Wide-Sense-Stationarity of Everyday Wireless Channels for Body-to-Body Networks

Abstract

The existence of wide-sense-stationarity (WSS) in narrowband wireless body-to-body networks is investigated for "everyday" scenarios using many hours of contiguous experimental data. We employ different parametric and non-parametric hypothesis tests for evaluating mean and variance stationarity, along with distribution consistency, of several body-to-body channels found from different on-body sensor locations. We also estimate the variation of power spectrum to evaluate the time independence of the auto-covariance function. Our results show that, with 95% confidence, the assumption of WSS is met for at most 90% of the cases with window lengths of 5 seconds for the channels between the hubs of different BANs. Additionally, in the best-case scenario, the hub-to-hub channel remains reasonably stationary (with more than 80% probability of satisfying the null hypothesis) for longer window lengths of more than 10 seconds. The short time power spectral variation for body-to-body channels is also shown to be negligible. Moreover, we show that body-to-body channels can be considered widesense-stationary over significantly longer periods than on-body channels.