Partner selection in indoor-to-outdoor cooperative networks: an experimental study

TL;DR

This study presents a novel partner selection protocol for indoor-to-outdoor cooperative wireless networks that significantly extends network lifetime by using real-time fading parameter estimation and a Bayesian approach, validated through real measurements.

Contribution

It introduces a Bayesian-based partner selection protocol leveraging a stochastic model of fading parameters, improving network lifetime without requiring full channel knowledge.

Findings

Infrequent K-factor estimation can double network lifetime.

The Bayesian estimator performs nearly as well as full channel knowledge.

The protocol is compatible with IEEE 802.15.4 networks.

Abstract

In this paper, we develop a partner selection protocol for enhancing the network lifetime in cooperative wireless networks. The case-study is the cooperative relayed transmission from fixed indoor nodes to a common outdoor access point. A stochastic bivariate model for the spatial distribution of the fading parameters that govern the link performance, namely the Rician K-factor and the path-loss, is proposed and validated by means of real channel measurements. The partner selection protocol is based on the real-time estimation of a function of these fading parameters, i.e., the coding gain. To reduce the complexity of the link quality assessment, a Bayesian approach is proposed that uses the site-specific bivariate model as a-priori information for the coding gain estimation. This link quality estimator allows network lifetime gains almost as if all K-factor values were known. Furthermore, it suits IEEE 802.15.4 compliant networks as it efficiently exploits the information acquired from the receiver signal strength indicator. Extensive numerical results highlight the trade-off between complexity, robustness to model mismatches and network lifetime performance. We show for instance that infrequent updates of the site-specific model through K-factor estimation over a subset of links are sufficient to at least double the network lifetime with respect to existing algorithms based on path loss information only.