Frame Structure and Resource Optimization for Hybrid Long and Short Packet NOMA-based Data Collection in IIoT with Imperfect SIC

Industrial internet of things (IIoTs), which contains different types of devices with heterogeneous quality of service (QoS) requirements, has encountered significant challenges on guaranteeing the needs of heterogeneous data collection utilizing limited resources. In this paper, we investigate the joint frame structure and resource optimization for the hybrid long and short packet non-orthogonal multiple access (NOMA) based data collection with imperfect successive interference cancellation (SIC) in IIoTs, where a number of short and long packets can multiplex the same time-frequency resource simultaneously to guarantee their respective heterogeneous QoS requirements. Specifically, the short-packet is firstly decoded to guarantee low latency, afterwards the superposed long-packet can be decoded to maintain high signal to interference plus noise ratio (SINR) performance. A joint short-packet scheduling, pilot length, blocklength and dynamic power allocation (JSLP) algorithm is proposed to minimize the maximum block error probability among short-packets and mitigate the impact of SIC error propagation in NOMA transmission while maintaining a high SINR of longpacket, with the assistance of the derived optimal closed-form short-packet scheduling results and pilot and block length expressions. Thanks to the closed-form expressions, the proposed JSLP algorithm demonstrates near-optimal performance and a significant complexity reduction compared to the exhaustive search, leading to fast convergence. Numerical results demonstrate that the designed hybrid NOMA-based frame structure and JSLP algorithm are robust against the SIC error propagation, and can maintain a high level of fairness by significantly mitigating the maximum block error probability gap among short-packets.