A Joint Pilot-Data Sparse Vector Transmission Framework for Short-Packet URLLC

Abstract

Sparse vector transmission (SVT) has garnered attention for short-packet communications due to its simple implementation and reliable transmission. However, its reliability in mobile scenarios remains an open challenge. To address this issue, a joint pilot-data SVT scheme tailored to the SVT block coding structure is proposed, in which both pilots and data are mapped into a unified sparse vector, and then forms a transmitted vector over an orthogonal frequency-division multiplexing (OFDM) symbol through random spreading. At the receiver, a joint iterative channel estimation and data detection algorithm is designed to progressively refine performance by selecting highly reliable decoded data as pseudo-pilots. Simulations indicate that the proposed SVT scheme outperforms both conventional separated pilot-data mapping and power-allocation-based superposition in block error rate (BLER) performance. Impressively, its first-iteration performance is already close to the converged performance of the power-allocation-based superposition scheme.