A Non-orthogonal Random Access Scheme Based on NB-IoT

Abstract

Narrowband Internet of Things (NB-IoT) is a Quasi-5G technology that will provide coverage for massive number of low-power consumption applications. Hence, the potential massive random access (RA) users call for a new RA scheme. In this paper, based on the key ideas of non-orthogonal multiple access and successive interference cancellation, the NB-IoT non-orthogonal random access (NB-NORA) scheme has been proposed to provide a more effective access performance for NB-IoT system. Further, the RA preamble (MSG1) transmission collision probability and initial layer 3 message (MSG3) outage performances are analyzed based on stochastic geometry, which is compared with the traditional NB-IoT orthogonal random access (NB-ORA) scheme. Simulation results show that the maximum throughput of MSG1 within NB-NORA scheme is double that within NB-ORA scheme, and the successful RA probability of NB-NORA is increased by about 75%.

Appendix

1.1 NB-NOMA System Model [13]

We assume a single-cell uplink transmission scenario, inwhich the evolved NodeB (eNB) is loacted at the center of the cell. The channel between the mth terminal and the eNB is denoted by $h_{m}$ and $h_{m} = \frac{{g_{m} }}{{l_{m} }}$, where $l_{m}$ and $g_{m}$ denotes pathloss and Rayleigh fading channel gain, respectively. To simplify the analysis, $l_{m}$ is modelled by Free-Space path loss model [18], i.e., $l_{m} = \left( {\frac{{\sqrt {G_{l} } \lambda }}{4\pi d}} \right)$, where $G_{l}$ is the product of the transmit and receive antenna field radiation patterns in the line-of-sight (LOS) direction, and $\lambda$ is the signal weave-length and $d$ denotes the distance between terminal and eNB. The probability density function (PDF) of $\left| {g_{m} } \right|^{2}$ can be written as

$$f_{{\left| {{\text{g}}_{\text{m}} } \right|^{ 2} }} (x) = \frac{1}{{2\mu^{2} }}e^{{ - \frac{x}{{2\mu^{2} }}}}$$

(17)

where $\mu$ is the variance of the normal distribution $N(0,\mu )$. Assuming that there are $M$ terminals sharing the same uplink channel simultaneously, the received signal at eNB is given by

$$Y = \sum\limits_{i = 1}^{M} {h_{i} \sqrt {p_{i} } s_{i} } + n$$

(18)

where $p_{i}$ and $s_{i}$ are the transmit power and transmit messages from the ith terminal, respectively. $n$ denotes the additive noise at eNB. In order to split the overlapped signals, SIC receiver is carried out at eNB. Before eNB detects the mth terminal’s message, it decodes the prior ith $(i < m)$ terminals’ message first, then remove the message from its observation, in a successive manner. The rest (M-m) terminals’ messages are regarded as interferences. As a result, the achievable data rate of the mth terminal is

$$R_{m} = \log \left( {1 + \frac{{p_{m} \left| {h_{m} } \right|^{2} }}{{\sum\nolimits_{i = m + 1}^{M} {p_{i} \left| {h_{i} } \right|^{2} + \sigma^{2} } }}} \right)$$

(19)

Assuming $\hat{R}_{m}$ is the target data rate of the mth terminal, then the event that eNB successfully detects the mth terminal’s message can be defined as

$$R_{m} \ge \hat{R}_{m}$$

(20)

Note that the pre-condition of (13) is $R_{m} \ge \hat{R}_{m} (i < m)$ i.e., eNB needs to correctly decode the prior $(m - 1)$ terminals’ messages before detecting the mth terminal’s message, where $R_{i}$ and $\hat{R}_{i}$ are the achievable data rate and the target data rate of the ith terminal, respectively.