Pilot-Free Unsourced Random Access Via Dictionary Learning and Error-Correcting Codes

TL;DR

This paper introduces a pilot-free unsourced random access protocol for massive cellular networks using dictionary learning and error-correcting codes, eliminating the need for pilot signals and improving energy efficiency.

Contribution

It proposes a novel URA scheme with a codebook, sparse transmission, and collision resolution that operates without pilot signals, suitable for massive MIMO systems.

Findings

No pilot overhead required, reducing energy consumption.

Effective collision resolution in dense device scenarios.

Numerical results confirm practical viability.

Abstract

Massive machine-type communications (mMTC) or massive access is a critical scenario in the fifth generation (5G) and the future cellular network. With the surging density of devices from millions to billions, unique pilot allocation becomes inapplicable in the user ID-incorporated grant-free random access protocol. Unsourced random access (URA) manifests itself by focusing only on unwrapping the received signals via a common codebook. In this paper, we propose a URA protocol for a massive access cellular system equipped with multiple antennas at the base station. The proposed scheme encompasses a codebook enabling construction of sparse transmission frame, a receiver equipped with dictionary learning and error-correcting codes and a collision resolution strategy for the collided codeword. Discrepant to the existing schemes with necessary overhead for preamble signals, no overhead or pre-defined pilot sequences are needed in the proposed scheme, which is favorable for energy-efficient transmission and latency reduction. Numerical results verify the viability of the proposed scheme in practical massive access scenario.