Link quality prediction for wireless devices with multiple radios

Abstract

Communication between wireless devices ought to be as simple as possible; they should be able to seamlessly switch between different radios and network stacks on the fly in order to better serve the user. To make this a possibility, we consider the challenging problem of predicting link quality in a changing mobile environment. In this paper we present an algorithm that uses Weighted Least Squares Regression to predict whether a given link can meet application requirements in terms of throughput, delay, and jitter. We use a simulation study to demonstrate that our algorithm is able to predict link quality accurately and stably in a frequently changing mobile environment. The prediction algorithm is more accurate than several alternative algorithms, and the overhead caused by the link measurements is negligible in terms of throughput and power consumption.