An extension of the frame-asynchronous coded slotted ALOHA (FA-CSA) protocol for grant-free access, based on packet fragmentation and fragment erasure coding, is investigated. In the analyzed scheme, active users employ maximum distance separable erasure codes at fragment level and transmit encoded fragments on orthogonal subcarriers. Fragment decoding errors at physical layer, even in absence of collisions, are included in the analysis. Asymptotic load threshold values approaching the upper limit are obtained by density evolution over a large efficiency range. At finite frame length, FA-CSA configurations with a very good reliability versus scalability tradeoff are obtained, together with accurate error floor estimation.

We introduce a massive multiple access scheme designed to meet different trade-offs between reliability, scalability, and latency. To maximize the number of successfully decoded users, the scheme builds upon coded random access, incorporating both grant-free and grant-based procedures, along with a massive acknowledgment phase conducted at the base station. The main design premise is the establishment of two distinct latency deadlines: the first one guaranteeing high reliability (e.g., between 99% and 99.99%), and the second one enforcing ultra-high reliability, even above 99.9999%. This dual-latency approach, supplemented with massive MIMO, enables the system to support a higher number of active users per frame while meeting stringent reliability requirements. Throughout the paper, we present a theoretical analysis and derive performance bounds to guide and support effective system design. The approach opens the door for the development of critical services that bridge the gap between massive machine-type communication (mMTC) and ultra-reliable and low-latency communication (URLLC), providing a more flexible and efficient framework for next-generation systems.

In this paper, feedback-aided coded random access (CRA) protocols for grant-free massive access, with and without exploitation of the power domain, are investigated. The developed schemes can support non-instantaneous acknowledgment messages, along with their time resources, with very little penalty in terms of the achieved tradeoff between scalability, reliability, and latency. The new CRA-type protocols rely on waiting slots introduced between consecutive transmissions from the same device to pipeline the feedback reception without degrading the throughput. Their performance can be further enhanced by exploitation of the power domain, in particular by introduction of a deterministic power selection scheme designed for transmission of different packet replicas over a short time window. The system performance is investigated assuming a realistic wireless channel model, a massive MIMO base station, and a realistic processing, via simulation and analysis. The achieved results show that the realistic feedback-aided protocols with deterministic power selection can support 12.8 grant-free users per slot guaranteeing a packet loss rate of 10−4while a massive MIMO base station equipped with 128 antennas is employed

This paper elaborates on the idea of building grant-free channel access schemes from non-orthogonal multiple access ones, and proposes an explicit such scheme based on sparse code multiple access (SCMA). In the designed protocol, SCMA codebooks and pilots are chosen by users in a fully uncoordinated fashion, with multiple pilots associated with the same codebook to aid codebook detection. A modified two-stage SCMA decoder is proposed, where a low-complexity collision resolution algorithm, working on a super-constellation, and an SCMA message passing detector are iteratively applied. Numerical results, integrated by analysis in the high signal-to-noise ratio regime, highlight a potential for the proposed scheme in the context of massive uncoordinated machine-type uplink.

In this paper, new coded random access schemes for massive IoT, capable of accommodating a non-instantaneous feedback from the receiver, are developed. The proposed schemes are based on the introduction of waiting slots between any two consecutive replicas transmitted by the same active device. The waiting window can be exploited by the device to receive acknowledgment messages that, otherwise, would consume uplink resources with a consequent performance degradation. The achievable performance of the developed schemes is investigated, by analysis and simulation, over a realistic wireless channel model and realistic signal processing at the base station, showing negligible losses with respect to previously proposed coded random access systems with idealized instantaneous feedback.