Channel-Aware Random Access in the Presence of Channel Estimation Errors

TL;DR

This paper analyzes how imperfect channel state information affects the stability and performance of a decentralized random access protocol called CARA, comparing it to centralized scheduling in multi-user wireless networks.

Contribution

It provides the exact stability region of CARA under channel estimation errors and compares its performance with centralized policies considering multipacket reception capabilities.

Findings

CARA's stability region is explicitly characterized with estimation errors.

CARA can outperform centralized policies with high MPR capability.

Imperfect CSI impacts the stability and efficiency of decentralized access.

Abstract

In this work, we consider the random access of nodes adapting their transmission probability based on the local channel state information (CSI) in a decentralized manner, which is called CARA. The CSI is not directly available to each node but estimated with some errors in our scenario. Thus, the impact of imperfect CSI on the performance of CARA is our main concern. Specifically, an exact stability analysis is carried out when a pair of bursty sources are competing for a common receiver and, thereby, have interdependent services. The analysis also takes into account the compound effects of the multipacket reception (MPR) capability at the receiver. The contributions in this paper are twofold: first, we obtain the exact stability region of CARA in the presence of channel estimation errors; such an assessment is necessary as the errors in channel estimation are inevitable in the practical situation. Secondly, we compare the performance of CARA to that achieved by the class of stationary scheduling policies that make decisions in a centralized manner based on the CSI feedback. It is shown that the stability region of CARA is not necessarily a subset of that of centralized schedulers as the MPR capability improves.