By reducing the complicated signaling overhead, grant-free multiple access (GF-MA)

technique is deemed as a promising uplink access solution in future massive Internetof-

Things (IoT) communications. In this thesis, grant-free non-orthogonal multiple

access (GF-NOMA) techniques are investigated, which enable massive connectivity

with low access latency. In particular, two types of GF-NOMA techniques are studied,

contention-free and contention-based, respectively. Specifically, in contention-free

GF-NOMA schemes, each user is pre-allocated a unique signature sequence for the incoming

uplink access. At the receiver, compressed sensing (CS) theory is applied to

carry out user activity identification and data detection, by taking advantage of the

sporadic transmission feature. On the other hand, contention-based GF-MA schemes,

also known as uncoordinated random access (URA) schemes, allow for an arrive-andgo

uplink access manner, without the need for prior coordination and resource preallocation.

However, due to the lack of central coordination, when multiple users select

the same resource block to transmit simultaneously, frequent collisions lead to low system

throughput, which becomes the bottleneck. The integration of NOMA techniques

can alleviate these impediments, thereby enhancing system spectrum efficiency.

The thesis contains four technical chapters. The first two contributions aim at the

design of low-complexity CS-based MUD design in the contention-free GF-NOMA systems,

while the remaining two focus on the performance analysis and traffic load regulation

in contention-based GF-NOMA systems.

In the first work, we propose a low-complexity CS-based sparsity adaptive block gradient

pursuit (SA-BGP) algorithm in uplink GF-NOMA systems. The proposed SA-BGP

algorithm is capable of jointly carrying out channel estimation, user activity detection

and data detection without knowing the user sparsity level. By exploiting the inherent

sparsity of transmission signal and gradient descend, the proposed method enjoys a

good detection performance with substantial reduction of computational complexity.

In the second work, block coordinate descend (BCD) method is employed in uplink GFNOMA

systems, aiming at reducing the computational complexity of the CS-MUD. We

design two modified BCD based algorithms, called enhanced BCD (EBCD) and complexity

reduction enhanced BCD (CR-EBCD), respectively. In particular, by incorporating

a novel candidate set pruning mechanism into the original BCD framework, the

proposed EBCD algorithm achieves remarkable CS-MUD performance improvement. In

addition, the proposed CR-EBCD algorithm further ameliorates the proposed EBCD

by eliminating the redundant matrix multiplications during the iteration process. Consequently,

compared with the proposed EBCD algorithm, the proposed CR-EBCD

algorithm enjoys two orders of magnitude complexity saving without sacrificing any

CS-MUD performance, rendering it a viable solution for future massive machine-type

communication (mMTC) scenarios.

In the third work, a novel uplink URA protocol is proposed, which integrates with sparse

code multiple access (SCMA) technique, referred to as SCMA-aided slotted-ALOHA

(S-ALOHA) scheme. Specifically, each active user transmits an SCMA codeword to

the access point (AP) in an arbitrary time slot whenever they want without the need

of prior scheduling. However, due to the lack of central management in the proposed

URA-based scheme, codebook collisions become inevitable, making the decoding challenging

at the AP. To cope with this issue, we propose an interference-canceling (IC)

first decoding strategy at the AP, which is able to partially address collision problems,

contributing to a higher system throughput. For the proposed IC-first decoding strategy,

a theoretical expression of the throughput has been derived. Moreover, to alleviate

the throughput degradation under the congested user traffic, a user barring mechanism

is introduced to control the traffic load. Based on the estimated real-time load, the

AP adaptively adjusts the access probability and redistributes the actual access load.

Thus, the system throughput can maintain steadily close to the maximum value.

In the fourth work, an efficient URA scheme with power domain non-orthogonal multiple

access (PD-NOMA), namely PD-NOMA aided multi-channel slotted ALOHA (MSALOHA),

is investigated. We first analytically study the throughput performance and

propose a more accurate theoretical bound for the system throughput, compared to

the existing results. Then, the idle subchannel probability, which can be used to indicate

the real-time traffic load, is reformulated by analysing the observable non-utilized

power level (ONU-PL) and a concise analytical expression is further derived. Furthermore,

in order to effectively manage the challenges posed by congested traffic loads, a

user load control mechanism is introduced. This mechanism is devised to ensure that

the real-time traffic load remains in close proximity to the optimal level by adaptively

controlling the user access probability, leading to a stable system throughput.