Accelerated Maximum-Likelihood for Massive MIMO Unsourced Random Access

You, Juntao; Wang, Wenjie; Liang, Shansuo; Han, Wei; Bai, Bo

doi:10.1109/globecom48099.2022.10001287

Cited by 1 publication

(2 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Second, an additional CB-MLD detection is performed in the first time slot in the MUD of IM-MSOP-II, compared with the MUD of IM-MSOP-I. Here, the two-stage CB-MLD consisting of one-step hard threshold and iterative gradient descent [17] is employed to reduce the complexity since 2 B1 is very large here. The complexity of the two-stage CB-MLD is O N 2 N 2 1 Iter 3 , where N 2 2 B1 is the size of candidate set after one-step hard threshold preprocessing and Iter 3 is the number of gradient descent iterations.…”

Section: Complexitymentioning

confidence: 99%

“…Moreover, an outer tree code is taken to introduce parity-check constraint in the information sequence [11]- [18]. In the multiuser detection (MUD), the non-Bayesian covariancebased maximum likelihood detection (CB-MLD) [13]- [17] or the approximate message passing (AMP) algorithm [18] [19] is employed to decode the inner CS code, and the outer tree decoder is utilized to stitch the information across different time slots. However, the high complexity of the tree decoder hinders the practical application of this concatenation scheme.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Index Modulation-Aided Unsourced Random Access in the Block-Fading Channel

Gao,

Zheng

2024

IEEE Open J. Commun. Soc.

View full text Add to dashboard Cite

In this article, we propose three index modulation (IM) aided unsourced random access schemes in the block-fading multiple-antenna channel. In all the proposed schemes, the signal frame is partitioned into multiple time slots, and only part of time slots are activated to transmit signal. The active time slots are determined by the active time slot pattern (ATP) information through utilizing the principle of IM. In the first two schemes, each time slot includes pilot sub-slot and data sub-slot. The multiplestage orthogonal pilots (MSOPs) are employed for both channel estimation and user identification, and the pilot-hopping pattern (PHP) of MSOPs determines the pilots transmitted over the active pilot sub-slots. Moreover, in the first scheme, the polar-coded signal is transmitted repeatedly in the data sub-slots of active time slots. In the second scheme, the ATP and PHP information is first transmitted through the sparseregression-code (SPARC) in the first time slot, and the polar-coded signal is divided into multiple parts which are transmitted in those active data sub-slots, respectively. At the receiver, the iterative orthogonal pilot-aided channel estimation and decoding with successive interference cancellation is presented. In the third scheme, the signal frame is also partitioned into multiple time slots where only two time slots are activated by each user. The first time slot is always activated to transmit the ATP and active channel-use pattern (ACP) information by the SPARC. The other ATP-guided active time slot transmits the remaining message by IM over the ACP-guided active channel uses. Different from the first two schemes, the third scheme has no pilot signal and thus the blind detection based on the bilinear generalized approximate message passing algorithm is employed to decode the message. Finally, computer simulations show that, compared with the state-of-the-art scheme, the first scheme has better performance for all the numbers of active users from 50 to 500, while the last two schemes much better only when less than about 200 active users exist.INDEX TERMS Unsourced random access, index modulation, orthogonal pilot, blind detection, blockfading channel.

show abstract

Section: Complexitymentioning

confidence: 99%