Optimal Coverage and Rate in Downlink Cellular Networks: A SIR Meta-Distribution Based Approach

TL;DR

This paper introduces a meta-distribution based analytical framework for optimizing coverage and spectral efficiency in downlink cellular networks, emphasizing the importance of considering user experience consistency over traditional first-order metrics.

Contribution

It develops a novel meta-distribution approach for SIR analysis, revealing that optimizing BS height based on coverage probability alone is suboptimal for network-wide user experience.

Findings

Meta-distribution analysis improves understanding of user experience variability.

Optimal BS height differs when considering meta-distribution versus coverage probability.

Resource allocation can be optimized using the proposed framework for consistent service.

Abstract

In this paper, we present a detailed analysis of the coverage and spectral efficiency of a downlink cellular network. Rather than relying on the first order statistics of received signal-to-interference-ratio (SIR) such as coverage probability, we focus on characterizing its meta-distribution. Our analysis is based on the alpha-beta-gamma (ABG) path-loss model which provides us with the flexibility to analyze urban macro (UMa) and urban micro (UMi) deployments. With the help of an analytical framework, we demonstrate that selection of underlying degrees-of-freedom such as BS height for optimization of first order statistics such as coverage probability is not optimal in the network-wide sense. Consequently, the SIR meta-distribution must be employed to select appropriate operational points which will ensure consistent user experiences across the network. Our design framework reveals that the traditional results which advocate lowering of BS heights or even optimal selection of BS height do not yield consistent service experience across users. By employing the developed framework we also demonstrate how available spectral resources in terms of time slots/channel partitions can be optimized by considering the meta-distribution of the SIR.