for NOMA schemes. For long codes based RSMA with DFT-s-OFDM waveform, time domain processing, e.g. Rake receiver, can be immediately applicable for NOMA multi-user detection, even for asynchronous scenario. However, for other NOMA schemes especially the ones based on short NOMA codes, the chip-level alignment across the UEs are crucial. It is unclear yet how to apply the short-codes based NOMA schemes for asynchronous transmission.” [Ref: R1-1804826]