Scheduling and Codeword Length Optimization in Time Varying Wireless Networks

TL;DR

This paper analyzes scheduling and codeword length optimization in time-varying wireless networks, demonstrating that considering channel variation in scheduling significantly improves throughput and approaches optimal performance as the number of users grows.

Contribution

It introduces a scheduling method that accounts for channel time variation, reducing the gap to maximum throughput in large user scenarios.

Findings

Optimizing codeword length enhances throughput.

Conventional scheduling's gap to maximum throughput increases with users.

Proposed scheduling reduces the throughput gap to near zero.

Abstract

In this paper, a downlink scenario in which a single-antenna base station communicates with K single antenna users, over a time-correlated fading channel, is considered. It is assumed that channel state information is perfectly known at each receiver, while the statistical characteristics of the fading process and the fading gain at the beginning of each frame are known to the transmitter. By evaluating the random coding error exponent of the time-correlated fading channel, it is shown that there is an optimal codeword length which maximizes the throughput. The throughput of the conventional scheduling that transmits to the user with the maximum signal to noise ratio is examined using both fixed length codewords and variable length codewords. Although optimizing the codeword length improves the performance, it is shown that using the conventional scheduling, the gap between the achievable throughput and the maximum possible throughput of the system tends to infinity as K goes to infinity. A simple scheduling that considers both the signal to noise ratio and the channel time variation is proposed. It is shown that by using this scheduling, the gap between the achievable throughput and the maximum throughput of the system approaches zero.