Abstract
This paper studies the sum throughput of the multi-user multiple-input-single-output networks in the cases with a large-but-finite number of transmit antennas and users. Considering continuous and bursty communication scenarios with different users' data request probabilities, we derive quasi-closed-form expressions for the maximum achievable throughput of the networks using optimal schedulers. The results are obtained in various cases with different levels of interference cancellation. Also, we develop an efficient scheduling scheme using genetic algorithms (GAs) and evaluate the effect of different parameters, such as channel/precoding models, number of antennas/users, scheduling costs, and power amplifiers' efficiency, on the system performance. Finally, we use the recent results on the achievable rates of finite block-length codes to analyze the system performance in the cases with short packets. As demonstrated, the proposed GA-based scheduler reaches (almost) the same throughput as in the exhaustive search-based optimal scheduler, with substantially less implementation complexity. Moreover, the power amplifiers' inefficiency and the scheduling delay affect the performance of the scheduling-based systems significantly.
Original language | English (US) |
---|---|
Article number | 8520925 |
Pages (from-to) | 152-166 |
Number of pages | 15 |
Journal | IEEE Transactions on Wireless Communications |
Volume | 18 |
Issue number | 1 |
DOIs | |
State | Published - Jan 2019 |
Bibliographical note
Publisher Copyright:© 2002-2012 IEEE.
Keywords
- Massive MIMO
- finite blocklength analysis
- genetic algorithm
- large-but-finite MIMO
- machine learning
- scheduling
ASJC Scopus subject areas
- Computer Science Applications
- Electrical and Electronic Engineering
- Applied Mathematics